Merge remote-tracking branch 'origin/master' into worktree/plan-review-layout

.agents/notes/implemented/architecture/2026-07-23-client-plugin-loading-model.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-23-client-plugin-loading-model.md: 4028f50bf5cf8a05063df3bf5e2b4b7f45a3c02e
-2026-07-23-client-plugin-loading-model.zh.md: 7dc96b47c950af07adb71ba7fe131d53e5c0e51d
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-23-client-plugin-loading-model.md
+2026-07-23-client-plugin-loading-model.md: b0873b7aa7bccd3d613f3113fa18207770952e5d
+2026-07-23-client-plugin-loading-model.zh.md: f3472dbfc5a78924e77337bf92ce5983c8492c4c

.agents/notes/implemented/architecture/2026-07-23-client-plugin-loading-model.md

@@ -116,7 +116,7 @@ One governance implementation runs on both sides of the wire; the browser-specif
 
 Costs accepted: the vendored Loader carries idle machinery in the browser (EntryTree persistence is a no-op, groups/isolation unused); every plugin edit in dev pays a bundle rebuild plus fiber remount; graph `inject` rows are informational — activation truth is service-level — so a mismatch surfaces at the settled sweep, not at graph validation; and the three not-yet-promoted libraries keep their static-import export surface until their DI conversions land.
 
-Roster endgame (landed 2026-07-25 with the config-tree boot move): the roster lives in `apps/cli/cordis.yml`, `mountWebPlugins` and the `CLIENT_PACKAGES` constant are gone, and recomposing a deployment means swapping the yml/overlay. The graph composer moved from a webserver-side registry into the `dsh-client-modules` node half (the package upgraded to dual-face per this note's promotion rule — its consumer now reaches it through cordis DI), and the transport split landed alongside: the webserver became a plain route-registration plugin, `/api/*` binding moved to the connection node half over the upgraded `api-gateway` plugin (`dsh-host-apiproxy` providing `ctx.apiProxy`), and the dev bundle watch + SSE channel moved to the hmr node half.
+Roster endgame (landed 2026-07-25 with the config-tree boot move): the roster lives in `apps/cli/config/web.cordis.yml`, `mountWebPlugins` and the `CLIENT_PACKAGES` constant are gone, and recomposing a deployment means swapping the yml/overlay. The graph composer moved from a webserver-side registry into the `dsh-client-modules` node half (the package upgraded to dual-face per this note's promotion rule — its consumer now reaches it through cordis DI), and the transport split landed alongside: the webserver became a plain route-registration plugin, `/api/*` binding moved to the connection node half over the upgraded `api-gateway` plugin (`dsh-host-apiproxy` providing `ctx.apiProxy`), and the dev bundle watch + SSE channel moved to the hmr node half.
 
 ## Alternatives considered
 

.agents/notes/implemented/architecture/2026-07-23-client-plugin-loading-model.zh.md

@@ -116,7 +116,7 @@ wire 两侧跑着同一份治理实现；浏览器特有的表面只是一套模
 
 接受的代价：vendored Loader 在浏览器里背着闲置机件（EntryTree 持久化是 no-op，分组/隔离未用）；开发期每次修改插件都要付一次 bundle 重建加 fiber 重挂；图中 `inject` 行仅是信息性说明——激活的真相在服务层——因此不匹配会在 settled 扫描时浮出，而不是在图校验时被拦下；三个尚未升格的库在各自的 DI 转换落地之前保持静态 import 的导出面。
 
-名册的终局（2026-07-25 随配置树 boot 迁移落地）：名册住 `apps/cli/cordis.yml`，`mountWebPlugins` 与 `CLIENT_PACKAGES` 常量已消失，重组一次部署等于换 yml/overlay。图的组合器从 webserver 侧的注册表迁进 `dsh-client-modules` 的 node 半（该包按本 note 的升级法则升格为双面——其消费方现经 cordis DI 到达），传输拆分同轮落地：webserver 变为朴素路由注册插件，`/api/*` 绑定迁到 connection 的 node 半、走升格后的 `api-gateway` 插件（`dsh-host-apiproxy` 提供 `ctx.apiProxy`），dev 的 bundle 监视与 SSE 通道迁到 hmr 的 node 半。
+名册的终局（2026-07-25 随配置树 boot 迁移落地）：名册住 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml`，`mountWebPlugins` 与 `CLIENT_PACKAGES` 常量已消失，重组一次部署等于换 yml/overlay。图的组合器从 webserver 侧的注册表迁进 `dsh-client-modules` 的 node 半（该包按本 note 的升级法则升格为双面——其消费方现经 cordis DI 到达），传输拆分同轮落地：webserver 变为朴素路由注册插件，`/api/*` 绑定迁到 connection 的 node 半、走升格后的 `api-gateway` 插件（`dsh-host-apiproxy` 提供 `ctx.apiProxy`），dev 的 bundle 监视与 SSE 通道迁到 hmr 的 node 半。
 
 ## Alternatives considered
 

.agents/notes/implemented/architecture/2026-07-24-web-config-tree-boot-and-transport-layering.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-24-web-config-tree-boot-and-transport-layering.md: 99e5d0d95f320464a6f12857e9d6e90e3587c777
-2026-07-24-web-config-tree-boot-and-transport-layering.zh.md: 89c1a578f0ad0841eea7512538e3de8095b82c4d
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-24-web-config-tree-boot-and-transport-layering.md
+2026-07-24-web-config-tree-boot-and-transport-layering.md: a2080024d36d54162f4f4aa79896e51efd708f59
+2026-07-24-web-config-tree-boot-and-transport-layering.zh.md: 1c430bf2939f9f556f378bdeb78872a0091a592d

.agents/notes/implemented/architecture/2026-07-24-web-config-tree-boot-and-transport-layering.md

@@ -12,7 +12,7 @@ English | [中文](2026-07-24-web-config-tree-boot-and-transport-layering.zh.md)
 
 ## Decision
 
-**Composition is one flat config tree.** `apps/cli/cordis.yml` holds every row — the host runtime (32 rows), the `api-gateway` row, the `webserver` row, and the `dshClient` rows (the browser roster; the modules row is simultaneously a host row). No spine bundle: every plugin is one row and every config field is yml-editable. `--dev` appends the `dsh-client-hmr` row in code before the settle sweep — prod and dev differ by exactly that row. Row order carries no load semantics; activation is service-availability driven, and the boot compensates with a fail-loud triple: `assertEntriesLoaded` (import failures), `installFailLoud` (late apply rejections), and an all-ACTIVE sweep (PENDING fibers — cordis inject waiting has no timeout).
+**Composition is one flat assembled tree.** `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml` holds every row — the host runtime (32 rows), the `api-gateway` row, the `webserver` row, and the `dshClient` rows (the browser roster; the modules row is simultaneously a host row). No spine bundle: every plugin is one row and every config field is yml-editable. That stance later became repository-wide, with the rows both surfaces share factored into `apps/cli/config/base.cordis.yml` and each surface reduced to an overlay ([shared-base overlays](../simplification/2026-07-29-shared-base-config-overlays.md)). `--dev` appends the `dsh-client-hmr` row in code before the settle sweep — prod and dev differ by exactly that row. Row order carries no load semantics; activation is service-availability driven, and the boot compensates with a fail-loud triple: `assertEntriesLoaded` (import failures), `installFailLoud` (late apply rejections), and an all-ACTIVE sweep (PENDING fibers — cordis inject waiting has no timeout).
 
 **Boot glue is a class pair.** `AppCLIEntry` (apps/cli) and `AppWebEntry` (the shell kernel) hold only what must exist independently of cordis: argv facts, the composed patch set, the parsed boot manifest, the module system instance, loading-page handles — everything else lives in plugins. `AppCLIEntry.run()` is three stages: layered env (ambient > cwd `.env` > `$DSH_HOME/.env`, closing the defect above) → patch composition → Loader include boot plus the triple. `AppWebEntry.run()` mirrors it browser-side: parse `window.__DSH_BOOT__` into a `BootManifest` (two views: npm-package rows for the module table, cordis-plugin rows for entry composition; malformed wire throws), build the module system, render the loading page, prefetch the `immediately` tier in parallel with Context/Loader setup, **await the prefetch before creating entries** (materialization is `tree.import`'s synchronous require, unprotected by fiber inject waiting; cross-package require edges such as i18n → runtime/client need every immediately-tier factory registered first — an empirically found 10–25% boot race otherwise), adopt the modules entry, create the graph rows, settle, sweep.
 

.agents/notes/implemented/architecture/2026-07-24-web-config-tree-boot-and-transport-layering.zh.md

@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-**组合是一棵平铺 config tree。** `apps/cli/cordis.yml` 持有全部行——host runtime（32 行）、`api-gateway` 行、`webserver` 行、`dshClient` 行（浏览器 roster；modules 行同时是 host 行）。不做 spine bundle：每插件一行、每个 config 字段 yml 可改。`--dev` 在 settle sweep 之前由代码追加 `dsh-client-hmr` 行——prod 与 dev 的全部差异就是这一行。行序无装载语义；激活由服务可用性驱动，boot 以 fail-loud 三件套补偿：`assertEntriesLoaded`（import 失败）、`installFailLoud`（迟到的 apply 拒绝）、all-ACTIVE sweep（PENDING fiber——cordis inject 等待没有超时）。
+**组合结果是一棵平铺配置树。** `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 共同持有全部行——host runtime（32 行）、`api-gateway` 行、`webserver` 行、`dshClient` 行（浏览器 roster；modules 行同时是 host 行）。不做 spine bundle：每插件一行、每个 config 字段 yml 可改。这一立场后来推广到全仓：两个 surface 共享的配置项被抽取进 `apps/cli/config/base.cordis.yml`，各 surface 则收敛为一份 overlay（[共享 base overlay](../simplification/2026-07-29-shared-base-config-overlays.md)）。`--dev` 在 settle sweep 之前由代码追加 `dsh-client-hmr` 行——prod 与 dev 的全部差异就是这一行。行序无装载语义；激活由服务可用性驱动，boot 以 fail-loud 三件套补偿：`assertEntriesLoaded`（import 失败）、`installFailLoud`（迟到的 apply 拒绝）、all-ACTIVE sweep（PENDING fiber——cordis inject 等待没有超时）。
 
 **boot 胶水是一对 class。** `AppCLIEntry`（apps/cli）与 `AppWebEntry`（壳内核）只持有独立于 cordis 必须提前存在的东西：argv 事实、合成的 patch 集、解析出的 boot manifest、模块系统实例、loading 页句柄——其余一律进插件。`AppCLIEntry.run()` 三段：分层 env（ambient > cwd `.env` > `$DSH_HOME/.env`，顺手关掉上述缺陷）→ patch 合成 → Loader include boot 加三件套。`AppWebEntry.run()` 在浏览器侧镜像它：把 `window.__DSH_BOOT__` 解析成 `BootManifest`（双视角：npm 包行给模块表、cordis 插件行给 entry 组合；畸形 wire 大声抛）、建模块系统、渲染 loading 页、immediately 层预取与 Context/Loader 准备并行、**create entry 之前等预取齐**（物化是 `tree.import` 的同步 require，不受 fiber inject 等待保护；i18n → runtime/client 这类跨包 require 边要求 immediately 层工厂全部注册完——否则有实测 10–25% 的 boot 竞态）、收编 modules entry、逐图行 create、settle、sweep。
 

.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md
-2026-07-28-directory-picker-capability-seam.md: 7c8f8cb67690cb4c5858cefb52b8cd79e649ec38
-2026-07-28-directory-picker-capability-seam.zh.md: 05545fc3cd758523814b31afa705249972d86464
+2026-07-28-directory-picker-capability-seam.md: 892bb4b2c4fe200df91866c4ec4cf8bb7c58e940
+2026-07-28-directory-picker-capability-seam.zh.md: d738773adc853dae7b3496f0dc2901eebd0f0a08

.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md

@@ -18,7 +18,9 @@ Placement and policy rulings folded into this decision:
 
 - **Not the `ctx.fs` seam.** `packages/fs/` is the model/session-facing storage stack (policy events, sandbox-swappable backends). Riding it would couple GUI browsing to the model's confinement backend — swapping `fs-sandbox` for the model must never change GUI behavior — and OS facts (home anchoring, hidden conventions) are not storage primitives. The picker seam stays presentation-free and model-free; `packages/host/` is its consumer-domain home.
 - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib.
-- **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the planned show-hidden toggle becomes a client-only change. Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself.
+- **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself.
+- **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets.
+- **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands.
 - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption.
 - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories.
 - **Whole-filesystem scope, no roots config.** `workspace.create` accepts arbitrary paths and the API serves bash-driving methods, so a browse root would be UX scoping, not a boundary; configurability without a consumer fails the evidence bar. Deferred until a deployment needs it.
@@ -30,6 +32,9 @@ Placement and policy rulings folded into this decision:
 - **One uniform seam method set (`pick(): path`).** Rejected: an in-app browser cannot be served behind a single host-side call — the browsing loop lives in the client and needs primitives on the wire; the native chooser cannot implement primitives. The interaction difference is irreducible, hence the discriminant.
 - **Direct stdlib calls inside apiproxy (no seam).** Rejected: keeps the gateway the only swap point (source edits), loses fixture/test backends, and contradicts the plugin doctrine that motivated the work.
 - **Adopting a file-manager/drive-enumeration dependency.** Rejected per the survey above; recorded here as the dependency policy requires.
+- **A flip-label show-hidden toggle ("Hide hidden files").** Rejected: a flipping action label is ambiguous between state and action and doubles the negative; the fixed label with a pressed presentation states both at once.
+- **Pure relatedTarget blur cancellation (no mousedown suppression).** Rejected: Safari does not focus buttons on pointer down, so a click's focusout carries a null `relatedTarget` and would cancel the editor before the click lands; editing-scoped mousedown suppression plus the card-anchored relatedTarget guard covers pointer and keyboard paths together.
+- **A wire `separator` field on `DirectoryListing` (host stamps `path.sep`).** Deferred, not rejected: it is the authoritative form — a POSIX home directory containing a backslash defeats the `listing.home` heuristic — but it touches the seam type and every backend; the browse client's `separatorOf` carries a TODO pointing at this alternative until a wire change is next scheduled.
 
 ## Consequences
 

.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md

@@ -18,7 +18,9 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick
 
 - **不用 `ctx.fs` seam。** `packages/fs/` 是面向模型／会话的存储栈（policy 事件、sandbox 可换后端）。骑上去会把 GUI 浏览耦合进模型的限制后端——为模型换 `fs-sandbox` 绝不能改变 GUI 行为——而 OS 事实（home 锚定、隐藏约定）也不是存储原语。picker seam 保持无展示、无模型；`packages/host/` 是它消费方域的家。
 - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。
-- **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，计划中的"显示隐藏"开关变成纯客户端改动。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。
+- **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。
+- **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。
+- **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。
 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。
 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。
 - **全盘可浏览，不做 roots 配置。** `workspace.create` 接受任意路径且 API 本就提供驱动 bash 的方法，浏览根只会是 UX 范围而非边界；没有消费方的可配置性过不了证据门槛。等到有部署需要再做。
@@ -30,6 +32,9 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick
 - **统一方法集的 seam（`pick(): path`）。** 否决：应用内浏览器无法藏在一次宿主侧调用后面——浏览循环在客户端，需要协议上的原语；而对话框实现不了原语。交互差异不可约，故用判别标签。
 - **apiproxy 里直接调标准库（不建 seam）。** 否决：换装点仍是改网关源码，失去 fixture／测试后端，与促成这项工作的插件教义相悖。
 - **引入文件管理器／盘符枚举依赖。** 按上文调研否决；依赖政策要求记录于此。
+- **动作标签随状态翻转的"显示隐藏"开关（"隐藏隐藏文件"）。** 否决：会翻转的动作标签在状态与动作之间有歧义，还把否定叠了两层；固定标签加按下态呈现一次说清两者。
+- **纯 relatedTarget 失焦取消（不做 mousedown 抑制）。** 否决：Safari 在指针按下时不给按钮聚焦，点击触发的 focusout 因而携带空 `relatedTarget`，会在点击落地前就取消编辑器；编辑期作用的 mousedown 抑制加上锚定卡片的 relatedTarget 守卫才能同时覆盖指针与键盘路径。
+- **在 `DirectoryListing` 上增设线上 `separator` 字段（宿主标注 `path.sep`）。** 延期而非否决：它才是权威形态——含反斜杠的 POSIX 家目录会击穿 `listing.home` 启发式——但它触及 seam 类型与每个后端；browse 客户端的 `separatorOf` 挂着指向本方案的 TODO，直到下次安排线上变更。
 
 ## 后果
 

.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md
-2026-07-28-dsh-native-typescript-source-launch.md: 1ba1dd2663038ad7c49af71f8b428245f7fa3e2b
-2026-07-28-dsh-native-typescript-source-launch.zh.md: 02f84f34820469ad9e810ae17d79d3fe12b0cd4c
+2026-07-28-dsh-native-typescript-source-launch.md: 773f831ec2b116d4908fcd5dc818df78c5deee5e
+2026-07-28-dsh-native-typescript-source-launch.zh.md: 0602aa1ba079fcd5bffb2fe3989da9a4f62763fb

.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md

@@ -20,9 +20,9 @@ The `dsh` TUI, Web, and headless source launches use `node --experimental-transf
 
 `scripts/tspath-loader.ts` registers only a module resolve hook. It uses `TSX_TSCONFIG_PATH` when set (resolving relative values from the invoking cwd) and otherwise reads the root `tsconfig.json`; `TsconfigPathsResolver` follows that config's `extends` chain through the repository's existing TypeScript development tool, selects exact or wildcard `paths` entries according to tsconfig rules, and maps matching workspace bare specifiers to `.ts`/`.mts`/`.cts` source files or directory index files. Node remains solely responsible for code transformation. The source-only loader is not part of the built CLI and `apps/cli` does not declare `typescript` as a runtime dependency.
 
-Source imports are redirected only when the target package is either the nearest package manifest's own name or one of that manifest's declared runtime dependencies. The Cordis Loader uses the configuration directory URL as the import parent; the resolver then searches upward for the workspace manifest that declares the plugin, so dependency ownership for `examples/tui-agent/cordis.yml` lies with `examples/package.json`, and dependency ownership for `apps/cli/cordis.yml` lies with `apps/cli/package.json`. Specifiers that do not match tsconfig paths, refer to undeclared dependencies, or are not bare all fall back to Node's default resolution.
+Source imports are redirected only when the target package is either the nearest package manifest's own name or one of that manifest's declared runtime dependencies. The Cordis Loader uses the configuration directory URL as the import parent; the resolver then searches upward for the workspace manifest that declares the plugin, so dependency ownership for the shipped `apps/cli/config/base.cordis.yml` plus its surface overlay lies with `apps/cli/package.json`. Specifiers that do not match tsconfig paths, refer to undeclared dependencies, or are not bare all fall back to Node's default resolution.
 
-`verify-cordis-config` performs a one-way completeness check on both resolver manifests: every bare plugin package in a configuration must appear in the corresponding manifest's `dependencies`, while the manifest may contain extra dependencies not referenced by that configuration. The root `AGENTS.md` makes updating the configuration and dependencies together a standing rule.
+`verify-cordis-config` performs a one-way completeness check on the resolver manifest: every bare plugin package in a configuration must appear in the corresponding manifest's `dependencies`, while the manifest may contain extra dependencies not referenced by that configuration. The root `AGENTS.md` makes updating the configuration and dependencies together a standing rule.
 
 After the Loader settles, the shared `dsh-app-boot` checks every enabled entry that has no fiber and rejects startup with `plugin(s) failed to load: ...; Cordis startup failed because these plugin(s) could not be resolved`, listing all failed plugins. This diagnostic lives at the app layer and does not change the vendored Loader's startup behavior.
 

.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md

@@ -20,9 +20,9 @@ Cordis 配置还引入了另一条解析边界。`cordis.yml` 中的 bare plugin
 
 `scripts/tspath-loader.ts` 只注册一个模块 resolve hook。设置 `TSX_TSCONFIG_PATH` 时，它会使用该路径（相对路径从调用方的 cwd 解析），否则读取根 `tsconfig.json`；`TsconfigPathsResolver` 使用仓库已有的 TypeScript 开发工具沿该配置的 `extends` 链解析，按 tsconfig 规则选择精确或 wildcard `paths` 条目，并将命中的 workspace bare specifier 映射到 `.ts`/`.mts`/`.cts` 源文件或目录 index 文件。代码转换始终只由 Node 负责。该源码专用 loader 不属于构建后的 CLI，`apps/cli` 也不会把 `typescript` 声明为运行时依赖。
 
-只有当目标包是最近 package manifest 的自身名称或其已声明的运行时依赖时，源码 import 才会重定向。Cordis Loader 使用配置目录 URL 作为 import parent；此时 resolver 会向上查找声明该插件的 workspace manifest。因此，`examples/tui-agent/cordis.yml` 的依赖由 `examples/package.json` 持有，`apps/cli/cordis.yml` 的依赖由 `apps/cli/package.json` 持有。未命中 tsconfig paths、引用未声明依赖或不是 bare specifier 的说明符全部交回 Node 默认解析。
+只有当目标包是最近 package manifest 的自身名称或其已声明的运行时依赖时，源码 import 才会重定向。Cordis Loader 使用配置目录 URL 作为 import parent；此时 resolver 会向上查找声明该插件的 workspace manifest。因此，已交付的 `apps/cli/config/base.cordis.yml` 及其界面覆盖层所需依赖由 `apps/cli/package.json` 持有。未命中 tsconfig paths、引用未声明依赖或不是 bare specifier 的说明符全部交回 Node 默认解析。
 
-`verify-cordis-config` 对这两个解析方 manifest 执行单向完整性检查：配置中的每个 bare plugin package 都必须出现在对应 manifest 的 `dependencies` 中，manifest 可以包含该配置未引用的额外依赖。根 `AGENTS.md` 将同步更新配置和依赖定为常驻规则。
+`verify-cordis-config` 对该解析方 manifest 执行单向完整性检查：配置中的每个 bare plugin package 都必须出现在对应 manifest 的 `dependencies` 中，manifest 可以包含该配置未引用的额外依赖。根 `AGENTS.md` 将同步更新配置和依赖定为常驻规则。
 
 Loader 完成结算后，共享的 `dsh-app-boot` 会检查每个已启用但没有 fiber 的 entry，并以 `plugin(s) failed to load: ...; Cordis startup failed because these plugin(s) could not be resolved` 拒绝启动，同时列出全部加载失败的插件。该诊断位于应用层，不改变 vendor 中 Loader 的启动行为。
 

.agents/notes/implemented/architecture/2026-07-28-launcher-owned-resume-identity.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-launcher-owned-resume-identity.md
+2026-07-28-launcher-owned-resume-identity.md: e11431716305ff18e951ca45921c71dcd105974c
+2026-07-28-launcher-owned-resume-identity.zh.md: e4078023cb1519288f0367f8973788f361aaa0bd

.agents/notes/implemented/architecture/2026-07-28-launcher-owned-resume-identity.md

@@ -0,0 +1,64 @@
+# Agent Note: Launcher-owned session identity and exit line
+
+Status: implemented
+
+English | [中文](2026-07-28-launcher-owned-resume-identity.zh.md)
+
+## Problem
+
+Two facts a launcher owns were shipped as deployment config keys on the TUI app bundle: `resumeSessionId` (which session `main` binds to) and `resumeCommand` (the exit hint template, with `{session}` interpolated). Neither varies by deployment — both are properties of how the process was invoked, which only the launcher knows.
+
+Routing them through YAML made them silently droppable. `@cordisjs/plugin-include` applies a targeted patch by replacing whole top-level keys (`target[key] = value`), so a personal `~/.dsh/config.yaml` patching the `tui-agent` entry's `config` replaces the shipped block entirely. A user overlay written to change provider and model therefore deleted every resume key it did not restate, and nothing reported it: absent `resumeCommand` legitimately means "no fallback configured".
+
+Both failures were live in one real overlay. The exit hint stopped printing, because the overlay omitted `resumeCommand`. Worse, the overlay carried `resumeSessionId: !!js process.env.RESUME_SESSION_ID` — a stale line from before [the env-var bridge was removed](../../archived/architecture/2026-07-24-dsh-commander-argument-adapter.md) — which overwrote the shipped `!!js "typeof resumeSessionId === 'string' ? …"` intake with a read of a variable nothing sets. `dsh --resume <valid-id>` then started a *fresh* session and said nothing, reproduced directly: the banner showed a newly minted id, not the requested one. The [`dsh meta`](../feature/2026-07-28-dsh-meta-source-workspace.md) note had recorded this silent resume as an unexplained pre-existing defect; the overlay's shallow replacement is the cause.
+
+A config key cannot express these facts safely, because the deployment is not the authority on them.
+
+## Decision
+
+Session identity and the exit line are launcher-owned context slots, provided before any Loader entry mounts. Neither appears in any `cordis.yml` nor in any plugin's `Config`.
+
+Both sit beside the existing `tuiResumeHost` host capability, which set the precedent — a resume host has always been a provided capability rather than config. Each slot is declared by the package that consumes it:
+
+- `CONFIGURED_AGENT_IDENTITIES_KEY` (`dsh-agent-loop`) carries launcher identities keyed by configured-agent `id`, each a `LauncherAgentIdentity` (`{ id: SessionId, resume: boolean }`). `agent-loop` applies the matching identity over its configured agent, replacing both identity keys, and takes the history-loading `resumeSessionId` path only when `resume` is set, because that path requires an existing log and fails loud without one. An absent slot leaves the configured identity untouched. The `tui` row resolves the same id through its own `sessionId` key, so the front door renders exactly the agent that was bound.
+- `TUI_GOODBYE_MESSAGE_KEY` (`dsh-tui`) carries the complete line printed once the terminal is released on exit. Absent prints nothing.
+
+Identity belongs to `agent-loop` because that is the plugin which creates configured agents, and because a patch replaces a row's whole `config`: an overlay repointing the agent row's model route would erase a launcher-set identity key. See [the shared-base overlay note](../simplification/2026-07-29-shared-base-config-overlays.md).
+
+`apps/cli` mints or selects the id and builds the line from the invocation it is reproducing, sharing one `resumeArgs` helper with the `/resume` execve handoff so the printed command and the in-place handoff cannot diverge. The line now names `--config` when one was passed, and reproduces `dsh meta --resume <id>` in meta mode — closing the mode-aware hint deferred by the `dsh meta` note, where a copied hint previously only worked from the checkout.
+
+**`ctx.provide` is the only channel from launcher argv into a Loader-mounted plugin.** Config `!!js` expressions evaluate as `with (entry.ctx) { eval(expr) }` (`vendor/loader/src/config/utils.ts`), so a bare identifier resolves against the entry's context and nothing else reaches it. The slot therefore cannot be removed while the app bundle is mounted from YAML; what changes is that it is now internal launcher↔app plumbing instead of a documented key a config author must wire correctly.
+
+The message is a plain string, not a callback. That forces the launcher to know the id before boot, which is why minting moved out of the app bundle — and it keeps exit free of awaited work after the terminal is released.
+
+The TUI owns rendering, not wording: it applies `displayText` before its own `palette.muted`, so a hostile `--config` path cannot inject terminal escapes into the exit line. Sanitizing means the launcher cannot embed its own ANSI.
+
+## Alternatives considered
+
+**Keep the keys and add built-in defaults in the app bundle.** Rejected: a default in code survives an overlay, but two ways to state one fact remain, and a config author can still set the key wrong — which is exactly how the stale `process.env.RESUME_SESSION_ID` line disabled resume.
+
+**Merge the app bundle into `apps/cli` and delete the slot entirely.** Rejected here, then [adopted later](../simplification/2026-07-29-shared-base-config-overlays.md) in a form this note did not consider: the composition moved into flat config files (`apps/cli/config/base.cordis.yml` plus a per-surface overlay) rather than into CLI code, so it never entered the `v8 ignore` process-wiring block, and the overlay extension points survive as ordinary row patches. The slot itself was not deleted — it moved to `dsh-agent-loop`, because a launcher fact still cannot travel through a replaceable config key.
+
+**Put the goodbye message on `TuiResumeHost`.** Rejected: an exit line is not a handoff capability, and a host that cannot replace its process may still want to print one. They are independent slots.
+
+**Have the host supply only the command text and let the TUI keep the `To resume this session:` prefix.** Rejected: the TUI would retain resume vocabulary for a string it no longer understands, and meta mode proves the launcher is the only component that knows what the command should say.
+
+**Let the TUI keep suppressing the line until the session is durably persisted.** Rejected: that check is why the exit path queried persistence and swallowed listing failures. A plain string cannot consult persistence, and misuse now fails loud through `agent-loop/config-start-failed` rather than silently resuming nothing.
+
+**A callback (`goodbyeMessage(agent)`) so the host could decide at exit time.** Rejected: it restores async work after `ui.stop()`, reintroducing a hang risk during teardown for a string that is already knowable at boot.
+
+## Consequences
+
+- Removing two published `Config` keys is a breaking config change: a stale config naming either now fails schema validation at boot instead of degrading silently. Intended, and acceptable pre-release.
+- `TuiResumeHost` is unchanged, but `TuiRuntime` gains `goodbyeMessage`; `apps/cli` is the only provider.
+- The exit line prints even for a session with no log (launch, quit immediately). Using it then fails loud rather than starting a surprise session. This is the deliberate cost of dropping the persistence check.
+- `dsh-tui` no longer reads `sessionPersistence` at all: `currentResumeCommand`, `listWorkspaceSessions`, and its swallowed-error path are deleted, and the `/resume` selector's `sessionQuery` reads are now the only session discovery in the TUI.
+- The launcher mints session ids for its own app, so a non-CLI host that provides no slot keeps the bundle's own minting.
+
+## Testing
+
+`packages/ui/tui/tests/tui.spec.ts` pins the printed line, the absent-slot silence, and escape sanitization of a hostile message; the former two exit-suppression tests are replaced, since suppression is the behavior this change removes. `packages/core/agent-loop/tests/` drives the identity slot for the resume, launcher-minted, and no-slot cases.
+
+The load-bearing coverage is `apps/cli/tests/tui-keyless-smoke.e2e.ts`, which launches the real `apps/cli/src/bin.ts` in a PTY: one test asserts the exit line carries `--config`, and a regression test seeds a personal `config.yaml` that replaces the entire `agent-loop` config block and asserts the line still prints — encoding "an overlay cannot drop resume" as an executed contract rather than a comment.
+
+Verified live in tmux against the real personal overlay: the defect reproduced on unmodified staging (requested id ignored, fresh id in the banner), and on this branch the same overlay yields a printed exit line, a `--resume` that restores the prior turn, and a `/resume` selector marking the session `current · live · persisted`. A wrong id now fails loud.

.agents/notes/implemented/architecture/2026-07-28-launcher-owned-resume-identity.zh.md

@@ -0,0 +1,64 @@
+# Agent Note：由启动器持有的会话身份与退出行
+
+Status: implemented
+
+[English](2026-07-28-launcher-owned-resume-identity.md) | 中文
+
+## Problem
+
+有两项本应由启动器持有的事实，却被作为 TUI 应用组合包上的部署配置键交付：`resumeSessionId`（`main` 绑定到哪个会话）与 `resumeCommand`（退出提示的模板，其中 `{session}` 会被插值）。二者都不随部署而变——它们都是进程被如何调用的属性，而这一点只有启动器知道。
+
+把它们经由 YAML 传递，使其可被静默丢弃。`@cordisjs/plugin-include` 施加定向补丁的方式是替换整个顶层键（`target[key] = value`），因此一份对 `tui-agent` 条目的 `config` 打补丁的个人 `~/.dsh/config.yaml`，会把交付时的整块内容整体替换掉。于是，一份为改动 provider 和 model 而写的用户 overlay，会删掉它未重述的每一个 resume 键，且没有任何东西报告这一点：缺失 `resumeCommand` 合法地意味着「未配置回退」。
+
+两处失效在同一份真实的 overlay 中同时存在。退出提示不再打印，因为该 overlay 省略了 `resumeCommand`。更糟的是，该 overlay 带着 `resumeSessionId: !!js process.env.RESUME_SESSION_ID`——一行来自 [env 变量桥被移除](../../archived/architecture/2026-07-24-dsh-commander-argument-adapter.md)之前的陈旧代码——它用一次对某个无人设置的变量的读取，覆盖掉了交付时的 `!!js "typeof resumeSessionId === 'string' ? …"` 入口。此后 `dsh --resume <valid-id>` 会开启一个*全新*会话且什么都不说，并被直接复现：banner 显示的是一个新铸造的 id，而非所请求的那个。[`dsh meta`](../feature/2026-07-28-dsh-meta-source-workspace.md) note 曾把这次静默的 resume 记为一处无法解释的既有缺陷；而 overlay 的浅层替换正是其成因。
+
+一个配置键无法安全地表达这些事实，因为部署方并非它们的权威。
+
+## Decision
+
+会话身份与退出行是由启动器持有的上下文槽位，在任何 Loader 条目挂载之前提供。二者都不出现在任何 `cordis.yml` 中，也不出现在任何插件的 `Config` 中。
+
+这两个槽位与既有的 `tuiResumeHost` 宿主能力并列，后者确立了先例——resume 宿主一直是一项被提供的能力，而非配置。每个槽位都由消费它的包声明：
+
+- `CONFIGURED_AGENT_IDENTITIES_KEY`（`dsh-agent-loop`）按所配置 agent 的 `id` 承载启动器身份，每项为一个 `LauncherAgentIdentity`（`{ id: SessionId, resume: boolean }`）。`agent-loop` 将匹配的身份覆盖到其所配置的 agent 上，替换两个身份键；并且仅当 `resume` 被置位时才走加载历史的 `resumeSessionId` 路径，因为该路径要求存在一份日志、否则会明确报错。槽位缺失则保留配置中的身份不变。`tui` 配置项通过自身的 `sessionId` 键解析同一个 id，因此前端入口渲染的正是被绑定的那个 agent。
+- `TUI_GOODBYE_MESSAGE_KEY`（`dsh-tui`）承载退出时终端释放后打印一次的完整行。缺失则什么都不打印。
+
+身份归属于 `agent-loop`，因为它才是创建所配置 agent 的插件；也因为 patch 会整体替换配置项的 `config`：重新指向 agent 配置项模型路由的 overlay 会抹掉启动器设置的身份键。参见[共享 base overlay note](../simplification/2026-07-29-shared-base-config-overlays.md)。
+
+`apps/cli` 铸造或选定 id，并依据它所复现的那次调用构建该行，与 `/resume` 的 execve 移交共用同一个 `resumeArgs` 助手，从而使打印出的命令与原地移交不会分歧。该行现在会在传入了 `--config` 时命名它，并在 meta 模式下复现 `dsh meta --resume <id>`——从而收口了 `dsh meta` note 所推迟的随 mode 变化的提示，在那里被复制的提示此前只有在检出目录中才有效。
+
+**`ctx.provide` 是从启动器 argv 进入被 Loader 挂载的插件的唯一通道。** 配置的 `!!js` 表达式会以 `with (entry.ctx) { eval(expr) }`（`vendor/loader/src/config/utils.ts`）求值，因此一个裸标识符会针对该条目的上下文解析，别无它物可达。于是只要应用 bundle 仍从 YAML 挂载，这个槽位就无法被移除；变化之处在于它现在是启动器↔应用之间的内部管线，而不再是一个配置作者必须正确接线的、有文档记载的键。
+
+该消息是一个纯字符串，而非回调。这迫使启动器在启动前就知道 id，也正是铸造从应用 bundle 中移出的原因——并且它让退出在终端释放之后免于任何被 await 的工作。
+
+TUI 持有渲染，而非措辞：它在自己的 `palette.muted` 之前先应用 `displayText`，因此一个恶意的 `--config` 路径无法把终端转义序列注入退出行。做净化意味着启动器无法嵌入自己的 ANSI。
+
+## Alternatives considered
+
+**保留这些键，并在应用组合包中加入内建默认值。** 拒绝：代码中的默认值能在 overlay 下存活，但表达同一事实的两种途径依然并存，而配置作者仍可把键设错——这正是那行陈旧的 `process.env.RESUME_SESSION_ID` 使 resume 失效的方式。
+
+**把应用组合包合并进 `apps/cli` 并彻底删除该槽位。** 此处拒绝，但后来以本 note 未曾设想的形式[被采纳](../simplification/2026-07-29-shared-base-config-overlays.md)：组合被搬进平铺的配置文件（`apps/cli/config/base.cordis.yml` 加各 surface 一份 overlay），而非搬进 CLI 代码，因此从未进入 `v8 ignore` 进程接线块，overlay 的扩展点也作为普通配置项 patch 保留了下来。槽位本身并未被删除——它迁移到了 `dsh-agent-loop`，因为启动器的事实依然不能经由一个可被整体替换的配置键传递。
+
+**把 goodbye 消息放到 `TuiResumeHost` 上。** 拒绝：退出行不是一项移交能力，而一个无法替换自身进程的宿主仍可能想要打印一行。它们是相互独立的槽位。
+
+**让宿主只提供命令文本，而由 TUI 保留 `To resume this session:` 前缀。** 拒绝：TUI 将为一个它已不再理解的字符串保留 resume 词汇，而 meta 模式证明启动器才是唯一知道该命令应当说什么的组件。
+
+**让 TUI 继续在会话被持久化之前抑制该行。** 拒绝：这项检查正是退出路径要查询持久化并吞掉列举失败的原因。一个纯字符串无法查询持久化，而误用现在会经由 `agent-loop/config-start-failed` 明确报错，而不是静默地恢复了个空。
+
+**用一个回调（`goodbyeMessage(agent)`）让宿主能在退出时决定。** 拒绝：它会在 `ui.stop()` 之后恢复异步工作，为一个在启动时就已可知的字符串，重新引入拆解期间的挂起风险。
+
+## Consequences
+
+- 移除两个已发布的 `Config` 键是一次破坏性配置变更：一份命名了任一键的陈旧配置，现在会在启动时的 schema 校验中明确报错，而不再静默降级。这是有意为之，且在预发布阶段可以接受。
+- `TuiResumeHost` 保持不变，但 `TuiRuntime` 新增 `goodbyeMessage`；`apps/cli` 是唯一的提供方。
+- 即便某会话没有日志（启动后立即退出），退出行也会打印。此时使用它会明确报错，而不是开启一个意外的会话。这是丢弃持久化检查的有意代价。
+- `dsh-tui` 完全不再读取 `sessionPersistence`：`currentResumeCommand`、`listWorkspaceSessions` 及其吞错路径都被删除，`/resume` 选择器的 `sessionQuery` 读取如今是 TUI 中唯一的会话发现途径。
+- 启动器为其自身的应用铸造会话 id，因此一个不提供任何槽位的非 CLI 宿主，仍保留 bundle 自带的铸造逻辑。
+
+## Testing
+
+`packages/ui/tui/tests/tui.spec.ts` 钉住打印出的行、槽位缺失时的静默，以及对恶意消息的转义净化；此前那两个退出抑制测试被替换，因为抑制正是本次改动移除的行为。`packages/core/agent-loop/tests/` 为 resume、启动器铸造与无槽位三种情形驱动身份槽位。
+
+承重的覆盖是 `apps/cli/tests/tui-keyless-smoke.e2e.ts`，它在一个 PTY 中拉起真实的 `apps/cli/src/bin.ts`：一个测试断言退出行携带 `--config`，一个回归测试植入一份个人 `config.yaml` 来替换整块 `agent-loop` 配置块并断言该行仍会打印——把「overlay 不能丢掉 resume」编码为一条被执行的契约，而非一句注释。
+
+在 tmux 中针对真实的个人 overlay 做过实测：该缺陷在未修改的 staging 上复现（所请求的 id 被忽略，banner 里是新的 id），而在本分支上同一份 overlay 会产出一行打印的退出行、一个能恢复上一轮次的 `--resume`，以及一个把该会话标记为 `current · live · persisted` 的 `/resume` 选择器。错误的 id 现在会明确报错。

.agents/notes/implemented/architecture/2026-07-28-user-settings-seam.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-user-settings-seam.md
+2026-07-28-user-settings-seam.md: bf93f95168b1b6d0dec5a9fc2c9aac5531f0564a
+2026-07-28-user-settings-seam.zh.md: 8cd4dfcbb2facdd590b2c24d453ad79e9badda4d

.agents/notes/implemented/architecture/2026-07-28-user-settings-seam.md

@@ -0,0 +1,35 @@
+# Agent Note: user-settings seam (`ctx.settings`) and the file provider
+
+Status: implemented
+
+English | [中文](2026-07-28-user-settings-seam.zh.md)
+
+> Scope: the `packages/settings/` capability family — the abstract seam, the file-backed provider, and the composition boundary between user settings and `cordis.yml`. The [web config-tree note](2026-07-24-web-config-tree-boot-and-transport-layering.md) recorded "the profile write path" as a deferral; this seam is that write path's owner. Consumer migrations (theme, locale, default model route) and the web `settings.*` RPC surface are follow-ups, not part of this note's shipped scope.
+
+## Problem
+
+User-editable configuration had no owner: `dsh web` read a cwd-anchored profile json through a static whitelist with no write path, the TUI read `$DSH_HOME/config.yaml` raw loader patches, and both froze at boot. A personal-settings page (web GUI) needs one cross-surface user layer with schema validation, a write path, and hot propagation — and peer products (Codex, Claude Code, Kimi, OpenCode, Pi) all converged on separating user preferences from extension composition. Loader-reactive config updates cannot carry this: `fiber.update` swaps entry config in place, so a plugin that read config at construction observes nothing and no callback tells it otherwise.
+
+## Decision
+
+**Two planes with a litmus test.** `cordis.yml` (+ Include patches) stays the composition plane: which plugins exist, wiring, deployment config, owned by the orchestrator and upgraded with the product. A settings namespace carries only the user-editable subset; the test is "should the personal config page edit it?" Values live in both planes without ambiguity because layering is the contract: schema defaults, then the registrant's composition `base` (its entry-config subset), then the user document section.
+
+**Three-package seam mirroring `session-persistence/`.** `dsh-settings` owns the abstract `Settings` service: namespace registry, layered resolution, schema validation, per-namespace deep-equal change detection, and the `settings/updated` commit event. Providers implement only `writable`/`load()`/`persist(ns, section)` and push externally observed documents through the protected `publish(doc)` — so hot-update semantics are identical across providers, and a network configuration-center backend (nacos-style, possibly read-only) is a sibling package away. `dsh-settings-local` is the file provider: YAML/JSON under `resolveSpec` (explicit defaulting to `<DSH_HOME>/settings.yaml`), chokidar watch, read-modify-write persists under a cross-process writer lock with atomic `0600` tmp+rename commits, leaf-level diff patching of the written namespace (comments survive untouched nodes), and content-equality self-write suppression ([write-path integrity note](2026-07-30-settings-write-path-integrity.md)).
+
+**Registrations are caller-fiber effects.** `register()` runs through the service proxy, so `this.ctx` is the registrant's context and the registration rides `ctx.effect`: disposing the registrant removes the namespace and its watchers (proven by the HMR disposal test), while the user's section keeps living in storage for the next owner.
+
+**Fail loud at rest, last-good in motion.** Boot-time and registration-time validation throw (invalid stored section fails the registering plugin; an existing-but-unparsable document fails provider load). Once live, a bad external edit warns and keeps the last good state per namespace — a hot reload must never take the process down. This asymmetry mirrors `Include.refresh()` and Kimi's safe runtime reload.
+
+**Consumers stay optional-by-construction.** A consumer registers inside `ctx.inject(['settings'], …)`; without a mounted provider it keeps resolving entry config alone, so every existing composition, demo, and snapshot works unchanged and migration is per-plugin.
+
+## Alternatives considered
+
+- **Include write-back as the user layer** (per-plugin config pages writing loader entry files, cordis-webui style): write-back would target per-composition files, binding user preferences to one `cordis.yml`; a per-user layer must survive template upgrades and serve TUI and web from one document.
+- **Loader-reactive `fiber.update` as the propagation channel**: constructor-time reads observe nothing; the seam's explicit `watch()` makes hot-update a consumer contract instead of framework magic.
+- **A domain-aware settings service** (getters per product area): the coupling objection from design review stands; the service stores, validates, and publishes — domain meaning stays with the registrant that owns the schema.
+- **Multi-layer precedence now** (system/managed/project tiers à la Codex/Claude Code): deferred until a real second layer exists; the resolve step is the single place layering would extend.
+- **A cross-process lockfile now** (Pi's proper-lockfile): initially deferred as "atomic replace plus watcher convergence until real contention shows up" — review showed convergence loses unobserved sibling namespaces, so the deferral is superseded by the [write-path integrity note](2026-07-30-settings-write-path-integrity.md)'s hand-rolled writer lock.
+
+## Consequences
+
+Deferred, in dependency order: the web `settings.raw`/`settings.describe`/`settings.update` RPC surface (which must redact `role('secret')` fields before exposure); first consumer migrations (`ui-theme`, locale, api-gateway default route) retiring `PROFILE_MAPPINGS` and the profile json; `${env:VAR}` value indirection for secrets; provider-side layering. The keyless snapshot obligation lands with the first model- or product-user-visible consumer, not with this infrastructure step.

.agents/notes/implemented/architecture/2026-07-28-user-settings-seam.zh.md

@@ -0,0 +1,35 @@
+# Agent Note：用户设置 seam（`ctx.settings`）与文件 provider
+
+Status: implemented
+
+[English](2026-07-28-user-settings-seam.md) | 中文
+
+> 范围：`packages/settings/` 能力族——抽象 seam、文件 provider，以及用户设置与 `cordis.yml` 的组合边界。[web config-tree note](2026-07-24-web-config-tree-boot-and-transport-layering.md) 曾把"profile 写路径"记为延后项；本 seam 就是该写路径的归属。消费者迁移（主题、语言、默认模型路由）与 web `settings.*` RPC 面是后续工作，不在本 note 已交付范围内。
+
+## 问题
+
+用户可编辑配置没有归属：`dsh web` 经静态白名单读 cwd 锚定的 profile json 且无写路径，TUI 读 `$DSH_HOME/config.yaml` 裸 loader patch，两者都在启动时冻结。个人设置页（web GUI）需要一个跨 surface 的用户层，带 schema 校验、写路径与热传导——同类产品（Codex、Claude Code、Kimi、OpenCode、Pi）也全部收敛于"用户偏好与扩展组合分离"。Loader 的 reactive 配置更新承载不了这件事：`fiber.update` 原地替换 entry config，构造期读过配置的插件毫无感知，也没有任何回调通知它。
+
+## 决策
+
+**两个面，一条判定。**`cordis.yml`（+ Include patches）仍是组合面：有哪些插件、接线、部署配置，归 orchestrator 所有并随产品升级。settings namespace 只承载用户可编辑子集；判定是"个人配置页应该能改它吗？"值可同时存在于两个面而不歧义，因为分层就是契约：schema 默认值，然后注册方的组合 `base`（其 entry 配置子集），最后用户文档分节。
+
+**镜像 `session-persistence/` 的三包 seam。**`dsh-settings` 拥有抽象 `Settings` 服务：namespace 注册表、分层解析、schema 校验、按 namespace 深相等变更检测，以及 `settings/updated` 提交事件。provider 只实现 `writable`/`load()`/`persist(ns, section)`，并通过受保护的 `publish(doc)` 推入外部观察到的文档——因此热更新语义对所有 provider 一致，网络配置中心后端（nacos 类，可能只读）只是一个平级包的距离。`dsh-settings-local` 是文件 provider：`resolveSpec` 显式默认到 `<DSH_HOME>/settings.yaml` 的 YAML/JSON、chokidar 监听、跨进程写锁下以 `0600` tmp+rename 原子提交的读-改-写 persist、对被写 namespace 的叶子级 diff 修补（未触碰节点的注释得以保留）、按内容相等抑制自写（[write-path integrity note](2026-07-30-settings-write-path-integrity.md)）。
+
+**注册是调用方 fiber 上的 effect。**`register()` 经服务代理调用，`this.ctx` 即注册方 context，注册挂在 `ctx.effect` 上：dispose 注册方即移除 namespace 及其观察者（HMR disposal 测试证明），而用户的分节继续留在存储中等待下一任 owner。
+
+**静止时响亮报错，运行中保留最后可用值。**启动期与注册期校验直接抛错（非法存量分节使注册插件加载失败；存在但不可解析的文档使 provider 加载失败）。运行中坏的外部编辑只告警并按 namespace 保留最后可用状态——热重载绝不拖垮进程。该不对称镜像 `Include.refresh()` 与 Kimi 的安全运行时重载。
+
+**消费者天然可选。**消费者在 `ctx.inject(['settings'], …)` 内注册；不挂 provider 时仍只按 entry 配置解析，因此所有既有组合、demo、snapshot 原样工作，迁移按插件渐进。
+
+## Alternatives considered
+
+- **以 Include 写回为用户层**（cordis-webui 式的按插件配置页写 loader entry 文件）：写回目标是按组合的文件，会把用户偏好绑死在某个 `cordis.yml` 上；用户层必须在模板升级中存活，并以同一文档服务 TUI 与 web。
+- **以 Loader reactive `fiber.update` 为传导通道**：构造期读取毫无感知；seam 的显式 `watch()` 把热更新变成消费者契约而非框架魔法。
+- **领域化的 settings 服务**（按产品域的 getter）：设计评审中的耦合反对成立；服务只做存储、校验、发布——领域含义留给拥有 schema 的注册方。
+- **现在就做多层优先级**（Codex/Claude Code 式 system/managed/project 层级）：延后到真实第二层出现；resolve 步骤是分层未来唯一的扩展点。
+- **现在就上跨进程锁**（Pi 的 proper-lockfile）：最初以"原子替换加 watcher 收敛，真实冲突出现再说"为由延后——评审发现收敛会丢失未观察到的同级 namespace，因此该延后已被 [write-path integrity note](2026-07-30-settings-write-path-integrity.md) 的手写写锁取代。
+
+## 后果
+
+按依赖顺序延后：web `settings.raw`/`settings.describe`/`settings.update` RPC 面（暴露前必须对 `role('secret')` 字段脱敏）；首批消费者迁移（`ui-theme`、语言、api-gateway 默认路由）并退役 `PROFILE_MAPPINGS` 与 profile json；面向密钥的 `${env:VAR}` 值间接引用；provider 侧分层。keyless snapshot 义务随第一个模型或产品用户可见的消费者落地，而非本基础设施步骤。

.agents/notes/implemented/architecture/2026-07-30-settings-write-path-integrity.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-30-settings-write-path-integrity.md
+2026-07-30-settings-write-path-integrity.md: 07bd095162879c8e7866846cf562f6a13307e5fc
+2026-07-30-settings-write-path-integrity.zh.md: 5d02177073d482b61750d7bdfbbd0866bc227a6a

.agents/notes/implemented/architecture/2026-07-30-settings-write-path-integrity.md

@@ -0,0 +1,35 @@
+# Agent Note: settings write-path integrity and observer lifecycle
+
+Status: implemented
+
+English | [中文](2026-07-30-settings-write-path-integrity.zh.md)
+
+> Scope: the third review round over `packages/settings/` — write-path data integrity in `dsh-settings-local` (operation chain, read-modify-write, cross-process writer lock, diff-shaped YAML edits) and observer lifecycle in `dsh-settings` (watch disposal, async listener containment, the JSON-shape write boundary). This note reverses one deferral recorded in the [user-settings seam note](2026-07-28-user-settings-seam.md): the cross-process lockfile now ships.
+
+## Problem
+
+Review found the provider's write path could destroy state it never observed, and the seam's observer lifecycle leaked past disposal. Concretely: watcher reloads and document writes ran on two independent promise chains while every write rendered the whole next document from the cached text, so an external edit still inside the debounce window was overwritten — and the follow-up reload no-oped because the post-rename content matched the cache, erasing the edit without a trace. The initial `load()` raced the watcher's own setup, leaving a startup window whose changes never fire an event. Two processes sharing a harness home rendered from independent caches, last writer winning whole namespaces. On the seam side, a `watch()` disposer only removed the observer from its set — an invocation already chained onto the watcher tail still ran after disposal, and nothing drained started invocations at service dispose; the `settings/updated` manual fan-out caught only synchronous throws, so an async listener's rejection escaped as an unhandled rejection; and `structuredClone` admitted Dates, Maps, BigInts, and cycles that YAML/JSON storage silently distorts on the reload round-trip (a Date lands as a timestamp string, a BigInt as a plain number). YAML writes replaced the whole namespace node, deleting every comment inside the section a comment-preserving provider had promised to keep.
+
+## Decision
+
+**One operation chain, and every write is a read-modify-write.** Watcher refreshes and persists from every namespace queue share a single settled chain, and `persistSection` begins by reconciling the on-disk text into the seam — publishing any unobserved difference first — before rendering against that fresh text. A write can no longer resurrect a stale document, and an on-disk document that turned invalid fails the write loud rather than being overwritten (the reload path keeps its warn-and-keep-last-good policy; the shared `reconcileFromDisk` throws and each caller picks its policy). The watcher's `ready` signal queues one extra reconcile, closing the startup gap between the initial load and the watcher becoming active.
+
+**Writes hold a `wx`-created `<file>.lock` sibling.** The read-render-rename cycle runs under a cross-process writer lock with exponential backoff, a 2 s acquisition deadline, and stale takeover after 5 s (a crashed holder, broken with a warning). Readers never lock — the rename commit is atomic — so contention is writer-only and resolves in milliseconds. The lock constants are protocol invariants, not config: a holder rewrites one small document, so the deadline and stale age derive from that bound, not from deployment taste.
+
+**Observer disposal is quiescent.** Watchers carry an `active` flag checked when a queued invocation would start, so a disposer that ran while the invocation waited prevents the start entirely; started invocations register in a service-level `pendingTails` set that the dispose drain awaits beside the write queues. The `settings/updated` fan-out contains a returned thenable's rejection through the same listener diagnostic as a sync throw, and the event contract now states that the `INVARIANT` rethrow serves synchronous listeners only — invariant companions must stay sync, which the shipped companion already is.
+
+**The write boundary admits JSON data only.** The call-time snapshot is a single `cloneJsonShaped` walk that detaches the patch and rejects any non-JSON value — Date, Map, BigInt, non-finite number, function, symbol, class instance, `undefined` array entry, circular reference — with its `$`-rooted path before anything persists. Object entries that are explicitly `undefined` still skip (the sparse-patch contract), now enforced at the boundary instead of inside `mergeLayers`.
+
+**YAML edits are leaf-level diffs.** `renderYaml` diffs the stored section against the next one and applies only `setIn` for changed values and `deleteIn` for removed keys, recursing through maps. Comments, anchors, and formatting survive on every untouched node and on the key node of every changed pair; arrays and other non-map values replace wholesale when unequal (`deepEqualJson` is the shared predicate), taking comments inside them along.
+
+## Alternatives considered
+
+- **`proper-lockfile` instead of a hand-rolled lock** — the dependency-over-hand-rolling policy was weighed: the library is barely maintained, its stale/retry policy is broader than this one-file protocol needs, and the shipped lock is ~40 lines with deterministic tests (including injected `EEXIST`/`stat` races). The policy favors dependencies that delete owned code; this one would replace 40 explained lines with an opaque peer.
+- **Revision/CAS instead of a lock** — rename cannot express compare-and-swap, so a CAS needs a version sidecar or content re-hash and a retry loop in every writer; the lock achieves the same serialization with one primitive and keeps readers free.
+- **Merging external edits into the in-flight write's own section** — the seam merges patches over the state visible at call time, so a same-namespace external edit racing a write still resolves last-write-wins; folding it in would need three-way merge semantics no consumer has asked for. The write publishes the external state first, so the loser is at least observed before being superseded.
+- **Declaring async `settings/updated` listeners unsupported** — the typed signature is `void` and lint flags misused promises, but an unlinted JS plugin can still register an async listener; a contract note cannot un-throw an unhandled rejection, so containment is the only defense that holds at runtime.
+- **Keeping `structuredClone` and validating in the provider** — the seam is the durable boundary's owner (every provider stores JSON-shaped documents), and rejecting at call time gives the caller the offending path; a provider-side check would reject after merge, blaming the merged section instead of the caller's value.
+
+## Consequences
+
+`update()` gained a documented failure mode (lock deadline, invalid on-disk document) and the rejection messages carry `$`-rooted paths. Remaining, documented in the provider README: same-namespace concurrent edits stay last-write-wins (no per-value merge or revision check), a watcher event the OS never delivers leaves the cache stale until the next signal or write, and comments inside replaced arrays or attached inline to changed scalar values go with the value they described. The [user-settings seam note](2026-07-28-user-settings-seam.md)'s deferred-lockfile alternative is superseded by this note. The same defect classes exist in `dsh-credentials-local` (two chains over one `.env`, cached whole-file write-back, post-persist emit) and in the `llm/adapters-updated` fan-out on the stacked branches; those fixes belong to the PRs that introduce the packages and follow this template on merge-up.

.agents/notes/implemented/architecture/2026-07-30-settings-write-path-integrity.zh.md

@@ -0,0 +1,41 @@
+# Agent Note: settings 写路径完整性与观察者生命周期
+
+Status: implemented
+
+[English](2026-07-30-settings-write-path-integrity.md) | 中文
+
+> 范围：对 `packages/settings/` 的第三轮评审——`dsh-settings-local` 的写路径数据完整性（操作链、读-改-写、跨进程写锁、diff 形态的 YAML 编辑）与 `dsh-settings` 的观察者生命周期（watch 的 dispose（资源释放）、异步监听器收容、JSON 形态写入边界）。本 note 推翻了[用户设置 seam note](2026-07-28-user-settings-seam.md)所记录的一项延后决定：跨进程锁文件现已交付。
+
+## 问题
+
+评审发现，提供方的写路径可能销毁它从未观察到的状态，而 seam 的观察者生命周期会泄漏到 dispose 之后。具体而言：watcher 重载与文档写入跑在两条相互独立的 promise 链上，而每次写入都从缓存文本渲染出完整的下一份文档，于是仍处于防抖窗口内的外部编辑会被覆盖——随后的重载又因 rename 后的内容与缓存一致而成为空操作，这次编辑就被无痕抹去。初始 `load()` 与 watcher 自身的建立过程存在竞态，留下一个启动窗口：落在这个窗口内的变更永远不会触发事件。共享同一 harness home 的两个进程各自从独立的缓存渲染，后写者以整个 namespace 为单位胜出。
+
+在 seam 一侧，`watch()` 的释放器只把观察者从集合中移除——已经接到 watcher 链尾的调用在 dispose 之后照常运行，服务 dispose 时也没有任何环节排空已启动的调用；`settings/updated` 的手动扇出只捕获同步抛错，异步监听器的 rejection 会以 unhandled rejection 的形式逃逸；`structuredClone` 则放行 Date、Map、BigInt 与循环引用，而 YAML/JSON 存储会在重载往返中悄悄扭曲这些值（Date 会变成时间戳字符串，BigInt 会变成普通数字）。
+
+YAML 写入则整体替换 namespace 节点，把分节内的每条注释都删掉——而这个保注释的提供方承诺过要保住它们。
+
+## 决策
+
+**单一操作链，且每次写入都是读-改-写。**watcher 的刷新与来自各 namespace 队列的持久化共享同一条结算链；`persistSection` 会先把磁盘上的文本对账进 seam——任何未被观察到的差异都先发布出去——然后才对照这份新鲜文本渲染。写入不再可能复活一份陈旧文档；磁盘上已变非法的文档会让写入响亮失败，而不是被覆盖（重载路径保持其“告警并保留最后可用值”策略；共享的 `reconcileFromDisk` 抛错，各调用方自选策略）。watcher 的 `ready` 信号会额外排入一次对账，弥合初始加载与 watcher 生效之间的启动缺口。
+
+**写入持有以 `wx` 创建的同目录 `<file>.lock`。**读-渲染-rename 循环在一把跨进程写锁下运行：指数退避、2 s 获取截止时间、5 s 后陈旧接管（持有者已崩溃；打破旧锁时给出告警）。读方从不加锁——rename 提交是原子的——因此竞争只发生在写方之间，毫秒级即可化解。锁的各项常量是协议不变式，不是配置：持有者只是重写一份小文档，截止时间与陈旧时限都从这一上界推得，而非出自部署偏好。
+
+**观察者 dispose 达到完全停稳。**watcher 携带一个 `active` 标志，排队的调用即将启动时先检查它，因此在调用等待期间已经运行过的释放器能让这次启动彻底不发生；已启动的调用会登记进服务级的 `pendingTails` 集合，dispose 排空除了等待各写队列，还会等待该集合。`settings/updated` 扇出会把监听器返回的 thenable 的 rejection 收容进与同步抛错相同的监听器诊断；事件契约现已写明 `INVARIANT` 重抛只服务同步监听器——不变式配套插件必须保持同步，而已交付的那个配套插件本就是同步的。
+
+**写入边界只放行 JSON 数据。**调用时刻的快照就是一次 `cloneJsonShaped` 遍历：它把 patch 从调用方分离出来，并在任何内容持久化之前拒绝一切非 JSON 值——Date、Map、BigInt、非有限数值、函数、symbol、类实例、值为 `undefined` 的数组元素、循环引用——拒绝时附带该值以 `$` 为根的路径。显式为 `undefined` 的对象条目仍会跳过（稀疏 patch 契约），这一契约如今在边界处强制执行，而不再放在 `mergeLayers` 内部。
+
+**YAML 编辑是叶子级 diff。**`renderYaml` 对比已存储分节与下一份分节，只对变化的值应用 `setIn`、对移除的键应用 `deleteIn`，并沿 map 递归。注释、锚点与格式在每个未触碰节点上、以及每个被改键值对的键节点上全部保留；数组等非 map 值在不相等时整体替换（`deepEqualJson` 是共享的判定谓词），其内部注释随之一并被带走。
+
+## 曾考虑的替代方案
+
+- **用 `proper-lockfile` 取代手写锁**——按“依赖优先于手写”政策做过权衡：该库几乎无人维护，其陈旧／重试策略比这个单文件协议所需的更宽泛，而已交付的锁约 40 行并带确定性测试（含注入的 `EEXIST`/`stat` 竞态）。该政策偏向能删除自有代码的依赖；这个依赖只会把 40 行带解释的代码换成一个不透明的等价物。
+- **用修订号/CAS 取代锁**——rename 表达不了 compare-and-swap，因此 CAS 需要一个版本伴随文件或内容重哈希，外加每个写方里的一个重试循环；锁用一个原语实现同样的串行化，还让读方完全免锁。
+- **把外部编辑合并进正在进行的写入自身的分节**——seam 是在调用时刻可见的状态之上合并 patch 的，因此与写入竞态的同 namespace 外部编辑仍按后写胜出解决；要把外部编辑并进来，需要三方合并语义，而没有任何消费方提出过这种需求。写入会先发布外部状态，落败一方至少在被取代之前被观察到。
+- **宣布不支持异步 `settings/updated` 监听器**——类型签名是 `void`，lint 也会标记误用的 promise，但未经 lint 的 JS 插件仍能注册异步监听器；契约里的一句说明无法收回已经抛出的 unhandled rejection，收容是唯一在运行时守得住的防线。
+- **保留 `structuredClone`、在提供方里做校验**——seam 才是持久化边界的所有者（每个提供方存储的都是 JSON 形态文档），而且在调用时刻拒绝能把违规值的路径给到调用方；提供方侧的检查要到合并之后才拒绝，归咎的是合并后的分节，而不是调用方传入的值。
+
+## 后果
+
+`update()` 有了成文的失败模式（锁截止时间到期、磁盘文档非法），rejection 消息携带以 `$` 为根的路径。仍然存在、且已记录在提供方 README 中的有：同 namespace 并发编辑仍是后写胜出（没有逐值合并，也没有修订号检查）；OS 从未投递的 watcher 事件会让缓存保持陈旧，直到下一个信号或下一次写入；被替换数组内部的注释、以及行内附着在被改标量值上的注释，会随其描述的值一起消失。
+
+[用户设置 seam note](2026-07-28-user-settings-seam.md)里“延后锁文件”那条替代方案已被本 note 取代。同类缺陷还存在于 `dsh-credentials-local`（两条链共用一个 `.env`、按缓存整文件写回、持久化之后才发事件）与堆叠分支上的 `llm/adapters-updated` 扇出；这些修复归引入相应包（package）的那些 PR（Pull Request）所有，向上合并时按本模板处理。

.agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.md
-2026-07-28-themed-scrollbars-and-reserved-gutter.md: 38228c868bb00210118e8110feb722fb81d0d56c
-2026-07-28-themed-scrollbars-and-reserved-gutter.zh.md: 9fafe1faa9303b5e2e23a1b3064904f71494026d
+2026-07-28-themed-scrollbars-and-reserved-gutter.md: b45f70b126d083916c756afb88a8b646a4e9bb85
+2026-07-28-themed-scrollbars-and-reserved-gutter.zh.md: 8afa36429ce7e6e061b014d63dcb20e5a642a84c

.agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.md

@@ -20,13 +20,13 @@ The rules sit on `body`, not `html`. `design-platform.css` declares the `--dsw-a
 
 The two renderings are mutually exclusive, and the exclusion is enforced rather than assumed. A non-`auto` `scrollbar-width` or `scrollbar-color` makes Chromium and Safari discard every `::-webkit-scrollbar*` rule for that element, `::-webkit-scrollbar-thumb:hover` included. Declaring both unconditionally therefore leaves the hover token rendering nowhere at all: the engines that implement the hover pseudo-element are exactly the ones the standard properties silence, and Firefox has no hover pseudo-element to fall back on. The standard properties consequently sit inside `@supports not selector(::-webkit-scrollbar)`, which is true only where the pseudo-element is unimplemented, so Firefox takes the standard path and WebKit-based engines take the pseudo-element path. The WebKit rules are not gated in turn: an engine without those pseudo-elements drops them as unknown selectors, so a gate would only restate what selector matching already does. An engine too old for the `selector()` function makes the condition invalid, which evaluates false and selects the pseudo-element path — the correct side for the pre-16.4 Safari that is the realistic case for that reading.
 
-Both paths read one indirection pair, `--dsh-scrollbar-thumb` and `--dsh-scrollbar-thumb-hover`, bound on `body` to the l1 (base-surface) tokens. **This is the rebinding contract, and it is the part the CSS alone does not state**: an elevated surface sets `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` and `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)` on its own container, and that one rebind reaches the standard properties and the WebKit pseudo-elements together. The pair is rebound as a pair; rebinding the resting thumb alone leaves the hover state on the base-surface token. Eight surfaces rebind today: the command popup, the slash menu, the model-select panel, the settings panel, the shared `ui-primitives` menu card, the composer input card, the question composer card, and the todo panel. Most declare it on the elevated card rather than on the scrolling descendant, because the elevation is a property of the surface and custom properties inherit down to whichever child actually scrolls.
+Both paths read one indirection pair, `--dsh-scrollbar-thumb` and `--dsh-scrollbar-thumb-hover`, bound on `body` to the l1 (base-surface) tokens. **This is the rebinding contract, and it is the part the CSS alone does not state**: an elevated surface sets `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` and `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)` on its own container, and that one rebind reaches the standard properties and the WebKit pseudo-elements together. The pair is rebound as a pair; rebinding the resting thumb alone leaves the hover state on the base-surface token. The mechanically discoverable subset is owned by `packages/client/ui-theme/tests/scrollbar-styles.spec.ts`: any sheet that both scrolls and paints an elevated surface must rebind, so this note no longer maintains a complete surface inventory. Most declare the pair on the elevated card rather than on the scrolling descendant, because elevation belongs to the surface and custom properties inherit to whichever child actually scrolls.
 
-The last four were missed in the first implementation and found in review, which is why the rebinding contract is now checked mechanically rather than by inspection: a sheet that scrolls somewhere and paints an elevated surface somewhere must rebind.
+Four surfaces — `Menu`, `InputBar`, `QuestionComposer`, and `TodoPanel` — were missed in the first implementation and found in review, which is why the per-sheet rebinding contract is checked mechanically rather than by inspection.
 
 The elevated set is resolved from the palette's own dark elevation ladder — the surface tokens whose dark value lands on `bg-layer-2` or `bg-layer-3`, which is the step the l1/l2 split encodes. Deriving it instead from the sheets that already rebind was the first attempt and is unsound: such a set can only confirm what someone already remembered, and a surface nobody has rebound yet — exactly the case the check exists for — defines itself as unelevated. `--dsw-specific-tip` proved it, resolving to the menu surface's rung while the todo panel scrolled on it unrebound and the derived check stayed green.
 
-Scope is by token family, not by geometry: only `--dsw-alias-bg-*` and `--dsw-specific-*` name a surface. `--dsw-alias-button-*`, `--dsw-alias-interactive-*`, and `--dsw-alias-markdown-*` reach the same rungs while naming a control or an inline span that no scroll container renders its bar against. Shape cannot make that call, since a floating button legitimately carries a radius, a shadow, and a fixed size. The check is per sheet rather than per rule because the card and the descendant that scrolls are separate rules, and CSS text does not express which contains which.
+Scope is by token family, not by geometry: only `--dsw-alias-bg-*` and `--dsw-specific-*` name a surface. `--dsw-alias-button-*`, `--dsw-alias-interactive-*`, and `--dsw-alias-markdown-*` reach the same rungs while naming a control or an inline span that no scroll container renders its bar against. Shape cannot make that call, since a floating button legitimately carries a radius, a shadow, and a fixed size. The check is per sheet rather than per rule because the card and the descendant that scrolls are separate rules. That approximation cannot detect a scrolling component embedded in an elevated card painted by another package's stylesheet, as `DirectoryBrowser` inside `Modal` demonstrated; cross-sheet composition remains a review and assembled-UI responsibility.
 
 The track and the corner stay transparent, so the thumb reads against whatever surface scrolls under it; only the thumb and its hover state carry a token color.
 

.agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.zh.md

@@ -20,13 +20,13 @@ Status: implemented
 
 两种渲染互斥，而这种互斥是被强制的，不是假定的。`scrollbar-width` 或 `scrollbar-color` 只要取非 `auto` 值，Chromium 与 Safari 就会丢弃该元素上的全部 `::-webkit-scrollbar*` 规则，`::-webkit-scrollbar-thumb:hover` 也在其中。因此无条件地同时声明会让 hover token 在任何地方都得不到渲染：实现了 hover 伪元素的引擎，恰恰就是被标准属性静音的那些，而 Firefox 没有 hover 伪元素可作退路。于是标准属性写在 `@supports not selector(::-webkit-scrollbar)` 之内，该条件只在伪元素未被实现处为真，因此 Firefox 走标准属性路径，WebKit 系引擎走伪元素路径。WebKit 规则不再反向加门禁：不实现这些伪元素的引擎会把它们当作未知选择器丢弃，因此加门禁只是重述选择器匹配本身已经做的事。对于旧到不支持 `selector()` 函数的引擎，该条件无效，从而求值为假并选中伪元素路径——对于这条判断下现实存在的 16.4 之前的 Safari，这正是正确的一侧。
 
-两条路径都读取同一组间接变量 `--dsh-scrollbar-thumb` 与 `--dsh-scrollbar-thumb-hover`，它们在 `body` 上绑定到 l1（基础表面）token。**这就是重新绑定契约，也是单看 CSS 无法得知的部分**：抬升表面在自己的容器上设置 `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` 与 `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)`，这一次重新绑定同时作用于标准属性和 WebKit 伪元素。这组变量必须成对重新绑定；只改静止态滑块会让 hover 状态仍留在基础表面的 token 上。目前有八处抬升表面做了重新绑定：命令浮层、斜杠菜单、模型选择面板、设置面板、`ui-primitives` 共用菜单卡片、输入条卡片、提问组件卡片与待办面板。多数把声明写在抬升卡片上而非滚动的后代元素上，因为抬升层级是这个表面的属性，而自定义属性会继承到真正滚动的那个子元素。
+两条路径都读取同一组间接变量 `--dsh-scrollbar-thumb` 与 `--dsh-scrollbar-thumb-hover`，它们在 `body` 上绑定到 l1（基础表面）token。**这就是重新绑定契约，也是单看 CSS 无法得知的部分**：抬升表面在自己的容器上设置 `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` 与 `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)`，这一次重新绑定同时作用于标准属性和 WebKit 伪元素。这组变量必须成对重新绑定；只改静止态滑块会让 hover 状态仍留在基础表面的 token 上。可由机械检查发现的子集归 `packages/client/ui-theme/tests/scrollbar-styles.spec.ts` 所有：任何既滚动又绘制抬升表面的样式表都必须重新绑定，因此本 note 不再维护完整的表面清单。多数把这组变量声明在抬升卡片上而非滚动的后代元素上，因为抬升层级属于这个表面，而自定义属性会继承到真正滚动的那个子元素。
 
-后四处在最初的实现里被漏掉、由评审发现，因此重新绑定契约现在由机械检查把关，而不再依赖人工审阅：一张样式表只要在某处滚动、又在某处绘制抬升表面，就必须重新绑定。
+`Menu`、`InputBar`、`QuestionComposer` 与 `TodoPanel` 这四个表面在最初的实现里被漏掉、由评审发现，因此逐样式表的重新绑定契约由机械检查而非人工审阅把关。
 
 抬升表面集合是从调色板自身的暗色抬升阶梯解析出来的——暗色取值落在 `bg-layer-2` 或 `bg-layer-3` 上的那些表面 token，而这一档正是 l1/l2 之分所编码的层级差。最初的做法是从已经做了重新绑定的样式表反向推导，那是不成立的：这样得到的集合只能确认别人已经记得的部分，而尚无人重新绑定的表面——恰恰就是这项检查存在的理由——会把自己定义成「非抬升」。`--dsw-specific-tip` 证明了这一点：它解析到与菜单表面相同的那一档，待办面板在它上面滚动却没有重新绑定，而推导式的检查依然是绿的。
 
-判定范围依据 token 家族而非几何形状：只有 `--dsw-alias-bg-*` 与 `--dsw-specific-*` 表述的是表面。`--dsw-alias-button-*`、`--dsw-alias-interactive-*` 与 `--dsw-alias-markdown-*` 会落到相同档位，但它们表述的是控件或行内片段，没有任何滚动容器会把滚动条画在它们之上。形状无法做这个判断，因为悬浮按钮本来就会带圆角、阴影和固定尺寸。这项检查以样式表为粒度而非以规则为粒度，因为卡片与真正滚动的后代元素是两条不同的规则，而 CSS 文本无法表达谁包含谁。
+判定范围依据 token 家族而非几何形状：只有 `--dsw-alias-bg-*` 与 `--dsw-specific-*` 表述的是表面。`--dsw-alias-button-*`、`--dsw-alias-interactive-*` 与 `--dsw-alias-markdown-*` 会落到相同档位，但它们表述的是控件或行内片段，没有任何滚动容器会把滚动条画在它们之上。形状无法做这个判断，因为悬浮按钮本来就会带圆角、阴影和固定尺寸。这项检查以样式表为粒度而非以规则为粒度，因为卡片与真正滚动的后代元素是两条不同的规则。这种近似检查无法检测嵌在由另一个包的样式表绘制的抬升卡片中的滚动组件，`Modal` 内的 `DirectoryBrowser` 就证明了这一点；跨样式表的组合仍需在评审和组装后 UI 层面把关。
 
 轨道与两条滚动条相交的角落保持透明，因此滑块是以其下滚动的任何表面为背景被看到；只有滑块及其 hover 状态带 token 颜色。
 

.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md
+2026-07-29-human-transcript-append-origin.md: a296b93d538d9c28bd61ee8fd0530863b4bfd878
+2026-07-29-human-transcript-append-origin.zh.md: 96e0cd1038fe8904dfd4c1eceaae9b25339c5dca

.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.md

@@ -0,0 +1,53 @@
+# Agent Note: The human transcript projects append-origin events
+
+Status: implemented
+
+English | [中文](2026-07-29-human-transcript-append-origin.zh.md)
+
+## Problem
+
+The terminal and the host history gateway both treated the model-visible surface as the human transcript. A successful compaction replaces a surface range with one checkpoint node, so the moment that replacement landed the terminal dropped every message it shadowed — conversation the user had already read — and re-ran that destructive rebuild on any later replacement. The same confusion reached pagination: `maxMessages` counted every `user/message`, `assistant/message`, and `steering/message` in the window, so a model-only replacement copy consumed a page slot the human never filled, and the cut could land between a compaction's log-only provenance and the replacement that cites it.
+
+Nothing was lost from the log. `Session.events` still held every original message and full tool result; the surface only decides what the model is sent next. The defect was entirely in the projection.
+
+## Decision
+
+Model and human projections are separate, and the event's own marker decides which one an event belongs to. `dsh-session` exports the marker split `isAppendSurfaceEvent(event)` and `isReplacementSurfaceEvent(event)` over the two `SurfaceOp` variants, from the browser-safe `surface` module. Append-origin events are the durable source for a transcript; replacement copies stay model-only. Everything that must send exactly what the model sees — `deriveMessages`, token accounting, the compaction backends, tool pairing, injected-context liveness, cross-session reference projection — keeps reading `session.surface`.
+
+The terminal replays the transcript from append-origin surface events and keeps a shadowed step's tool cards paired through `transcriptToolCallIds`, which reads the append-origin `assistant/message` rather than surface membership. A landed compaction contributes one dim `… earlier context was compacted …` row at its own log position: the marker reports where the model stopped seeing that history instead of erasing it. The framed checkpoint payload never renders, and both paths classify a surface event by the same marker, so a compaction that arrives live and the same log replayed after resume produce the same transcript. Only replay re-derives `tool/call` pairing: a call event carries no marker of its own and inherits membership from the `assistant/message` that advertised it, which the live listener has necessarily just rendered.
+
+A checkpoint is recognized through the compaction seam's own contract — `isCompactCheckpointSource`, the backend-independent marker `CompactService` requires on the replacement user message — so the terminal depends on the declared vocabulary, not on the shape of the replacement. `dsh-session-reference` already consumes that predicate to project another session's log; this is the same question asked by a different reader. Other replacements are silent: a pruned `tool/result` and a regenerated `assistant/message` rewrite one node for the model and mark no boundary in the conversation.
+
+`session.history` counts only append-origin messages toward `maxMessages`. Each page remains one contiguous raw event range, so a compaction's `compact/summary` provenance stays on the page of the replacement that cites it.
+
+No persisted event, RPC envelope, compaction transaction, or model-visible surface changed, and no migration is required.
+
+## Deferred
+
+The browser client still builds its conversation from the model surface through `FoldAdapter`, so compaction still collapses web history to a single context row. The same predicate is the fix there, together with an append-order transcript projection and a marker component; that work is a separate change against `packages/client/runtime` and `packages/client/ui-conversation`.
+
+That work must handle a page whose checkpoint cites a `surfaceOp.start` outside the window: pagination no longer spends quota on the checkpoint, so it never cuts on the checkpoint's provenance group, and `FoldAdapter` pads absent events with a non-surface sentinel — so `SurfaceManager` rejects the range and `nodes()` falls back to `degradedSeqs()` with a logged error. The hole predates this change (counting could already run past a checkpoint into the range it shadows), but the old rule accidentally covered the case where the checkpoint was the oldest counted message and pulled the whole shadowed range onto its page. `degradedSeqs()` — every surface-eligible event in append order — is already close to the transcript projection A2 needs, which is the shape to build deliberately rather than reach as a degradation. Rendering compaction *progress* — a terminal indicator while a compaction runs — needs the bracket-first ordering that the queued manual `/compact` work introduces, and is likewise out of scope here. The marker also carries no scale: the checkpoint's `sourceEventSeqs` already hold the shadowed count, so a count or range would tell a reader how much each row folded. That belongs with progress, where the reader meets the other half of the same information. Whoever takes it should fold the terminal's two replacement branches — replay and the live listener, textually identical and 600 lines apart — into one `renderReplacement(event)` first, so the marker's content has a single home.
+
+## Alternatives considered
+
+**Recognize a checkpoint by shape (a replacement `user/message`).** Rejected: it reads a coincidence of today's producers instead of a declared contract, and any future producer that replaces a range with a user message would silently inherit the compaction marker. The seam already publishes `COMPACT_CHECKPOINT_SOURCE` precisely so consumers can recognize a checkpoint independently of the backend.
+
+**Keep rendering the checkpoint as an injected-context card.** Rejected: the framed checkpoint is an instruction envelope written for the model, not human conversation content. Showing it while hiding the history it replaced inverts what the reader needs.
+
+**Persist a second display transcript.** Rejected: the append-only log already contains the authoritative source material, so a parallel record buys nothing and adds migration and consistency work.
+
+**Derive the marker from the `compact/*` bracket instead of the checkpoint.** Rejected for the transcript: the bracket is a pair of time-point markers around an operation, while the transcript needs the position where the surface actually changed. The bracket is the right source for progress and duration, which this change does not render.
+
+**Classify events by re-folding the log, as `session-query` does for search (`current` / `shadowed` / `log-only`).** Rejected: a fold answers a whole-log question, while a projection asks a per-event one that the event's own marker already answers in constant time.
+
+## Consequences
+
+Compaction no longer erases terminal history; a session compacted several times shows one marker per landed compaction, in log order. Pagination pages can carry more raw events than before, because quota is spent only on messages a human or model actually produced.
+
+`rebuildTranscript` now materializes a component per append-origin event in the whole log, and it runs on mount, on a terminal color-scheme change, and on every reasoning toggle. Compaction used to bound that work for exactly the long sessions compaction serves, so the cost now grows with session length instead of with the surface. That is the trade the fix exists to make — preserved history is the point — but a windowing or reuse strategy belongs to whoever first measures a slow rebuild, not to a later profiler wondering why the work grew.
+
+`dsh-tui` gains a dependency on the `dsh-compact` seam for one pure predicate, mirroring `dsh-session-reference`'s existing use. The terminal still needs no compaction backend at runtime.
+
+Two behaviors changed with their tests. The surface-replacement terminal test previously pinned erasure ("hides shadowed tool calls") and now pins preservation plus exactly one marker, including a pruned result copy, a regenerated assistant message, and a foreign plugin's replacement all rendering nothing. The compaction snapshot scenario wrote a `workspace-context` source while claiming to pin compaction; it now writes a real checkpoint source, and its three fixtures are re-recorded to show the preserved prompt, the full tool card, and the marker.
+
+The live/replay equivalence above is fixture-pinned, not only asserted here: `surface-replayed-compaction` mounts with the replacement already stored and records byte-identical to the live path's `surface-after-compaction-wide`. Changing either path breaks that equality, which is the point — the resume projection is what regressed for users, and the two fixtures must move together.

.agents/notes/implemented/bug-fix/2026-07-29-human-transcript-append-origin.zh.md

@@ -0,0 +1,53 @@
+# Agent Note: 人类可读记录投影追加来源的事件
+
+Status: implemented
+
+[English](2026-07-29-human-transcript-append-origin.md) | 中文
+
+## Problem
+
+终端与宿主历史网关都把模型可见的 surface 当作人类可读记录（transcript）。一次成功的压缩（compaction）会用一个检查点节点替换一段 surface 范围，因此该替换一落地，终端就丢弃了它所遮蔽的每条消息——那些是用户已经读过的对话——并在此后任何替换到来时重新执行这次破坏性重建。同样的混淆也波及分页：`maxMessages` 统计窗口内的每个 `user/message`、`assistant/message` 和 `steering/message`，于是仅供模型使用的替换副本占用了一个人类从未填充的页面额度，而切分点还可能落在压缩的仅日志溯源信息与引用它的替换之间。
+
+日志本身没有丢失任何内容。`Session.events` 仍保存着每条原始消息和完整的工具结果；surface 只决定接下来发送给模型的内容。缺陷完全在投影层。
+
+## Decision
+
+模型投影与人类投影是分开的，而事件属于哪一种由事件自身的标记决定。`dsh-session` 在浏览器安全的 `surface` 模块中导出按两种 `SurfaceOp` 变体划分的谓词 `isAppendSurfaceEvent(event)` 与 `isReplacementSurfaceEvent(event)`。追加来源的事件是记录的持久来源，替换副本仅供模型使用。凡是必须准确发送模型所见内容的部分——`deriveMessages`、token 记账、压缩后端、工具配对、注入上下文的存活判断、跨会话引用投影——都继续读取 `session.surface`。
+
+终端从追加来源的 surface 事件回放记录，并通过 `transcriptToolCallIds` 让被遮蔽步骤的工具卡片保持配对：该函数读取追加来源的 `assistant/message`，而不是 surface 成员关系。已落地的压缩会在其自身日志位置贡献一行暗色 `… earlier context was compacted …`：这行标记报告模型从何处起不再看到那段历史，而不是把它抹掉。带框的检查点载荷从不渲染，且两条路径都按同一个标记对 surface 事件分类，因此实时到达的压缩与恢复后回放同一份日志会产生相同的记录。只有回放会重新推导 `tool/call` 的配对关系：调用事件自身不携带标记，其归属继承自公布它的 `assistant/message`，而实时监听器必然刚刚渲染过后者。
+
+检查点通过压缩接缝自身的契约来识别——`isCompactCheckpointSource`，即 `CompactService` 要求替换用户消息携带的、与后端无关的标记——因此终端依赖的是已声明的词汇，而不是替换的形态。`dsh-session-reference` 已经在用该谓词投影另一个会话的日志；这里只是另一个读者提出同样的问题。其他替换保持静默：被裁剪的 `tool/result` 与重新生成的 `assistant/message` 只是为模型重写一个节点，并不在对话中标出边界。
+
+`session.history` 只把追加来源的消息计入 `maxMessages`。每一页仍是一段连续的原始事件区间，因此压缩的 `compact/summary` 溯源信息会与引用它的替换留在同一页。
+
+持久事件、RPC 信封、压缩事务与模型可见的 surface 都没有变化，也不需要迁移。
+
+## Deferred
+
+浏览器客户端仍通过 `FoldAdapter` 从模型 surface 构建会话，因此压缩在 Web 端仍会把历史折叠成一行上下文。那里的修复用的是同一个谓词，另需按追加顺序的记录投影与一个标记组件；该工作是针对 `packages/client/runtime` 与 `packages/client/ui-conversation` 的独立变更。
+
+该工作必须处理这样一页：其检查点引用的 `surfaceOp.start` 落在窗口之外。分页不再为检查点消耗额度，因此永远不会按检查点的溯源分组切分；而 `FoldAdapter` 会用一个非 surface 的哨兵事件填补缺失事件——于是 `SurfaceManager` 拒绝该范围，`nodes()` 退化为 `degradedSeqs()` 并记录一条错误。这个缺口早于本次变更（此前计数就可能越过检查点进入它所遮蔽的范围），但旧规则恰好覆盖了这样一种情形：检查点是最旧的被计数消息，其溯源分组把整段被遮蔽的范围一起拉到该页。`degradedSeqs()`——按追加顺序的每个 surface 可入事件——已经很接近 A2 所需的记录投影，因此那正是应当刻意构建的形态，而不是作为退化路径被动落到的结果。渲染压缩*进度*——压缩运行期间的终端指示——需要排队式手动 `/compact` 工作引入的“先开括号”顺序，同样不在本次范围内。标记同样不携带规模信息：检查点的 `sourceEventSeqs` 已经包含被遮蔽的数量，因此一个计数或区间可以告诉读者每一行折叠了多少内容。这件事属于进度那一侧，读者正是在那里遇到同一份信息的另一半。接手者应当先把终端里两处替换分支——回放与实时监听器，文本完全相同却相隔 600 行——合并为一个 `renderReplacement(event)`，让标记的内容只有一个归处。
+
+## Alternatives considered
+
+**按形态识别检查点（一个替换型 `user/message`）。** 被否决：那读取的是当前生产者的巧合而非已声明的契约，而未来任何用用户消息替换一段范围的生产者都会静默地继承压缩标记。接缝已经发布 `COMPACT_CHECKPOINT_SOURCE`，正是为了让消费方与后端无关地识别检查点。
+
+**继续把检查点渲染为注入上下文卡片。** 被否决：带框的检查点是为模型撰写的指令信封，不是人类对话内容。展示它却隐藏它替换掉的历史，正好颠倒了读者的需要。
+
+**持久化第二份展示用记录。** 被否决：仅追加的日志已经包含权威源材料，平行记录换不来任何东西，反而增加迁移与一致性工作。
+
+**用 `compact/*` 括号而不是检查点来推导标记。** 就记录而言被否决：括号是围绕一次操作的一对时间点标记，而记录需要的是 surface 真正发生变化的位置。括号适合作为进度与耗时的来源，而本次变更并不渲染这些。
+
+**像 `session-query` 为搜索所做的那样重新折叠日志来分类事件（`current`／`shadowed`／`log-only`）。** 被否决：折叠回答的是整份日志的问题，而投影问的是逐事件的问题，事件自身的标记已能以常数时间给出答案。
+
+## Consequences
+
+压缩不再抹掉终端历史；被压缩多次的会话会按日志顺序显示每次落地压缩对应的一行标记。分页的每一页可以携带比以前更多的原始事件，因为额度只花在人类或模型真正产生的消息上。
+
+`rebuildTranscript` 现在会为整份日志中的每个追加来源事件物化一个组件，并在挂载时、终端配色方案变化时以及每次切换 reasoning 时运行。压缩此前正好为压缩所服务的那些长会话限制了这项工作量，因此这份开销现在随会话长度增长，而不再随 surface 增长。这正是本次修复要做的取舍——保留历史才是目的——但窗口化或复用策略属于第一个真正测到重建变慢的人，而不属于日后某个疑惑工作量为何增长的性能分析者。
+
+`dsh-tui` 为一个纯谓词新增了对 `dsh-compact` 接缝的依赖，与 `dsh-session-reference` 现有用法一致。终端在运行时仍然不需要任何压缩后端。
+
+两项行为随其测试一起改变。表层替换的终端测试此前钉住的是抹除（“隐藏被遮蔽的工具调用”），现在钉住的是保留加恰好一行标记，其中被裁剪的结果副本、重新生成的 assistant 消息以及来自其他插件的替换都不渲染任何内容。压缩快照场景此前声称钉住压缩，却写入了 `workspace-context` 来源；现在它写入真实的检查点来源，并重新录制三份 fixture，以显示被保留的提示、完整的工具卡片和那行标记。
+
+上文的实时／回放等价性由 fixture 钉住，而不只是在此断言：`surface-replayed-compaction` 在挂载时替换已经存在，其录制结果与实时路径的 `surface-after-compaction-wide` 逐字节一致。改动任一路径都会破坏这项相等——这正是要点：回放投影才是当初对用户造成回归的部分，两份 fixture 必须一起变动。

.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md
-2026-07-08-self-referential-cordis-toolset.md: 40934fe0e2975c4e068df6ef8f31ed7921df3230
-2026-07-08-self-referential-cordis-toolset.zh.md: 13662b9359aa85895ce85391ebd5a5902cc451cc
+2026-07-08-self-referential-cordis-toolset.md: 335d5e808016ebc37c8457c5dcf4d7da9d8b8c93
+2026-07-08-self-referential-cordis-toolset.zh.md: 8d2a105a2189aa23915f718c6440069d1a2aa9ed

.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md

@@ -12,7 +12,7 @@ First, model-written registration must be validated where it happens: a malforme
 
 ## Decision
 
-The toolset ships as [`@deepseek-ai/dsh-tool-cordis`](../../../../packages/cordis/tool-cordis/README.md) and is demoed by [`examples/cordis-agent`](../../../../examples/cordis-agent/README.md). It gives the model three tools over the live Cordis runtime in the current DSH process: inspect it, mount an in-memory temporary Plugin, and unmount that Plugin to quiescence.
+The toolset ships as [`@deepseek-ai/dsh-tool-cordis`](../../../../packages/cordis/tool-cordis/README.md) and is demoed by `examples/web-cordis`. It gives the model three tools over the live Cordis runtime in the current DSH process: inspect it, mount an in-memory temporary Plugin, and unmount that Plugin to quiescence.
 
 The vm isolates accidental global pollution, and the context façade hides framework internals. Neither restricts the authority of exposed services: a temporary Plugin can call `ctx.bash` with the host executor's privileges and reach the real filesystem and web services. It runs in the shared DSH runtime and may affect other sessions in that process. This is an opt-in development tool with bash-equivalent trust, not a security boundary or product default.
 

.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.zh.md

@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-该工具集以 [`@deepseek-ai/dsh-tool-cordis`](../../../../packages/cordis/tool-cordis/README.md) 发布，并由 [`examples/cordis-agent`](../../../../examples/cordis-agent/README.md) 演示。它为模型提供三个工具，操作当前 DSH 进程中的活跃 Cordis 运行时：审视它、挂载一个仅存于内存的临时 Plugin，再将该 Plugin 卸载至完全停稳。
+该工具集以 [`@deepseek-ai/dsh-tool-cordis`](../../../../packages/cordis/tool-cordis/README.md) 发布，并由 `examples/web-cordis` 演示。它为模型提供三个工具，操作当前 DSH 进程中的活跃 Cordis 运行时：审视它、挂载一个仅存于内存的临时 Plugin，再将该 Plugin 卸载至完全停稳。
 
 vm 隔离了意外的全局污染，上下文门面隐藏了框架内部细节。但二者都不限制已暴露服务的权限：临时 Plugin 可以调用 `ctx.bash` 以宿主执行器的权限运行命令，也能访问真实的文件系统和网络服务。它运行在共享 DSH runtime 中，可能影响同一进程的其他 session。这是一个需要显式启用的开发工具，信任等级与 bash 相当，不是安全边界，也不是产品默认配置。
 

.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
-2026-07-17-dedicated-full-screen-tui-front-door.md: a3f3d6b51e85ad20218ce5aebf526bd96946be55
-2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 6a0c2f12815e9418a2b5bcb237d3db3616ccd133
+2026-07-17-dedicated-full-screen-tui-front-door.md: c011a0284ea0efe59693785c038f814a866068ac
+2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 5aea4ac0c5a3b29c098c297e514fab49caf643ff

.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md

@@ -14,17 +14,17 @@ The interactive channel must remain a Cordis plugin over the same agent, session
 
 DeepSeek Harness ships [`@deepseek-ai/dsh-tui`](../../../../packages/ui/tui/README.md) as a dedicated Cordis plugin. It owns terminal input and presentation only; agent lifecycle, session persistence, tool execution, and the model-facing question tool remain separate composition entries. The plugin requires both stdin and stdout to be TTYs and fails instead of silently changing to line-oriented behavior.
 
-The app layer has one terminal front door. `@deepseek-ai/dsh-tui-demo` mounts the TUI before the configured agent, and `examples/tui-agent` owns the interactive coding composition and Code Mode overlay directly. Non-interactive tasks use `@deepseek-ai/dsh-cli-demo`; ACP remains a separate automation protocol.
+There is one terminal front door. `@deepseek-ai/dsh-tui` mounts before the configured agent, and `apps/cli/config/tui.cordis.yml` — an overlay over the shared `base.cordis.yml` — owns the interactive coding composition. Non-interactive tasks use the official headless surface; ACP remains a separate automation protocol and owns the supported Code Mode demo.
 
 The selected front door receives the exact generated or resumed `SessionId` used by the pre-created agent. It mounts before the agent composition, waits for the matching root agent, and enters full-screen mode only after that agent exists. A matching `agent-loop/config-start-failed` event is therefore reported before screen takeover and exits with status 1.
 
 ### Session projection and interaction
 
-The TUI rebuilds the transcript from the active `session.surface` and reprojects it whenever an event carries a `surfaceOp`, so resumed and compacted history matches the model-visible conversation. It renders Markdown text and reasoning, including fenced code with hidden Markdown markers, a dim optional language label, and a code-colored body, token totals, the latest `todo/write` plan, and tool cards produced through each tool definition's `presentCall` and `presentResult` methods. Long card bodies retain a configurable head/tail preview with the hidden-line count; one terminal control expands or collapses every card. Pending chunks and tool calls update the same components that completed events settle.
+The TUI rebuilds the transcript from the append-origin session events, so resumed history keeps every message the reader already saw; a compacted range stays readable behind one marker instead of matching the model-visible conversation ([append-origin transcript](../bug-fix/2026-07-29-human-transcript-append-origin.md)). It renders Markdown text and reasoning, including fenced code with hidden Markdown markers, a dim optional language label, and a code-colored body, token totals, the latest `todo/write` plan, and tool cards produced through each tool definition's `presentCall` and `presentResult` methods. Long card bodies retain a configurable head/tail preview with the hidden-line count; one terminal control expands or collapses every card. Pending chunks and tool calls update the same components that completed events settle.
 
 Editor input calls `agent.send()` while idle and `agent.steer()` while a turn is running. Cancellation, reasoning visibility, tool-card expansion, redraw, transcript clearing, and exit are terminal-only controls. `/exit` and `/quit` share the same exit path: they cancel an active turn, wait for idle, and then restore and close the terminal. The idle footer derives context occupancy from `tokenMeter` and shows the selected model and explicit reasoning effort; during a run, elapsed activity and the Escape interrupt hint replace that summary. `/status` remains available in either state and appends a detailed terminal-only snapshot: session identity and timestamps, selected model, reasoning effort/default state and reasoning visibility, lifecycle counts folded from the event log, the same deduplicated usage buckets and KV-cache rate as the footer, and context use from `tokenMeter` plus the selected model's advertised capacity. The plugin registers the shared `userInteraction` provider and presents queued questions in a wide bottom-left keyboard panel with batch progress, numbered options, and aligned descriptions; the panel's controls hint lists only actions meaningful for the current option count, omitting navigation when exactly one option is shown; agent behavior and answer logging remain owned by their existing services.
 
-The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. Each model row owns the adapter-advertised reasoning-effort order and default: Shift+Tab cycles that row's efforts, includes provider-default behavior when the adapter advertises no default, and leaves models without selectable metadata unchanged. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model/reasoning-effort target per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
+The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. The selector carries a filter box above the list: typing narrows the rows to a case-insensitive substring over each row's `provider/model` label, model name, and description, keeping the selection on the previously highlighted row when it survives the filter; Escape clears a non-empty filter before a second Escape cancels the selector. Each model row owns the adapter-advertised reasoning-effort order and default: Shift+Tab cycles that row's efforts, includes provider-default behavior when the adapter advertises no default, and leaves models without selectable metadata unchanged. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model/reasoning-effort target per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
 
 ### Terminal ownership
 
@@ -47,6 +47,6 @@ The implemented [TUI terminal-state snapshot Agent Note](../testing/2026-07-18-t
 
 - Interactive terminal work has a stateful Markdown, card, plan, and question interface with no second terminal protocol to keep aligned.
 - The TUI carries a pi-tui dependency and a strict TTY requirement; non-TTY deployments use the Headless app or a structured protocol.
-- Session projection makes resume and compaction consistent with the durable conversation, but one configured session owns the transcript and editor.
+- Session projection makes resume consistent with the durable conversation, but one configured session owns the transcript and editor.
 - Tool packages extend terminal cards through their existing presentation methods without adding tool-specific branches to the TUI.
 - Model and reasoning-effort selection use adapter-advertised metadata without turning catalog membership into request validation; unused selections are not durable state.

.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md

@@ -14,17 +14,17 @@ Status: implemented
 
 DeepSeek Harness 将 [`@deepseek-ai/dsh-tui`](../../../../packages/ui/tui/README.md) 作为独立的 Cordis 插件交付。该插件只负责终端输入与呈现；agent 生命周期、会话持久化、工具执行以及模型可见的提问工具仍由不同组合项负责。插件要求 stdin 和 stdout 均为 TTY；条件不满足时会失败，不会静默切换为逐行输出。
 
-应用组合层只有一个终端入口。`@deepseek-ai/dsh-tui-demo` 在已配置 agent 之前挂载 TUI，`examples/tui-agent` 直接拥有交互式 coding 组装及其 Code Mode overlay。非交互任务使用 `@deepseek-ai/dsh-cli-demo`；ACP 仍是独立的自动化协议。
+只有一个终端入口。`@deepseek-ai/dsh-tui` 在已配置 agent 之前挂载，而 `apps/cli/config/tui.cordis.yml`——叠加在共享 `base.cordis.yml` 之上的 overlay——拥有交互式 coding 组装。非交互任务使用官方 headless 界面；ACP 仍是独立的自动化协议，并拥有受支持的 Code Mode demo。
 
 所选入口接收预创建 agent 使用的同一个新建或恢复 `SessionId`。入口先于 agent 组合挂载，等待相符的根 agent 出现，然后才进入全屏模式。因此，相符的 `agent-loop/config-start-failed` 事件会在接管屏幕前报告，并以状态码 1 退出。
 
 ### 会话投影与交互
 
-TUI 从活跃的 `session.surface` 重建 transcript（文本记录），并在事件携带 `surfaceOp` 时重新投影，因此恢复或压缩后的历史与模型可见会话保持一致。TUI 渲染 Markdown 文本与推理（其中围栏代码块隐藏 Markdown 标记、保留一个暗色的可选语言标签，并使用代码配色的正文）、token 用量、最新 `todo/write` 计划，以及各工具定义通过 `presentCall` 和 `presentResult` 方法生成的工具卡片。较长的工具卡片正文会保留可配置的头尾预览，并显示隐藏行数；一个终端控制可以展开或收起全部卡片。进行中的分片与工具调用会更新同一组组件，随后由完成事件收束状态。
+TUI 从追加来源的会话事件重建 transcript（文本记录），因此恢复后的历史会保留读者已经看到的每条消息；被压缩的范围不再与模型可见会话保持一致，而是留在一行标记之后仍可阅读（[追加来源的 transcript](../bug-fix/2026-07-29-human-transcript-append-origin.md)）。TUI 渲染 Markdown 文本与推理（其中围栏代码块隐藏 Markdown 标记、保留一个暗色的可选语言标签，并使用代码配色的正文）、token 用量、最新 `todo/write` 计划，以及各工具定义通过 `presentCall` 和 `presentResult` 方法生成的工具卡片。较长的工具卡片正文会保留可配置的头尾预览，并显示隐藏行数；一个终端控制可以展开或收起全部卡片。进行中的分片与工具调用会更新同一组组件，随后由完成事件收束状态。
 
 agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调用 `agent.steer()`。取消、推理显隐、工具卡片展开、重绘、清空 transcript 和退出都只是终端控制。`/exit` 和 `/quit` 共用同一条退出路径：先取消进行中的轮次，等待 agent 空闲，然后恢复并关闭终端。空闲态页脚根据 `tokenMeter` 得出上下文占用率，并显示所选模型和显式选定的推理强度；agent 运行期间，该摘要会替换为带已用时长的活动指示和 Escape 中断提示。`/status` 在这两种状态下均可用，并会追加一份仅在终端显示的详细快照，其中包括会话标识与时间戳、所选模型、推理强度（或默认状态）及推理显隐状态、从事件日志归并得出的生命周期计数、与页脚一致的去重用量分项和 KV 缓存命中率，以及 `tokenMeter` 给出的上下文用量和所选模型公布的容量。插件注册共享的 `userInteraction` 提供方，在左下角宽幅键盘操作面板中呈现排队的问题，面板显示批次进度、带编号的选项和对齐的描述；面板的操作提示只列出在当前选项数量下有意义的操作，仅有一个选项时不显示导航项；agent 行为和答案日志仍由既有服务负责。
 
-`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。每个模型行都持有适配器公布的推理强度顺序和默认值：按 Shift+Tab 可循环切换该行的推理强度；如果适配器没有公布默认值，循环中还会包含提供方默认行为；没有可选元数据的模型则保持不变。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个步骤快照一次同一个提供方/模型/推理强度目标，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
+`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。选择器在列表上方设有一个过滤框：输入内容会按对每行 `provider/model` 标签、模型名称和描述的大小写不敏感子串匹配来缩小行集，并在原先高亮行仍通过过滤时保持其选中状态；Escape 会先清空非空的过滤内容，再次按 Escape 才取消选择器。每个模型行都持有适配器公布的推理强度顺序和默认值：按 Shift+Tab 可循环切换该行的推理强度；如果适配器没有公布默认值，循环中还会包含提供方默认行为；没有可选元数据的模型则保持不变。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个步骤快照一次同一个提供方/模型/推理强度目标，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
 
 ### 终端所有权
 
@@ -47,6 +47,6 @@ agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调
 
 - 交互式终端拥有带状态的 Markdown、卡片、计划和提问界面，无需再对齐第二套终端协议。
 - TUI 会引入 pi-tui 依赖并严格要求 TTY；非 TTY 部署使用 Headless app 或结构化协议。
-- 会话投影使恢复和压缩与持久会话保持一致，但只有一个已配置会话拥有 transcript 和编辑器。
+- 会话投影使恢复与持久会话保持一致，但只有一个已配置会话拥有 transcript 和编辑器。
 - 工具包通过既有呈现方法扩展终端卡片，无需在 TUI 中增加工具专用分支。
 - 模型和推理强度选择使用适配器公布的元数据，但不会把目录成员关系变成请求校验；未使用的选择不属于持久化状态。

.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md
-2026-07-20-dsh-cli-personal-config.md: cc965438214b68078647596af5a28fd666e7bd95
-2026-07-20-dsh-cli-personal-config.zh.md: 02f9c578061e1c25044c377c3ec8f80594275ae7
+2026-07-20-dsh-cli-personal-config.md: 3331e36c86002d91fb272868268707fed014d01a
+2026-07-20-dsh-cli-personal-config.zh.md: 172d84b075ed7ecc127c317b47e30581db67f89a

.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md

@@ -14,10 +14,10 @@ Two coupled pieces, aligned with the `apps/` assembly tier proposed by the `dsh
 
 **The `dsh` CLI (`apps/cli`, npm name `@deepseek-ai/dsh`).** `apps/*` joins the workspaces as the product-assembly tier over `packages/*` libraries. The bin's dispatch reserves `web` and `-p`/`--prompt` for PR #443 (they exit with a pointer) so the two branches merge as a near-union; everything else runs the default surface: the interactive TUI, booting the shipped `examples/tui-agent/cordis.yml` (or an explicit config argument) with the invoking directory as the workspace. The committed `bin/dsh` launcher resolves the checkout through its own real path and runs the bin **from source** through Node's native TypeScript transform plus the app-owned tsconfig-paths loader, so `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` installs a command that always executes the current working tree. `pnpm run demo:tui` runs the same entry.
 
-**Personal config (`dsh-app-boot`).** The personal overlay lives in the Harness home — `$DSH_HOME`, else `~/.dsh` — resolved by the shared [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md) (`@deepseek-ai/dsh-paths`), the same single root skills and AGENTS.md resolve against. The dsh TUI surface consumes its two optional files; the demo bins boot their committed trees verbatim:
+**Personal config (`dsh-app-boot`).** The personal overlay lives in the Harness home — `$DSH_HOME`, else `~/.dsh` — resolved by the shared [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md) (`@deepseek-ai/dsh-paths`), the same single root skills and AGENTS.md resolve against. The official dsh surfaces consume its two optional files; the demo bins boot their committed trees verbatim:
 
 - `.env` — loaded after the invoking directory's `.env`; `process.loadEnvFile` never overrides, so precedence is ambient > project `.env` > personal `.env`.
-- `config.yaml` — a top-level YAML array of `@cordisjs/plugin-include` `PatchOptions`, parsed with the include's own `!!js` dialect (`loadPersonalPatches`) and passed to `boot()`, which forwards it as the root include's `patches`. Patch semantics are exactly the committed overlay semantics (the Code Mode overlay is the template): an id-targeted patch replaces the named entry's whole `config`, `insert` appends entries, an unmatched id warns and is skipped.
+- `config.yaml` — a top-level YAML array of `@cordisjs/plugin-include` `PatchOptions`, parsed with the include's own `!!js` dialect (`loadPersonalPatches`) and passed to `boot()`, which forwards it as the root include's `patches`. Patch semantics match the shipped surface overlays: an id-targeted patch replaces the named entry's whole `config`, `insert` appends entries, and an unmatched id is a silent no-op.
 - A missing file means no overlay; a present-but-unreadable, unparsable, or non-array file throws at boot (misconfiguration fails loud, never a silent skip).
 
 The PTY smoke's launcher isolates `$DSH_HOME` to a per-test directory, exactly as it already isolates `DSH_AGENTS_HOME`, so a developer's real personal overlay cannot leak into fixtures; only the dsh CLI reads personal config, so no other test launcher needed changes.

.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.zh.md

@@ -14,10 +14,10 @@ Status: implemented
 
 **`dsh` CLI（`apps/cli`，npm 名 `@deepseek-ai/dsh`）。** `apps/*` 作为 `packages/*` 库之上的产品装配层加入 workspaces。bin 的分发把 `web` 和 `-p`/`--prompt` 保留给 PR #443（它们以指引退出），使两个分支能以接近并集的方式合并；其余一切都运行默认表面：交互式 TUI，加载随仓库提供的 `examples/tui-agent/cordis.yml`（或显式的配置参数），并以调用目录为工作区。已提交的 `bin/dsh` 启动器通过自身真实路径解析 checkout，通过 Node 的原生 TypeScript 转换和应用自身持有的 tsconfig-paths loader **从源码**运行该 bin，因此 `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` 安装的命令永远执行当前工作树。`pnpm run demo:tui` 运行同一入口。
 
-**个人配置（`dsh-app-boot`）。** 个人 overlay 存放在 Harness home——`$DSH_HOME`，否则 `~/.dsh`——由共享的 [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md)（`@deepseek-ai/dsh-paths`）解析，与 skills、AGENTS.md 解析所依据的单一根目录相同。dsh 的 TUI 表面消费其中两个可选文件；各示例 bin 仍然逐字节按已提交的配置树启动：
+**个人配置（`dsh-app-boot`）。** 个人 overlay 存放在 Harness home——`$DSH_HOME`，否则 `~/.dsh`——由共享的 [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md)（`@deepseek-ai/dsh-paths`）解析，与 skills、AGENTS.md 解析所依据的单一根目录相同。dsh 的官方界面消费其中两个可选文件；各示例 bin 仍然逐字节按已提交的配置树启动：
 
 - `.env`——在调用目录的 `.env` 之后加载；`process.loadEnvFile` 从不覆盖已有值，因此优先级为环境变量 > 项目 `.env` > 个人 `.env`。
-- `config.yaml`——顶层 YAML 数组，元素为 `@cordisjs/plugin-include` 的 `PatchOptions`，用 include 自己的 `!!js` 方言解析（`loadPersonalPatches`）并传给 `boot()`，由它作为根 include 的 `patches` 转发。补丁语义与已提交 overlay 完全一致（Code Mode overlay 是模板）：按 id 定位的补丁替换该配置项的整个 `config`，`insert` 追加配置项，未匹配的 id 记录警告并跳过。
+- `config.yaml`——顶层 YAML 数组，元素为 `@cordisjs/plugin-include` 的 `PatchOptions`，用 include 自己的 `!!js` 方言解析（`loadPersonalPatches`）并传给 `boot()`，由它作为根 include 的 `patches` 转发。补丁语义与交付的 surface overlay 一致：按 id 定位的补丁替换该配置项的整个 `config`，`insert` 追加配置项，未匹配的 id 静默不执行任何操作。
 - 文件缺失即无 overlay；文件存在但不可读、不可解析或非数组则在启动时抛出（配置错误响亮失败，绝不静默跳过）。
 
 PTY 冒烟测试的启动器把 `$DSH_HOME` 隔离到每个测试自己的目录，与它已有的 `DSH_AGENTS_HOME` 隔离方式完全一致，开发者真实的个人 overlay 不可能泄漏进 fixture；只有 dsh CLI 读取个人配置，因此其他测试启动器无需改动。

.agents/notes/implemented/feature/2026-07-21-tui-resume-command.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-21-tui-resume-command.md: 86f62e16f5e2ee83e2ed36f0ed675ca2a1422c4b
-2026-07-21-tui-resume-command.zh.md: 06e58f81445aaaf5299282714148194c1d2aacf4
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-21-tui-resume-command.md
+2026-07-21-tui-resume-command.md: c8cb855378d793a168e1f87d6b41f9a48db7dc14
+2026-07-21-tui-resume-command.zh.md: 2a7e74d1106499cb7d7232dc13954ae126246550

.agents/notes/implemented/feature/2026-07-21-tui-resume-command.md

@@ -10,17 +10,17 @@ The original `/resume` printed shell commands. It did not let a keyboard user in
 
 ## Decision
 
-`/resume` uses the TUI's existing interactive overlay seam as a full-viewport picker rather than a centered dialog. The flat page keeps the search field, workspace, candidates, and shortcut footer in stable screen regions; only the active row uses the accent role. Its search editor starts immediately after the search glyph and emits pi-tui's cursor marker, so terminal IME composition remains anchored in the field. Escape clears a non-empty query before a second Escape closes the picker. It lists the current workspace by last logged activity and searches log-backed title or id. Each candidate displays current/live/persisted state, last turn outcome, recent provider/model, durable goal phase when present, and the id as secondary text. The current session and sessions already live in this runtime remain visible but disabled.
+`/resume` uses the TUI's existing interactive overlay seam as a full-viewport picker rather than a centered dialog. The flat page keeps the search field, workspace scope line, candidates, and shortcut footer in stable screen regions; only the active row uses the accent role. Its search editor starts immediately after the search glyph and emits pi-tui's cursor marker, so terminal IME composition remains anchored in the field. Escape clears a non-empty query before a second Escape closes the picker. It orders candidates by last logged activity and searches log-backed title or id. Each candidate displays current/live/persisted state, last turn outcome, recent provider/model, durable goal phase when present, and the id as secondary text. The current session and sessions already live in this runtime remain visible but disabled. The picker opens on the current workspace and reaches every other one through the scope toggle the [cross-workspace resume](2026-07-28-cross-workspace-resume.md) note owns.
 
-`session-query.readSession()` supplies a detached complete log validated by the same core replay boundary used by resume. The TUI folds title and goal state from that log. A candidate load failure is local to that row; selecting a candidate revalidates the log, `cwd`, route, current agent's idle status, and the exclusions for the current session and sessions already live in this runtime, so a stale listing cannot bypass preflight. A missing adapter reports an intact session with an unavailable route. This preflight does not lock the target or exclude another process.
+`session-query.readSession()` supplies a detached complete log validated by the same core replay boundary used by resume. The TUI folds title and goal state from that log. A candidate load failure is local to that row; selecting a candidate revalidates the log, workspace, route, current agent's idle status, and the exclusions for the current session and sessions already live in this runtime, so a stale listing cannot bypass preflight. A missing adapter reports an intact session with an unavailable route. This preflight does not lock the target or exclude another process.
 
-After preflight, the TUI flushes the current session, confirms that its agent remains idle, then stops the terminal before calling `TuiRuntime.handoffResume`. The shipped `dsh` host disposes the root app and uses `process.execve` with a normalized `--resume` argument, atomically replacing the process rather than starting a child. The resumed app publishes the same `SessionId`; ordinary replay restores transcript, title, todos, and durable goal state. Goal activation is intentionally disarmed, and the TUI asks for human confirmation or `/goal resume`.
+After preflight, the TUI flushes the current session, confirms that its agent remains idle, then stops the terminal before calling `TuiRuntime.handoffResume` with the validated id and the target workspace. The shipped `dsh` host disposes the root app and uses `process.execve` with a normalized `--resume` argument, atomically replacing the process rather than starting a child. The resumed app publishes the same `SessionId`; ordinary replay restores transcript, title, todos, and durable goal state. Goal activation is intentionally disarmed, and the TUI asks for human confirmation or `/goal resume`.
 
-`resumeCommand` remains an exit and no-host fallback. The TUI substitutes `{session}` only for display and never executes arbitrary shell text. The exit hint still appears only after the current session is durable.
+The exit line is a launcher-owned context slot rather than a config template, and a host without in-place handoff reports that the session stays resumable instead of naming a command it cannot construct; the [launcher-owned resume identity](../architecture/2026-07-28-launcher-owned-resume-identity.md) note owns that ownership move and supersedes the `resumeCommand` config key this note originally shipped. The TUI still never executes shell text.
 
 ## Alternatives considered
 
-**Have the TUI spawn `resumeCommand`.** Rejected: the template is deployment text, not trusted argv, and the TUI does not own app teardown or process lifetime. The constrained host seam receives only a validated `SessionId`.
+**Have the TUI spawn the resume command.** Rejected: the text is display copy, not trusted argv, and the TUI does not own app teardown or process lifetime. The constrained host seam receives only a validated `SessionId`.
 
 **Construct the resumed agent inside the existing TUI.** Rejected: replacing one config-created agent would cross Loader ownership, scoped plugin setup, persistence retirement, and terminal lifecycle in the presentation layer. Root disposal plus process replacement reuses the supported startup path.
 
@@ -36,4 +36,4 @@ After preflight, the TUI flushes the current session, confirms that its agent re
 
 ## Testing
 
-TUI tests cover keyboard navigation, title/id search, search-clear/cancel behavior, running-agent refusal, refusal of the current session and sessions already live in this runtime, route absence, corrupt rows, preflight revalidation, fallback commands, and stop-before-handoff ordering. Session-query tests pin detached full-log validation. Agent-loop resume tests pin exact identity and history; title, todo, and goal replay suites pin restored projections and disarmed goal activation. The keyless TUI snapshot owns the full-viewport selector and its IME cursor anchor, and a real PTY smoke covers search plus handoff.
+TUI tests cover keyboard navigation, title/id search, search-clear/cancel behavior, running-agent refusal, refusal of the current session and sessions already live in this runtime, route absence, corrupt rows, preflight revalidation, the no-host warning, and stop-before-handoff ordering. Session-query tests pin detached full-log validation. Agent-loop resume tests pin exact identity and history; title, todo, and goal replay suites pin restored projections and disarmed goal activation. The keyless TUI snapshot owns the full-viewport selector and its IME cursor anchor, and a real PTY smoke covers search plus handoff.

.agents/notes/implemented/feature/2026-07-21-tui-resume-command.zh.md

@@ -10,17 +10,17 @@ Status: implemented
 
 ## Decision
 
-`/resume` 使用 TUI 现有的交互式浮层接口，但以占满 viewport 的选择页呈现，而不是居中弹窗。这个扁平页面把搜索框、workspace、候选项和快捷键页脚放在稳定的屏幕区域，只有当前行使用强调色。搜索编辑器紧跟搜索图标起始，并输出 pi-tui 的光标标记，因此终端输入法的组合文本会锚定在输入框中。查询非空时，第一次按 Escape 会清空查询，第二次才关闭选择页。页面按日志记录的最后活动时间列出当前 workspace 的会话，并支持按日志内标题或 id 搜索。每个候选项都会显示是否为当前会话、是否活跃、是否已持久化，最近一个轮次的结果，最近使用的提供方／模型，以及可用时的持久化目标阶段；id 作为次要信息显示。当前会话和已在本运行时中处于活跃状态的会话仍会显示，但不可选择。
+`/resume` 使用 TUI 现有的交互式浮层接口，但以占满 viewport 的选择页呈现，而不是居中弹窗。这个扁平页面把搜索框、workspace 作用域行、候选项和快捷键页脚放在稳定的屏幕区域，只有当前行使用强调色。搜索编辑器紧跟搜索图标起始，并输出 pi-tui 的光标标记，因此终端输入法的组合文本会锚定在输入框中。查询非空时，第一次按 Escape 会清空查询，第二次才关闭选择页。页面按日志记录的最后活动时间排列候选项，并支持按日志内标题或 id 搜索。每个候选项都会显示是否为当前会话、是否活跃、是否已持久化，最近一个轮次的结果，最近使用的提供方／模型，以及可用时的持久化目标阶段；id 作为次要信息显示。当前会话和已在本运行时中处于活跃状态的会话仍会显示，但不可选择。选择页打开时位于当前 workspace，并通过[跨 workspace 恢复](2026-07-28-cross-workspace-resume.md)记录所拥有的作用域切换到达其他每一个 workspace。
 
-`session-query.readSession()` 提供一份脱离运行时的完整日志，并通过恢复流程所用的同一核心回放边界完成验证。TUI 从该日志中折叠出标题和目标状态。候选项加载失败时只影响该行；选择候选项后会复查日志、`cwd`、路由、当前 agent 的空闲状态，以及针对当前会话和已在本运行时中处于活跃状态的会话的排除规则，避免陈旧列表绕过预检。适配器缺失时会报告会话完整但路由不可用。该预检不会锁定目标，也不会排除其他进程。
+`session-query.readSession()` 提供一份脱离运行时的完整日志，并通过恢复流程所用的同一核心回放边界完成验证。TUI 从该日志中折叠出标题和目标状态。候选项加载失败时只影响该行；选择候选项后会复查日志、workspace、路由、当前 agent 的空闲状态，以及针对当前会话和已在本运行时中处于活跃状态的会话的排除规则，避免陈旧列表绕过预检。适配器缺失时会报告会话完整但路由不可用。该预检不会锁定目标，也不会排除其他进程。
 
-预检通过后，TUI 会刷写当前会话，再次确认其 agent 仍处于空闲状态，然后停止终端并调用 `TuiRuntime.handoffResume`。已交付的 `dsh` 宿主会释放根应用，并使用带有规范化 `--resume` 参数的 `process.execve` 原子替换当前进程，而不是启动子进程。恢复后的应用发布相同的 `SessionId`；常规回放会还原 transcript（文本记录）、标题、待办事项和持久化目标状态。系统会有意解除目标的激活状态，TUI 则要求用户确认继续或执行 `/goal resume`。
+预检通过后，TUI 会刷写当前会话，再次确认其 agent 仍处于空闲状态，然后停止终端并以经过验证的 id 和目标 workspace 调用 `TuiRuntime.handoffResume`。已交付的 `dsh` 宿主会释放根应用，并使用带有规范化 `--resume` 参数的 `process.execve` 原子替换当前进程，而不是启动子进程。恢复后的应用发布相同的 `SessionId`；常规回放会还原 transcript（文本记录）、标题、待办事项和持久化目标状态。系统会有意解除目标的激活状态，TUI 则要求用户确认继续或执行 `/goal resume`。
 
-`resumeCommand` 保留为退出及无宿主时的回退方案。TUI 仅为显示目的替换 `{session}`，绝不执行任意 shell 文本。只有当前会话已经持久化时，退出提示才会出现。
+退出时打印的行是启动器拥有的上下文插槽，而非配置模板；不支持原地交接的宿主会说明会话仍可恢复，而不再给出它无法构造的命令。[由启动器持有的会话身份与退出行](../architecture/2026-07-28-launcher-owned-resume-identity.md)记录了这次所有权迁移，并取代本记录最初交付的 `resumeCommand` 配置键。TUI 仍然绝不执行 shell 文本。
 
 ## Alternatives considered
 
-**让 TUI 创建 `resumeCommand` 进程。** 否决：该模板是部署文本，不是可信的参数列表，且 TUI 不拥有应用拆卸或进程生命周期。受约束的宿主接口只接收经过验证的 `SessionId`。
+**让 TUI 创建恢复命令进程。** 否决：该文本是展示用文案，不是可信的参数列表，且 TUI 不拥有应用拆卸或进程生命周期。受约束的宿主接口只接收经过验证的 `SessionId`。
 
 **在现有 TUI 内构造恢复后的 agent。** 否决：在表现层替换由配置创建的 agent，会跨越 Loader 所有权、作用域插件初始化、持久化资源释放和终端生命周期。释放根应用并替换进程可以复用受支持的启动路径。
 
@@ -36,4 +36,4 @@ Status: implemented
 
 ## Testing
 
-TUI 测试覆盖键盘导航、标题／id 搜索、清空搜索后再取消、agent 运行期间拒绝恢复、拒绝恢复当前会话和已在本运行时中处于活跃状态的会话、路由缺失、损坏的候选行、预检复查、回退命令，以及停止终端先于宿主交接的顺序。session-query 测试固定脱离运行时的完整日志验证。agent-loop 恢复测试固定会话身份和历史完全一致；标题、待办事项和目标回放测试套件固定这些投影均可恢复，且目标激活状态已经解除。无密钥 TUI 快照固定全屏选择页和输入法光标锚点，真实 PTY smoke 则覆盖搜索与交接。
+TUI 测试覆盖键盘导航、标题／id 搜索、清空搜索后再取消、agent 运行期间拒绝恢复、拒绝恢复当前会话和已在本运行时中处于活跃状态的会话、路由缺失、损坏的候选行、预检复查、无宿主时的告警，以及停止终端先于宿主交接的顺序。session-query 测试固定脱离运行时的完整日志验证。agent-loop 恢复测试固定会话身份和历史完全一致；标题、待办事项和目标回放测试套件固定这些投影均可恢复，且目标激活状态已经解除。无密钥 TUI 快照固定全屏选择页和输入法光标锚点，真实 PTY smoke 则覆盖搜索与交接。

.agents/notes/implemented/feature/2026-07-24-configurable-tui-prompt-theme.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-24-configurable-tui-prompt-theme.md: 4008f23a3f545e9b4484f0fa3f8490ad9b2c5541
-2026-07-24-configurable-tui-prompt-theme.zh.md: daf15b54d7e04d3860eacad47b754b963cde36fa
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-24-configurable-tui-prompt-theme.md
+2026-07-24-configurable-tui-prompt-theme.md: f8815c6c1904c47ebb899c3da7ac62a50f7f88f0
+2026-07-24-configurable-tui-prompt-theme.zh.md: 831471860a9dcd8bb10408c492e1fc6299af6262

.agents/notes/implemented/feature/2026-07-24-configurable-tui-prompt-theme.md

@@ -16,7 +16,7 @@ The TUI theme groups `color`, `truecolor`, `leftPrompt`, `rightPrompt`, `inputPr
 
 Registered fragments are trusted ANSI-capable presentation output. Template literals and ordinary external content remain sanitized, but a prompt-value plugin may emit terminal controls. Composite values own coordinated background transitions and separators, so one `${powerline}` value can render a complete Powerline segment without coupling adjacent atomic providers.
 
-The built-in `cwd`, `git/worktree`, `token_meter/cache_hit_rate`, `model`, `context`, `timing`, styled `symbol` label, and `indicator` caret values use the same registry. Session and agent events update their handles, while the running timer updates `timing` and the animated `indicator` each tick. The shipped input template is `${symbol} ${indicator}`, preserving the existing `dsh > ` prefix.
+The built-in `cwd`, `git/worktree`, `token_meter/cache_hit_rate`, `model`, `context`, `queued`, styled `symbol` label, and `indicator` caret values use the same registry. Session and agent events update their handles, while the running timer updates `queued` — the steering-queue badge, unavailable unless a running turn has queued messages — and the animated `indicator` each tick. The shipped input template is `${symbol} ${indicator}`, preserving the existing `dsh > ` prefix.
 
 ## Alternatives considered
 

.agents/notes/implemented/feature/2026-07-24-configurable-tui-prompt-theme.zh.md

@@ -16,7 +16,7 @@ TUI 主题把 `color`、`truecolor`、`leftPrompt`、`rightPrompt`、`inputPromp
 
 注册的片段被视为可信的、允许携带 ANSI 的呈现输出。模板中的字面文本与普通外部内容仍会被清洗，但提供提示符值的插件可以输出终端控制序列。复合值自行负责协调背景色过渡与分隔符，因此一个 `${powerline}` 值就能渲染完整的 Powerline 段，而无需与相邻的原子提供方耦合。
 
-内置的 `cwd`、`git/worktree`、`token_meter/cache_hit_rate`、`model`、`context`、`timing`、带样式的 `symbol` 标签与 `indicator` 光标符值使用同一个注册表。会话与 agent（智能体）事件更新各自的句柄，运行计时器每一拍更新 `timing` 与带动画的 `indicator`。随附的输入模板为 `${symbol} ${indicator}`，保留了原有的 `dsh > ` 前缀。
+内置的 `cwd`、`git/worktree`、`token_meter/cache_hit_rate`、`model`、`context`、`queued`、带样式的 `symbol` 标签与 `indicator` 光标符值使用同一个注册表。会话与 agent（智能体）事件更新各自的句柄，运行计时器每一拍更新 `queued`——转向队列徽标，仅在运行中的一轮有排队消息时才可用——与带动画的 `indicator`。随附的输入模板为 `${symbol} ${indicator}`，保留了原有的 `dsh > ` 前缀。
 
 ## 曾考虑的替代方案
 

.agents/notes/implemented/feature/2026-07-24-web-session-model-selector.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-24-web-session-model-selector.md
-2026-07-24-web-session-model-selector.md: 4bd52bba8d3ba16c01fc59d2b561516f6fd0bb87
-2026-07-24-web-session-model-selector.zh.md: 35f3525b80e7d27fb573e854aefc317ed354df53
+2026-07-24-web-session-model-selector.md: 003c0b5ac1c6963701e1d93e4c3ff8615fc45dd5
+2026-07-24-web-session-model-selector.zh.md: ba76b87b9bd43fff97bf9ea76624f5f75b10e135

.agents/notes/implemented/feature/2026-07-24-web-session-model-selector.md

@@ -18,7 +18,7 @@ The browser `ModelService` owns one `ModelDirectory` per live session. Its snaps
 
 `@deepseek-ai/dsh-client-ui-conversation` declares the session-scoped single slot `conversation.input.model` as a child of its composer-bar entry. InputBar renders the seat in its trailing controls immediately before the pending indicator and primary button; the seat receives the bar's `locked` owner prop and session scope. `@deepseek-ai/dsh-client-ui-model` occupies that seat and also contributes `/model` over the same directory. Its compact trigger displays the catalog model name and effective reasoning label, falling back to ids when metadata is absent. The upward menu first offers Model and, when the current exact model supports it, Effort; Model drills into provider groups, while Effort drills into the adapter-ordered levels. The provider-default row appears only when the adapter does not configure a model default.
 
-The production browser roster is the flat config tree in `apps/cli/cordis.yml`; the model feature is one `dshClient` row rather than a package hardcoded in Web boot code. Its package manifest orders it after the runtime and command feature, while Cordis service injection waits for the conversation slot before registering the composer occupant.
+The production browser roster is assembled from `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml`; the model feature is one `dshClient` row rather than a package hardcoded in Web boot code. Its package manifest orders it after the runtime and command feature, while Cordis service injection waits for the conversation slot before registering the composer occupant.
 
 ## Alternatives considered
 

.agents/notes/implemented/feature/2026-07-24-web-session-model-selector.zh.md

@@ -18,7 +18,7 @@ Web Host 为每个新建或恢复的 agent（智能体）复用 `installAgentLlm
 
 `@deepseek-ai/dsh-client-ui-conversation` 将会话作用域的单实例 slot `conversation.input.model` 声明为其输入栏 entry 的子 slot。InputBar 在尾部控件区将该 seat 渲染于 pending 指示器与主按钮之前；该 seat 接收输入栏的 `locked` owner prop 与会话作用域。`@deepseek-ai/dsh-client-ui-model` 占用该 seat，并在同一目录上提供 `/model`。其紧凑型触发器显示目录中的模型名称与生效的推理强度标签；元数据缺失时则回退到相应 ID。向上展开的菜单首先提供 Model，并在当前精确模型支持时提供 Effort；Model 可深入提供方分组，Effort 可深入适配器排序的级别。仅当适配器没有配置模型默认值时，才显示提供方默认值行。
 
-生产环境的浏览器名册是 `apps/cli/cordis.yml` 中的平铺 config tree；模型功能对应其中一行 `dshClient` 配置项，而不是 Web boot 代码中硬编码的包。其包 manifest（元数据清单）将加载顺序置于运行时与命令功能之后；Cordis 服务注入则等待 conversation slot 可用，再注册 composer 占用方。
+生产环境的浏览器名册由 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 共同组装；模型功能对应其中一行 `dshClient` 配置项，而不是 Web boot 代码中硬编码的包。其包 manifest（元数据清单）将加载顺序置于运行时与命令功能之后；Cordis 服务注入则等待 conversation slot 可用，再注册 composer 占用方。
 
 ## 考虑过的替代方案
 

.agents/notes/implemented/feature/2026-07-27-tui-tool-card-header.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-tui-tool-card-header.md
-2026-07-27-tui-tool-card-header.md: 0868d8dcaf5ab1641a122ecda05578aef46f5f94
-2026-07-27-tui-tool-card-header.zh.md: 71db5fc6fc853039ea9be1cb1c51686f941c4cb7
+2026-07-27-tui-tool-card-header.md: 13f5e7fec1a82d02d5bf8cae4784505c02ab6f38
+2026-07-27-tui-tool-card-header.zh.md: 85f9f1244a7da568f5dcc892729021ac0fa270c9

.agents/notes/implemented/feature/2026-07-27-tui-tool-card-header.md

@@ -24,7 +24,7 @@ The redesign is TUI-only. It touches `ToolCardComponent` in `packages/ui/tui/src
 
 **Keep the presenter title in the header** (e.g. `Tool / read / Read src/index.ts`). Rejected: the verb duplicates the tool name, and non-bash tools have no genuinely distinct one-line description — the target belongs in the body, so only bash contributes a header desc.
 
-**A summary footer for every card type** (line counts, exit pills, diff counts as a uniform `└ …` line). Deferred: only the diff footer shipped. Terminal exit keeps its existing dim `[exit N]` line, long output keeps its existing head+tail middle-elision, an empty result stays header-only, and an error body stays plain (only the header color carries the error) — the current treatments were kept deliberately, not by omission.
+**A summary footer for every card type** (line counts, exit pills, diff counts as a uniform `└ …` line). Deferred: only the diff footer shipped. Terminal exit keeps its existing dim `[exit N]` line, long output keeps its existing head+tail middle-elision, an empty result stays header-only, and an error body stays plain (only the header color carries the error) — the current treatments were kept deliberately, not by omission. The body's flat default-foreground styling was later revisited: the [consolidated TUI presentation](../architecture/2026-07-28-consolidated-tui-presentation.md) recesses the whole body into one dim tone under this note's colored status header.
 
 ## Consequences
 

.agents/notes/implemented/feature/2026-07-27-tui-tool-card-header.zh.md

@@ -24,7 +24,7 @@ TUI 曾把每次工具调用渲染为 `{glyph} {title}`，其中 `title` 是 pre
 
 **把 presenter 标题保留在表头**（例如 `Tool / read / Read src/index.ts`）。已否决：动词与工具名重复，而非 bash 工具并没有真正独立的单行描述——操作对象属于正文，因此只有 bash 向表头贡献描述段。
 
-**为每一种卡片都加一条汇总页脚**（行数、退出码徽章、diff 计数统一为一条 `└ …` 行）。已推迟：仅 diff 页脚落地。终端退出保留其既有的变暗 `[exit N]` 行，长输出保留其既有的首尾中段省略，空结果保持仅表头，错误正文保持朴素（仅表头颜色承载错误）——这些既有处理是有意保留的，而非遗漏。
+**为每一种卡片都加一条汇总页脚**（行数、退出码徽章、diff 计数统一为一条 `└ …` 行）。已推迟：仅 diff 页脚落地。终端退出保留其既有的变暗 `[exit N]` 行，长输出保留其既有的首尾中段省略，空结果保持仅表头，错误正文保持朴素（仅表头颜色承载错误）——这些既有处理是有意保留的，而非遗漏。正文原本以默认前景色平铺，这种样式后来也经过调整：[整合后的 TUI 呈现](../architecture/2026-07-28-consolidated-tui-presentation.md)把整个正文收进本文所述彩色状态标题之下的同一种暗色调。
 
 ## Consequences
 

.agents/notes/implemented/feature/2026-07-28-cross-workspace-resume.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-28-cross-workspace-resume.md
+2026-07-28-cross-workspace-resume.md: 1ef6195b6f269a6dc8cf869902b7772ad48f00c2
+2026-07-28-cross-workspace-resume.zh.md: d31dc5f0470e20f86062edb4ad75c37a6160fe23

.agents/notes/implemented/feature/2026-07-28-cross-workspace-resume.md

@@ -0,0 +1,49 @@
+# Agent Note: Cross-workspace session resume
+
+Status: implemented
+
+English | [中文](2026-07-28-cross-workspace-resume.zh.md)
+
+## Problem
+
+`/resume` could only reach sessions started in the launch directory, so returning to yesterday's work in another project meant remembering its path, leaving the TUI, and relaunching there. Two independent causes produced that limit, and fixing either alone changes nothing.
+
+Storage was the binding one. The shipped TUI composition defaulted its persistence root to a relative `./.sessions`, so each launch directory owned a disjoint JSONL root and a disjoint derived `session-query.db`. Sessions from another project were not filtered out of the listing — they were absent from the store the listing reads. The JSONL backend already partitions per-cwd *inside* one root, so the partitioning was doubled: once by root, once within it.
+
+The picker then filtered again. It dropped records whose `cwd` differed from the current session before display, and `summarizeResumeCandidate` independently marked a differing `cwd` as `disabledReason: 'different workspace'`, so a foreign session that did reach the store was both hidden and refused.
+
+Finally, resume never changed directory. The host re-execs `dsh --resume=<id>` through `process.execve`, which inherits the cwd. Session *header* cwd is restored from the log, but process cwd is what `dsh-fs-local`, the bash executor, and glob/grep resolve against, so resuming a foreign session would have replayed its transcript while acting on the wrong project.
+
+## Decision
+
+The shared CLI configuration supplies one session root under the Harness home, the picker gains a workspace scope, and the handoff carries the target directory.
+
+**Storage.** The shared base owns the default in `apps/cli/config/base.cordis.yml`: its `session-persistence-jsonl` row resolves `sessions` beneath `DSH_HOME`, with the standard `~/.dsh` fallback when the environment has not been initialized. TUI, Web, and headless therefore consume one default without launcher patches or a boot slot. An overlay or personal patch that states an explicit root replaces that row's whole `config` and remains the deployment's authoritative choice.
+
+**Scope, not exclusion.** A workspace other than the current one is a display scope rather than a disabled reason. `showResume()` summarizes every record and the `ResumePicker` owns a `scope` of `'workspace' | 'all'`, defaulting to the current workspace so the common case is unchanged. Tab toggles; the scope line names the active scope and the count the other holds; each row in the all-workspaces scope reports its own workspace, and that label joins the searchable text only in the scope that shows it. A toggle clears the query and selection so the highlighted row always belongs to the visible list, and the per-row workspace line makes a row one terminal row taller in that scope, which the visible-count budget accounts for.
+
+`summarizeResumeCandidate` therefore drops `'different workspace'` and gains `'session has no recorded workspace'`. That is a real new refusal rather than a rename: a header without `cwd` names no directory for the host to enter, so it cannot be handed off even though its log is intact.
+
+**Handoff.** `TuiResumeHost.handoff` takes the target `cwd` beside the `SessionId`. `preflightResume` resolves both together and returns them, so the caller cannot re-derive a stale directory from the row it displayed — a record whose `cwd` moved between listing and preflight is resumed in the *re-read* directory, which is why the former "reject a moved cwd" behavior is now a handoff with the new path. The shipped host chdirs before disposing the app: an unreachable directory must reject while the caller can still restore the terminal, because after teardown no owner remains to report to. Resume always uses the default `dsh --resume` surface because `meta` rejects parent options; the handoff already enters the persisted target directory.
+
+## Alternatives considered
+
+**Patch `persistenceRoot` from the `dsh` launcher instead of changing the bundle default.** Rejected after finding that a loader patch assigns `config` wholesale. The personal `~/.dsh/config.yaml` overlay already patches the `tui-agent` row with a partial config, which is exactly why `persistenceRoot` was falling back to the bundle default in the first place; a launcher patch would either be erased by that overlay or have to win over it and make the overlay unable to set the field. Owning the default in the bundle survives any partial patch and keeps one home for the fact.
+
+**Keep `./.sessions` and additionally scan the Harness-home root.** Rejected: two roots means two SQLite indexes and a merged listing whose rows have different liveness and revision authorities, to preserve visibility of logs that the no-migration decision already gives up.
+
+**Migrate existing project-local logs into the shared root.** Rejected by the requester. Sessions under a project's `./.sessions` stay on disk and stay resumable by explicit `dsh --resume <id>` from that directory, but no longer appear in `/resume`.
+
+**One flat list of every workspace.** Rejected: it loses the "this project" default that the overwhelmingly common case wants, and in a busy home directory the current project's sessions would compete with unrelated ones.
+
+**Let the host infer the directory from the restored session header.** Rejected: the header is model- and prompt-facing state restored *after* boot, while the directory must be entered *before* `execve`. Passing it explicitly keeps the ordering visible at the seam.
+
+## Consequences
+
+- Sessions already stored under a project-local `./.sessions` disappear from `/resume`. This is the accepted cost of no migration.
+- A resumed session can change the process's working directory, so a foreign resume is not a pure transcript restoration — every path-resolving tool moves with it.
+- The Harness home now holds session logs for every project on the machine. Its growth is no longer bounded by one checkout, and no retention policy is introduced here.
+
+## Testing
+
+TUI tests cover the default scope hiding other workspaces while reporting their count, Tab revealing them with per-row workspace labels, Tab back clearing the query and selection, searching by workspace label, a cwd-less record staying visible but disabled, and the handoff receiving both the id and the workspace re-read at preflight. The former "reject a moved cwd" case now asserts the handoff carries the new directory. Built CLI PTY tests exercise the shared config default and the per-process derived query index. The keyless TUI snapshot pins both scopes of the selector, including the scope line, the per-row workspace lines, and the Tab hint in the footer. A manual cross-workspace resume verified at the process level that the replacement's working directory became the target workspace.

.agents/notes/implemented/feature/2026-07-28-cross-workspace-resume.zh.md

@@ -0,0 +1,49 @@
+# Agent Note: 跨 workspace 会话恢复
+
+Status: implemented
+
+[English](2026-07-28-cross-workspace-resume.md) | 中文
+
+## Problem
+
+`/resume` 只能触达在启动目录中创建的会话，因此要回到昨天在另一个项目里的工作，就得记住它的路径、退出 TUI、再到那里重新启动。造成这一限制的原因有两个，彼此独立，只修其中一个都不会有任何变化。
+
+存储是那个决定性的原因。已交付的 TUI 组合把持久化根默认成相对路径 `./.sessions`，于是每个启动目录都独占一份互不相交的 JSONL 根目录，以及一份互不相交的派生 `session-query.db`。来自另一个项目的会话并不是在列表中被过滤掉的——它们根本不存在于列表读取的存储中。JSONL 后端本来就会在*同一个*根目录*内部*按 cwd 分区，所以分区被叠加了两层：一层按根目录，一层在根目录内部。
+
+接着选择器又过滤了一次。它在展示前丢弃 `cwd` 与当前会话不同的记录，而 `summarizeResumeCandidate` 又独立地把不同的 `cwd` 标记为 `disabledReason: 'different workspace'`，于是一个确实进入了存储的外部会话既被隐藏，也会被拒绝。
+
+最后，恢复流程从不切换目录。宿主通过 `process.execve` 重新执行 `dsh --resume=<id>`，而它会继承 cwd。会话*头部*的 cwd 会从日志中还原，但 `dsh-fs-local`、bash 执行器以及 glob／grep 解析路径时依据的是进程 cwd，所以恢复一个外部会话会在回放它的 transcript（文本记录）的同时，作用到错误的项目上。
+
+## Decision
+
+共享 CLI 配置提供 Harness home 下的同一个会话根目录，选择器获得 workspace 范围，交接过程携带目标目录。
+
+**存储。** 共享 base 在 `apps/cli/config/base.cordis.yml` 中拥有默认值：其 `session-persistence-jsonl` 配置项把 `sessions` 解析到 `DSH_HOME` 下；环境尚未初始化时则使用标准的 `~/.dsh` 回退值。因此 TUI、Web 与 headless 使用同一个默认值，无需启动器补丁或启动槽位。若 overlay 或个人 patch 显式声明根目录，它会整体替换该配置项的 `config`，并继续作为部署的权威选择。
+
+**是范围，不是排除。** 当前 workspace 之外的 workspace 是一种展示范围，而不是禁用理由。`showResume()` 汇总每一条记录，`ResumePicker` 持有一个 `'workspace' | 'all'` 的 `scope`，默认为当前 workspace，因此常见场景毫无变化。Tab 切换范围；范围行会说明当前生效的范围，以及另一个范围下的数量；在全 workspace 范围中每一行都报告自己的 workspace，而该标签只在展示它的范围里才加入可搜索文本。切换范围会清空查询和选中项，使高亮行始终属于可见列表；而逐行的 workspace 行会让该范围下的每一行在终端里多占一行，可见条数预算已经把这一点计入。
+
+因此 `summarizeResumeCandidate` 去掉了 `'different workspace'`，并新增 `'session has no recorded workspace'`。这是一条真正新增的拒绝理由，而不是改名：没有 `cwd` 的头部没有指明任何目录供宿主进入，所以即便它的日志完好也无法完成交接。
+
+**交接。** `TuiResumeHost.handoff` 在 `SessionId` 之外还接收目标 `cwd`。`preflightResume` 把两者一起解析并一起返回，因此调用方无法从它展示过的那一行里重新推导出一个陈旧目录——在列表展示与预检之间 `cwd` 发生了移动的记录，会在*重新读取到的*目录中恢复，这也是原先「拒绝已移动的 cwd」的行为如今变成携带新路径完成交接的原因。已交付的宿主在释放应用之前切换目录：不可达的目录必须在调用方还能恢复终端时就拒绝，因为拆卸之后已经没有任何所有者可供汇报。恢复始终使用默认的 `dsh --resume` 界面，因为 `meta` 会拒绝父级选项；交接过程已经进入持久化保存的目标目录。
+
+## Alternatives considered
+
+**从 `dsh` 启动器给 `persistenceRoot` 打补丁，而不是改动组合包默认值。** 在发现 loader 补丁会整体赋值 `config` 之后否决。个人的 `~/.dsh/config.yaml` 覆盖层已经用一份局部配置给 `tui-agent` 那一项打了补丁，这恰恰就是 `persistenceRoot` 一开始会退回到组合包默认值的原因；启动器补丁要么会被该覆盖层擦除，要么必须压过它，从而让覆盖层再也无法设置这个字段。把默认值放在组合包里能经受任何局部补丁，并让这项事实只有一个归属。
+
+**保留 `./.sessions`，并额外扫描 Harness home 根目录。** 否决：两个根目录意味着两份 SQLite 索引，以及一份合并列表——其中各行的活跃状态与版本权威来源并不相同，而这一切只是为了保住不做迁移的决策本就已经放弃的那部分日志可见性。
+
+**把现有的项目本地日志迁移到共享根目录。** 被需求方否决。项目 `./.sessions` 下的会话仍留在磁盘上，从该目录显式执行 `dsh --resume <id>` 仍可恢复，只是不再出现在 `/resume` 中。
+
+**把所有 workspace 铺成一个扁平列表。** 否决：这会丢掉绝大多数场景想要的「本项目」默认值，而在一个繁忙的 home 目录里，当前项目的会话会和无关会话争夺注意力。
+
+**让宿主从还原后的会话头部推断目录。** 否决：会话头部是面向模型与提示词的状态，在启动*之后*才还原，而目录必须在 `execve` *之前*进入。显式传递它能让这个顺序在边界处保持可见。
+
+## Consequences
+
+- 已经存放在项目本地 `./.sessions` 下的会话会从 `/resume` 中消失。这是不做迁移所接受的代价。
+- 恢复一个会话可以改变进程的工作目录，因此恢复外部会话不是单纯的 transcript 还原——每个解析路径的工具都会随之移动。
+- Harness home 现在保存着这台机器上每个项目的会话日志。它的增长不再受单个 checkout 约束，而本记录也没有引入任何保留策略。
+
+## Testing
+
+TUI 测试覆盖默认范围隐藏其他 workspace 但报告其数量、Tab 显示它们并带上逐行 workspace 标签、再按 Tab 返回时清空查询与选中项、按 workspace 标签搜索、无 cwd 的记录仍可见但不可选，以及交接同时收到 id 和在预检时重新读取到的 workspace。原先「拒绝已移动的 cwd」的用例现在断言交接携带新目录。构建后的 CLI PTY 测试会运行共享配置默认值与每进程派生的查询索引。无密钥 TUI 快照固定选择器的两个范围，包括范围行、逐行 workspace 行，以及页脚中的 Tab 提示。手动执行的一次跨 workspace 恢复在进程层面验证了替换后进程的工作目录变为目标 workspace。

.agents/notes/implemented/feature/2026-07-28-dsh-guided-skill-session-commands.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-28-dsh-guided-skill-session-commands.md
+2026-07-28-dsh-guided-skill-session-commands.md: a11807dca9ad1640857cd95a8b528df691b8a27e
+2026-07-28-dsh-guided-skill-session-commands.zh.md: e48420ed77496d8c336031473364f9008fccced7

.agents/notes/implemented/feature/2026-07-28-dsh-guided-skill-session-commands.md

@@ -0,0 +1,43 @@
+# Agent Note: `dsh migrate`/`dsh upgrade` seed the first turn with a skill
+
+Status: implemented
+
+English | [中文](2026-07-28-dsh-guided-skill-session-commands.zh.md)
+
+## Problem
+
+Two recurring flows begin with the user manually invoking one skill and answering its questions: migrating from another coding agent, and upgrading this checkout. Both require the user to know the skill exists and to type `/skill:dsh-migrate` or `/skill:dsh-upgrade` as the session's first turn. A dedicated entry command that drops the user straight into that guided session removes the discovery step.
+
+## Decision
+
+`dsh migrate` and `dsh upgrade` boot the ordinary TUI as a fresh session whose first turn auto-invokes a bundled skill (`dsh-migrate`, `dsh-upgrade`), exactly as if the user typed `/skill:<name>` and pressed Enter.
+
+The seed reuses the existing TUI skill path, not a new one. `createTuiChat` already has `invokeSkill(name, instructions)` — the code a typed `/skill:<name>` runs, including the "Unknown skill" notice. The launcher passes the skill name to the TUI through a new boot-context slot `INITIAL_SKILL_KEY` (`tuiInitialSkill`), mirroring `CONFIGURED_AGENT_IDENTITIES_KEY`/`TUI_GOODBYE_MESSAGE_KEY`: `ctx.provide` is the only channel from launcher argv into a Loader-mounted plugin. The TUI's `apply()` reads the slot and folds it into `config.initialSkill`; after `ui.start()` succeeds, `createTuiChat` fires `invokeSkill(config.initialSkill, '')` once when set.
+
+**Freshness is gated in the launcher, not the TUI.** `runSkillSession` always mints a fresh session and provides the slot only when `resumeSessionId === undefined`, so a later `dsh --resume <id>` of that session is an ordinary TUI session with no re-injection. The TUI stays generic: it invokes whatever skill it is handed, once, at startup.
+
+**`migrate`/`upgrade` take no options.** Unlike `meta`, they carry no `--resume`, `--config`, or `-p`; a guided fresh-session entry has nothing to resume or reconfigure. Any leaked default-surface option fails loud, matching the `web`/`meta` rejection pattern in the Commander adapter. The two modes share one `SkillSessionInvocation` discriminant (`mode: 'migrate' | 'upgrade'`); `bin.ts` maps the mode to `dsh-${mode}`.
+
+The `dsh-migrate` skill is bundled under `skills/` (shipped through `DSH_BUNDLED_SKILL_DIR`, like `dsh-upgrade`). It asks which source agent (opencode/pi/Claude Code/Codex) if unstated, then maps each capability — workspace instructions, personal overlay, skills, hooks, MCP, API/env — to its DSH equivalent, grounded in the actual repo surfaces (the `hooks-claude`/`hooks-codex` bridges, `~/.dsh/{config.yaml,.env,AGENTS.md,skills/}`, `AGENTS.md`/`CLAUDE.md`, `mcporter`), and states plainly when a capability has no equivalent.
+
+## Testing
+
+`apps/cli/tests/args.spec.ts` gains routing for `migrate`/`upgrade` (bare discriminant) and exit-1 for every leaked option on either side of each subcommand.
+
+`packages/ui/tui/tests/tui.spec.ts` gains two fake-terminal cases in the existing skill describe block: `config.initialSkill` set delivers the rendered skill body as the first turn with no user input, and an unknown initial skill reports a notice without sending. `runSkillSession` itself is composition inside the module's `v8 ignore` block, like `runTui`/`runMeta`.
+
+No keyless PTY snapshot: per the maintainer's scope call for this change, unit coverage plus interactive verification suffices, and the seed rides the already-snapshotted `/skill:` render path. Both commands were verified interactively in tmux from a scratch cwd: `dsh migrate` loaded `dsh-migrate` and asked which source agent; `dsh upgrade` loaded `dsh-upgrade`, which pulled in `dsh-customize` and began checkout discovery.
+
+## Alternatives considered
+
+**Prefill the input and let the user press Enter.** Rejected: needs a new editor-prefill seam and still requires a keystroke. Auto-submit reuses `invokeSkill` and delivers the intended one-command entry.
+
+**Seed a natural-language instruction ("use the dsh-migrate skill…") instead of `/skill:<name>`.** Rejected here: the literal skill-invocation path renders the skill body into the first turn deterministically, identical to the manual command, rather than depending on the model choosing to load the skill.
+
+**Support `--resume` on `migrate`/`upgrade`.** Rejected: these are one-shot guided entries. A resumed session is an ordinary TUI session reachable through the default surface's `dsh --resume <id>`; re-injecting the skill on resume would duplicate the first turn.
+
+**Read `INITIAL_SKILL_KEY` outside the TUI (as `CONFIGURED_AGENT_IDENTITIES_KEY` is read by `agent-loop`) rather than in the TUI's `apply()`.** Not needed: `initialSkill` is a TUI `Config` field consumed in `createTuiChat`, so folding the slot into config at the TUI entry keeps it beside the other launcher-owned runtime reads (`tuiResumeHost`, `tuiGoodbyeMessage`) and touches no other plugin.
+
+## Consequences
+
+Migrating or upgrading is one command from anywhere, with the guiding skill already invoked. The launcher→TUI initial-skill slot is reusable by any future guided-session command; the TUI's contract is "invoke this named skill once at startup," and freshness/resume policy stays with the launcher that owns session identity. The [TUI skill slash command](2026-07-21-tui-skill-slash-command.md) remains the mechanism; this note adds a launcher-driven auto-invocation of it and does not supersede it.

.agents/notes/implemented/feature/2026-07-28-dsh-guided-skill-session-commands.zh.md

@@ -0,0 +1,43 @@
+# Agent Note：`dsh migrate`/`dsh upgrade` 以 skill 播种首轮
+
+Status: implemented
+
+[English](2026-07-28-dsh-guided-skill-session-commands.md) | 中文
+
+## 问题
+
+有两个反复出现的流程都以用户手动调用某个 skill 并回答其问题开始：从其他编码 agent 迁移，以及升级本 checkout。二者都要求用户知道该 skill 存在，并把 `/skill:dsh-migrate` 或 `/skill:dsh-upgrade` 作为会话首轮键入。一个专用入口命令若能让用户直接进入该引导式会话，便可省去这一发现步骤。
+
+## 决策
+
+`dsh migrate` 与 `dsh upgrade` 以全新会话启动普通 TUI，其首轮自动调用一个内置 skill（`dsh-migrate`、`dsh-upgrade`），效果等同于用户键入 `/skill:<name>` 并回车。
+
+播种复用现有的 TUI skill 路径，而非新增一条。`createTuiChat` 已有 `invokeSkill(name, instructions)`——即键入 `/skill:<name>` 所走的代码，包含“未知 skill”通知。启动器通过一个新的启动上下文槽 `INITIAL_SKILL_KEY`（`tuiInitialSkill`）把 skill 名称传给 TUI，与 `CONFIGURED_AGENT_IDENTITIES_KEY`/`TUI_GOODBYE_MESSAGE_KEY` 一致：`ctx.provide` 是从启动器 argv 进入 Loader 挂载插件的唯一通道。TUI 的 `apply()` 读取该槽并折叠进 `config.initialSkill`；`ui.start()` 成功后，`createTuiChat` 在其被设置时调用一次 `invokeSkill(config.initialSkill, '')`。
+
+**新鲜性在启动器而非 TUI 中把关。** `runSkillSession` 总是创建全新会话，且仅在 `resumeSessionId === undefined` 时提供该槽，因此之后 `dsh --resume <id>` 恢复该会话时是普通 TUI 会话，不会重复注入。TUI 保持通用：它只是把接到的 skill 在启动时调用一次。
+
+**`migrate`/`upgrade` 不接受任何选项。** 与 `meta` 不同，它们不带 `--resume`、`--config` 或 `-p`；引导式全新会话入口没有可恢复或可重配置的内容。任何泄漏的默认界面选项都会明确报错，与 Commander 适配器中 `web`/`meta` 的拒绝模式一致。两个 mode 共用一个 `SkillSessionInvocation` 判别式（`mode: 'migrate' | 'upgrade'`）；`bin.ts` 将 mode 映射为 `dsh-${mode}`。
+
+`dsh-migrate` skill 内置于 `skills/`（经 `DSH_BUNDLED_SKILL_DIR` 交付，与 `dsh-upgrade` 相同）。若未说明源 agent，它会先询问是哪个（opencode/pi/Claude Code/Codex），再把每项能力——workspace 指令、个人覆盖、skills、hooks、MCP、API/env——映射到对应的 DSH 等价物，并基于仓库实际的表面（`hooks-claude`/`hooks-codex` 桥、`~/.dsh/{config.yaml,.env,AGENTS.md,skills/}`、`AGENTS.md`/`CLAUDE.md`、`mcporter`）落地；当某能力无等价物时明确说明。
+
+## 测试
+
+`apps/cli/tests/args.spec.ts` 新增 `migrate`/`upgrade` 的路由（裸判别式），以及每个子命令两侧任一泄漏选项的退出码 1。
+
+`packages/ui/tui/tests/tui.spec.ts` 在既有 skill describe 块中新增两个伪终端用例：设置 `config.initialSkill` 时无需用户输入即把渲染后的 skill 正文作为首轮投递；未知的初始 skill 以通知形式报告且不发送。`runSkillSession` 本身是模块 `v8 ignore` 块内的组装，与 `runTui`/`runMeta` 相同。
+
+无 keyless PTY 快照：依据维护者对本次改动的范围裁定，单元覆盖加交互式验证已足够，且播种走的是已有快照的 `/skill:` 渲染路径。两个命令均已在 tmux 中从临时 cwd 交互式验证：`dsh migrate` 加载 `dsh-migrate` 并询问源 agent；`dsh upgrade` 加载 `dsh-upgrade`，后者引入 `dsh-customize` 并开始 checkout 发现。
+
+## 考虑过的替代方案
+
+**预填输入框并让用户按回车。** 已否决：需要新增编辑器预填 seam，且仍需一次按键。自动提交复用 `invokeSkill`，实现预期的一命令入口。
+
+**播种自然语言指令（“使用 dsh-migrate skill……”）而非 `/skill:<name>`。** 在此否决：字面 skill 调用路径会确定性地把 skill 正文渲染进首轮，与手动命令完全一致，而不依赖模型自行选择加载该 skill。
+
+**在 `migrate`/`upgrade` 上支持 `--resume`。** 已否决：它们是一次性引导入口。恢复的会话是可经默认界面 `dsh --resume <id>` 到达的普通 TUI 会话；恢复时重新注入 skill 会重复首轮。
+
+**在 TUI 之外读取 `INITIAL_SKILL_KEY`（如同 `agent-loop` 读取 `CONFIGURED_AGENT_IDENTITIES_KEY` 那样），而非在 TUI 的 `apply()` 中。** 无此必要：`initialSkill` 是在 `createTuiChat` 中消费的 TUI `Config` 字段，因此在 TUI 入口处把该槽位折叠进配置，可以让它与其他由启动器持有的运行时读取（`tuiResumeHost`、`tuiGoodbyeMessage`）并列，且不触及任何其他插件。
+
+## 后果
+
+迁移或升级从任何位置都只需一条命令，且引导 skill 已被调用。启动器→TUI 的初始 skill 槽可被未来任何引导式会话命令复用；TUI 的契约是“在启动时调用一次这个具名 skill”，而新鲜性/恢复策略留在拥有会话身份的启动器一侧。[TUI skill 斜杠命令](2026-07-21-tui-skill-slash-command.md)仍是该机制；本 note 在其之上新增了一个由启动器驱动的自动调用，并未取代它。

.agents/notes/implemented/feature/2026-07-28-dsh-meta-source-workspace.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-28-dsh-meta-source-workspace.md
+2026-07-28-dsh-meta-source-workspace.md: d65e0e6ff092b63931dd58c52fa76fb76a071dff
+2026-07-28-dsh-meta-source-workspace.zh.md: 72a0c65e6eeda7d63dccc2306a3d7625cdb54362

.agents/notes/implemented/feature/2026-07-28-dsh-meta-source-workspace.md

@@ -0,0 +1,49 @@
+# Agent Note: `dsh meta` boots the TUI over the harness checkout
+
+Status: implemented
+
+English | [中文](2026-07-28-dsh-meta-source-workspace.zh.md)
+
+## Problem
+
+`dsh` treats the invoking directory as the workspace, which is what makes it useful on arbitrary projects. Working on dsh itself therefore means `cd`-ing to the checkout first — and the checkout is not a memorable path: the source install keeps it under a container directory as a timestamped staging worktree (`~/.dsh/source/staging-<timestamp>`) behind a `current` symlink, so the target moves on every upgrade. The agent is already *told* where its source lives by the `harness:source` prompt section, and the `cordis` toolset can modify that runtime, but the human still had to locate the directory by hand to start a session there.
+
+## Decision
+
+`dsh meta` boots the ordinary TUI with the harness checkout as the workspace, from any directory.
+
+The target is `SOURCE_ROOT` in `apps/cli/src/tui.ts` — `fileURLToPath(new URL('../../..', import.meta.url))`, three hops up from `apps/cli/{src,lib}` — the same constant the `harness:source` prompt section already names, so the workspace and the path advertised to the model cannot drift. It follows the launcher's real path, so a PATH symlink through `current` resolves to whichever staging worktree is active.
+
+The mechanism is one `process.chdir(workspace)` inside `runTui`, guarded by a new optional third parameter that only `runMeta` passes. The cwd *is* the workspace seam in the shipped tree: `examples/tui-agent/cordis.yml` derives the session cwd (`!!js process.cwd()`), the `./.sessions` persistence root, and the HMR watch root (`root: ['.']`) from it, so one chdir moves all three together and meta sessions land in the checkout's gitignored `.sessions/`. It runs after both `.env` layers are loaded — the bin's invoking-directory load and the personal one — so the ambient > project > personal precedence is untouched. `DEFAULT_CONFIG` and `SOURCE_ROOT` are absolute and TUI mode passes no snapshot mode, so config resolution is chdir-independent.
+
+`meta` accepts only `--resume <id>`. `--config` would boot a foreign tree against the harness workspace, which is the `--config` case rather than this one; `-p` is not interactive. Both fail loud, as does an empty `--resume=` — matching the default surface, where a swallowed empty id would silently start a fresh session.
+
+**`meta` does not redeclare `--resume`.** Commander parses an option a subcommand shares with its parent into `program.opts()` and leaves the subcommand's own options object empty, so redeclaring it silently dropped the id (found by probing the adapter, not by review). The action reads `program.opts()`, which also accepts the flag on either side of the subcommand; `--help` still lists it among the parent's options.
+
+## Testing
+
+`apps/cli/tests/args.spec.ts` extends its two existing cases rather than adding a file: routing for `meta`, `meta --resume <id>`, and `--resume <id> meta` (pinning the shared-option behavior above), and exit-1 for `meta --resume=`, `meta --config`, and `meta -p`. `runMeta` itself is composition inside the module's existing `v8 ignore` block, like `runTui`.
+
+There is no keyless PTY smoke for this mode. The smoke harness gives each run a temp cwd, but `dsh meta` deliberately chdirs to the real checkout, so a smoke would write `.sessions/` into the live tree mid-test. Covering it properly needs an injectable target directory — a test-only seam this note declines to add for a one-line chdir.
+
+The mode was verified interactively instead. Launched from `$HOME`, a `pwd` tool call reports the checkout, git resolves to its branch, the session log lands under the checkout's `.sessions/` (leaving `~/.sessions` untouched and the tree free of unignored residue), and plain `dsh` from another directory still uses the invoking one.
+
+`dsh meta --resume <valid-id>` once started a *fresh* session instead of resuming — a pre-existing defect on the default surface, not one this mode introduced. [Launcher-owned resume identity](../architecture/2026-07-28-launcher-owned-resume-identity.md) found the cause and fixed it: a personal overlay had replaced the whole `tui-agent` config block, overwriting the shipped `resumeSessionId` intake with a read of an unset environment variable, so a valid id was silently ignored. Session identity is now a launcher-owned context slot that no config key can displace, and `meta` routes through it.
+
+## Alternatives considered
+
+**Thread an explicit workspace through `boot` and the config tree.** Avoids mutating process-wide state, but the shipped config reads the cwd in three places (`!!js process.cwd()`, `persistenceRoot`, HMR `root`), so each would need its own new plumbing and config key to stay consistent. `chdir` before boot expresses "this is the workspace" once, at the seam that already means it.
+
+**A `--meta` flag on the default surface.** Rejected: the default surface is option-only so that subcommands do not collide with a positional, and a flag that silently relocates the workspace reads as a modifier of the current directory rather than a different target. `meta` alongside `web` matches the existing shape.
+
+**Resolve `~/.dsh/source/current` instead of the launcher's own path.** Rejected: it would diverge from the `harness:source` prompt path whenever a non-installed checkout's `bin/dsh` is invoked directly, telling the model one source root while working in another.
+
+**Make the printed resume hint mode-aware.** Deferred here as a known cost, then delivered by [launcher-owned resume identity](../architecture/2026-07-28-launcher-owned-resume-identity.md): the exit line became a launcher-provided context slot, so meta mode prints `dsh meta --resume <id>` and a copied hint works from any directory. It previously came from static config as `dsh --resume {session}` and only worked when re-run from the checkout.
+
+## Consequences
+
+Starting a session on dsh's own source is `dsh meta` from anywhere, and the workspace is guaranteed to be the same checkout the model is told about. Meta sessions are isolated in the checkout's `.sessions/`, so `dsh meta --resume` sees only other meta sessions — intended, since a session's logged cwd belongs to its workspace.
+
+The resume hint was this mode's original cost and is now resolved. [Launcher-owned resume identity](../architecture/2026-07-28-launcher-owned-resume-identity.md) made both the printed line and the in-place `/resume` handoff reproduce the mode as `dsh meta --resume <id>` from one shared argv helper, so a copied hint works from any directory and the handoff no longer depends implicitly on `execve` preserving the process cwd.
+
+`runTui` gains an optional third parameter, so the workspace override is visible at the one function that owns TUI composition rather than hidden in a second copy of it.

.agents/notes/implemented/feature/2026-07-28-dsh-meta-source-workspace.zh.md

@@ -0,0 +1,49 @@
+# Agent Note：`dsh meta` 以 harness 检出为 workspace 启动 TUI
+
+Status: implemented
+
+[English](2026-07-28-dsh-meta-source-workspace.md) | 中文
+
+## Problem
+
+`dsh` 把调用目录视为 workspace，这正是它能作用于任意项目的原因。但因此，开发 dsh 自身就得先 `cd` 到检出目录——而该目录并不是一个好记的路径：源码安装会把它放在一个容器目录下、作为带时间戳的 staging 工作树（`~/.dsh/source/staging-<timestamp>`），并由 `current` 符号链接指向，因此每次升级后目标都会变化。`harness:source` 提示词段已经*告知* agent 其源码位置，`cordis` 工具集也能修改该运行时，但人类仍需手工定位该目录才能在其中开始会话。
+
+## Decision
+
+`dsh meta` 在任意目录下都以 harness 检出为 workspace 启动普通 TUI。
+
+目标是 `apps/cli/src/tui.ts` 中的 `SOURCE_ROOT`——`fileURLToPath(new URL('../../..', import.meta.url))`，从 `apps/cli/{src,lib}` 向上三级——与 `harness:source` 提示词段所用的常量完全相同，因此 workspace 与告知模型的路径不可能发生偏离。它跟随启动器的真实路径，所以经由 `current` 的 PATH 符号链接会解析到当前生效的那个 staging 工作树。
+
+机制是 `runTui` 内的一次 `process.chdir(workspace)`，由一个新的可选第三参数把守，只有 `runMeta` 会传入。在已交付的配置树中，cwd *就是* workspace 的接缝：`examples/tui-agent/cordis.yml` 由它派生出会话 cwd（`!!js process.cwd()`）、`./.sessions` 持久化根目录以及 HMR 监视根目录（`root: ['.']`），因此一次 chdir 会让三者一并移动，meta 会话则落在检出目录中被 gitignore 的 `.sessions/` 内。它在两层 `.env` 都加载之后执行——bin 对调用目录的加载与个人层加载——因此“环境中已有的值 > 项目 > 个人”的优先级不受影响。`DEFAULT_CONFIG` 与 `SOURCE_ROOT` 都是绝对路径，且 TUI 模式不传 snapshot mode，所以配置解析与 chdir 无关。
+
+`meta` 只接受 `--resume <id>`。`--config` 会以 harness workspace 启动其他配置树，那属于 `--config` 的场景而非本场景；`-p` 并非交互式。两者都会明确报错，空的 `--resume=` 亦然——与默认界面一致，在那里被吞掉的空 id 会静默开启一个新会话。
+
+**`meta` 不重新声明 `--resume`。** 对于子命令与父命令共享的选项，Commander 会将其解析进 `program.opts()`，而把子命令自身的 options 对象留空；因此重新声明会静默丢弃该 id（这是通过实测适配器发现的，而非评审发现）。action 读取 `program.opts()`，这同时也允许该标志出现在子命令的任意一侧；`--help` 仍会在父命令的选项中列出它。
+
+## Testing
+
+`apps/cli/tests/args.spec.ts` 扩展其已有的两个用例而非新增文件：`meta`、`meta --resume <id>` 与 `--resume <id> meta` 的路由（钉住上述共享选项行为），以及 `meta --resume=`、`meta --config`、`meta -p` 的退出码 1。`runMeta` 自身与 `runTui` 一样，属于该模块既有 `v8 ignore` 块内的组合代码。
+
+该 mode 没有 keyless PTY 冒烟测试。冒烟框架会为每次运行提供临时 cwd，但 `dsh meta` 刻意 chdir 到真实检出目录，因此冒烟测试会在测试中途把 `.sessions/` 写入实际工作树。要正确覆盖它需要一个可注入的目标目录——为了一行 chdir 而引入的测试专用 seam，本 note 不予采纳。
+
+取而代之的是交互式验证。从 `$HOME` 启动后，`pwd` 工具调用报告的是该检出目录，git 解析到其分支，会话日志落在该检出的 `.sessions/` 下（`~/.sessions` 未被触及，工作树也没有未被忽略的残留），并且从其他目录运行的普通 `dsh` 仍使用调用目录。
+
+`dsh meta --resume <有效 id>` 曾经开启一个*新*会话而非恢复——这是默认界面上既已存在的缺陷，并非本 mode 引入。[由启动器持有的会话身份与退出行](../architecture/2026-07-28-launcher-owned-resume-identity.md) 查明了原因并将其修复：一个个人 overlay 替换了整个 `tui-agent` 配置块，用对一个未设置的环境变量的读取覆盖了已交付的 `resumeSessionId` 入口，因此有效的 id 会被静默忽略。会话标识如今是一个启动器拥有的上下文槽位，没有任何配置键能取代它，而 `meta` 经由它进行路由。
+
+## Alternatives considered
+
+**通过 `boot` 与配置树显式传递 workspace。** 这可避免修改进程级状态，但已交付的配置在三处读取 cwd（`!!js process.cwd()`、`persistenceRoot`、HMR `root`），每一处都需要各自新增管线与配置键才能保持一致。启动前 chdir 只在本就表达该含义的接缝上表达一次“这就是 workspace”。
+
+**在默认界面上加一个 `--meta` 标志。** 拒绝：默认界面是纯选项形式，以免子命令与位置参数冲突；而一个会静默改变 workspace 的标志读起来像是对当前目录的修饰，而非另一个目标。`meta` 与 `web` 并列符合既有形态。
+
+**解析 `~/.dsh/source/current` 而非启动器自身路径。** 拒绝：当直接调用某个非安装检出的 `bin/dsh` 时，它会与 `harness:source` 提示词路径产生偏离——告知模型一个源码根目录，却在另一个目录中工作。
+
+**让打印的恢复提示随 mode 变化。** 在此作为已知代价推迟，随后由 [由启动器持有的会话身份与退出行](../architecture/2026-07-28-launcher-owned-resume-identity.md) 交付：退出行变成了一个启动器提供的上下文槽位，因此 meta 模式打印 `dsh meta --resume <id>`，被复制的提示在任意目录下都有效。它此前来自静态配置，固定为 `dsh --resume {session}`，且只有在检出目录中重新运行才有效。
+
+## Consequences
+
+在 dsh 自身源码上开启会话变成了在任意位置执行 `dsh meta`，且该 workspace 必然就是告知模型的那个检出目录。meta 会话被隔离在检出目录的 `.sessions/` 内，因此 `dsh meta --resume` 只能看到其他 meta 会话——这是预期行为，因为会话记录的 cwd 属于它的 workspace。
+
+恢复提示曾是本 mode 的原初代价，如今已解决。[由启动器持有的会话身份与退出行](../architecture/2026-07-28-launcher-owned-resume-identity.md) 让打印的行与原地 `/resume` 移交都从同一个共享的 argv 辅助函数将该 mode 复现为 `dsh meta --resume <id>`，因此被复制的提示在任意目录下都有效，且移交不再隐式依赖于 `execve` 保留进程 cwd。
+
+`runTui` 新增一个可选第三参数，因此 workspace 覆盖是在拥有 TUI 组合逻辑的那唯一一个函数上可见的，而不是隐藏在它的第二份副本中。

.agents/notes/implemented/feature/2026-07-30-web-context-injection-disclosure.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-web-context-injection-disclosure.md
+2026-07-30-web-context-injection-disclosure.md: 84c3259f3f226e501a671cc55cacf7d7d96f61fb
+2026-07-30-web-context-injection-disclosure.zh.md: 4d77e06e27badb02fb73ca2ea2a739b33c5804de

.agents/notes/implemented/feature/2026-07-30-web-context-injection-disclosure.md

@@ -0,0 +1,33 @@
+# Agent Note: Web context injection disclosure
+
+Status: implemented
+
+English | [中文](2026-07-30-web-context-injection-disclosure.zh.md)
+
+## Problem
+
+The Web conversation rendered every logged non-user message through the generic `JsonBlock`. That presentation used a textual triangle, compact label typography, a bordered JSON panel, and unrelated spacing, so context injection did not match the Tool calls disclosure shown in the product design. Restyling the generic primitive would also change unknown events and attachment fallbacks.
+
+## Decision
+
+`MessageItem` routes context nodes to `ContextInjectionRow`. The row starts collapsed, names the presentation `上下文注入`, uses the existing browse glyph, and exposes the whole 24px header as one pointer and keyboard disclosure target. Its expanded body begins 4px below the header at the shared 22px content indent and renders the design's 141px scrollport with 8px radius, code-block background, 11/16 code text, and no border.
+
+`ContextInjectionRow` serializes both logged `content` and `source` into one inline JSON value, preserving provenance alongside model-visible material. The display remains bounded by the existing 20,000-character truncation policy. It changes no session event, runtime fold, or context-producing plugin.
+
+The package-internal `DisclosureRow` owns the header geometry, icon-to-chevron transition, controlled open state, and Enter/Space behavior shared by context and `ToolRow`. `ToolRow` remains the semantic owner of tool state, summaries, file links, and expanded tool bodies. Context does not enter the keyed toolview slot and gains no context-specific slot while all context sources share one presentation.
+
+## Verification
+
+Conversation component tests pin the collapsed default, browse glyph, whole-row pointer and keyboard toggles, inline JSON shape, truncation, and unchanged generic unknown-event rendering. The keyless assembled-Web history scenario injects context through the real Agent API, records the collapsed row in its ARIA golden, and measures the design's icon, header, indent, gap, scrollport, padding, radius, typography, color, and overflow in Chromium.
+
+## Alternatives considered
+
+**Restyle `JsonBlock` globally.** Unknown surface events and miscellaneous content blocks use that primitive for a separate generic fallback, so a global visual change would couple unrelated presentations.
+
+**Render context as a read tool.** Reusing `ToolRow` directly would add false tool semantics, state and keyed dispatch to a logged non-user message.
+
+**Add a keyed context-view slot.** Every current context source uses the same title and provenance body. A registration seam has no present consumer and can be added without changing the row if distinct source-owned presentations emerge.
+
+## Consequences
+
+Context injection matches the Tool calls visual language without changing its durable meaning. The shared disclosure header prevents the two rows from drifting, while the dedicated context body and generic `JsonBlock` remain independently evolvable. The fixed-height body trades automatic expansion for a stable transcript rhythm and requires scrolling to inspect long injected instructions.

.agents/notes/implemented/feature/2026-07-30-web-context-injection-disclosure.zh.md

@@ -0,0 +1,33 @@
+# Agent Note: Web 上下文注入展开项
+
+Status: implemented
+
+[English](2026-07-30-web-context-injection-disclosure.md) | 中文
+
+## 问题
+
+Web 会话原本通过通用 `JsonBlock` 渲染每条已记录的非用户消息。这种呈现使用文本三角符号、紧凑的标签字体、有边框的 JSON 面板和另一套间距，因此上下文注入与产品设计中的 Tool calls 展开项不一致。修改通用原语的样式还会影响未知事件和附件的兜底呈现。
+
+## 决策
+
+`MessageItem` 将上下文节点路由至 `ContextInjectionRow`。该行初始折叠，标题为 `上下文注入`，使用现有的浏览图标，并使整个 24px 标题栏成为可通过指针和键盘操作的展开目标。其展开主体从标题栏下方 4px 处开始，与共用的 22px 内容缩进对齐，并渲染设计规定的 141px 滚动区；滚动区采用 8px 圆角、代码块背景、11/16 代码文本且无边框。
+
+`ContextInjectionRow` 将已记录的 `content` 和 `source` 序列化为一个内联 JSON 值，在模型可见内容旁保留来源信息。显示内容继续受现有的 20,000 字符截断策略约束。该变更不修改任何会话事件、运行时折叠逻辑或上下文生成插件。
+
+包内部的 `DisclosureRow` 负责上下文行和 `ToolRow` 共用的标题栏几何、图标至折叠箭头的过渡、受控打开状态，以及 Enter／Space 操作。`ToolRow` 仍是工具状态、摘要、文件链接和展开后工具主体的语义 owner。所有上下文来源共用同一套呈现；上下文不会进入键控 toolview slot，也不会获得上下文专用 slot。
+
+## 验证
+
+会话组件测试固定验证初始折叠状态、浏览图标、整行的指针与键盘切换、内联 JSON 形状、截断，以及通用未知事件渲染保持不变。无密钥的组装后 Web 历史场景通过真实 Agent API 注入上下文，在 ARIA 预期输出中记录折叠行，并在 Chromium 中测量设计规定的图标、标题栏、缩进、间隙、滚动区、内边距、圆角、字体排版、颜色和溢出行为。
+
+## 考虑过的替代方案
+
+**全局重新设置 `JsonBlock` 样式。** 之所以否决：未知 surface 事件和其他内容块使用该原语作为独立的通用兜底呈现，全局视觉变更会把无关的呈现耦合起来。
+
+**将上下文渲染为 read 工具。** 之所以否决：直接复用 `ToolRow` 会为已记录的非用户消息添加错误的工具语义、状态和键控分发。
+
+**新增键控 context-view slot。** 之所以否决：当前所有上下文来源都使用相同的标题和来源信息主体，注册 seam 暂无消费方。如果将来出现由不同来源拥有的呈现，仍可在不更改该行的情况下添加此 seam。
+
+## 后果
+
+上下文注入与 Tool calls 采用一致的视觉语言，同时不改变其持久保存的语义。共用的展开项标题栏可以防止这两种行逐渐偏离，而专用的上下文主体与通用 `JsonBlock` 仍可独立演进。固定高度的主体以无法随内容自动增高为代价，为 transcript（文本记录）维持稳定的排版节奏；查看较长的注入指令时必须滚动。

.agents/notes/implemented/process/2026-07-14-typescript-program-backed-semantic-gates.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-14-typescript-program-backed-semantic-gates.md: 43a7b9b5369feb199721f5f1348c03cde66ee411
-2026-07-14-typescript-program-backed-semantic-gates.zh.md: 1ab027d723e30007e6675ae1f3589fb594d10afc
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-14-typescript-program-backed-semantic-gates.md
+2026-07-14-typescript-program-backed-semantic-gates.md: 91639d53b660c68ae52c7ddcef6b3f594e82273e
+2026-07-14-typescript-program-backed-semantic-gates.zh.md: 2270408564f0fc90241255dfb86a65c30362d651

.agents/notes/implemented/process/2026-07-14-typescript-program-backed-semantic-gates.md

@@ -30,6 +30,8 @@ The wrapper owns config diagnostics, semantic compiler options, repository-relat
 
 Context and agent-dispatch calls contribute only finite string-literal event sets. Direct `EventsService.dispatch()` calls recover the event slot through array literals, constant aliases, conditional branches, and resolved call sites of non-exported local helpers. Generic forwarding parameters are not concrete producers: attribution stays with the call sites that supply a closed event value.
 
+Semantic queries run only where a branch can consume them: calls are prefiltered by the closed event-API method-name set before receiver classification, and helper call sites are indexed on demand instead of eagerly resolving every call in every package source. The demand-driven index proves locality per helper — a helper that is non-exported, sits in a real ES module, and whose every same-file reference is a direct callee has all of its calls in that file by module scoping, so only that file is indexed. Any unproven premise (an export modifier, a global script file, an aliasing or otherwise unclassifiable reference) falls back to the original full package-source index, which is the unchanged original semantics; the proof affects cost, never results. A lazy single global index was rejected because the helper-parameter path is reached on the current tree, so it would still pay nearly the whole `getResolvedSignature` sweep.
+
 Every declared harness event must have a discovered producer. A missing producer fails generation as dead vocabulary or an unsupported semantic dispatch shape; listener-free extension points remain valid. `internal/dispatch` instrumentation is not treated as a subscription to every event it observes, so the matrix contains direct product listeners rather than manually asserted indirect relationships.
 
 ### B. Scoped-event routing generates one typed resolver map

.agents/notes/implemented/process/2026-07-14-typescript-program-backed-semantic-gates.zh.md

@@ -30,6 +30,8 @@ Status: implemented
 
 Context 与 AgentEventDispatch 调用只贡献有限的字符串字面量事件集合。对于直接调用 `EventsService.dispatch()` 的路径，生成器会沿数组字面量、常量别名、条件分支和未导出本地辅助函数的已解析调用点恢复事件槽位。泛型转发参数不算作具体生产方：事件仍归属于传入封闭事件值的调用点。
 
+语义查询只在存在消费分支的位置运行：调用先经过封闭的事件 API 方法名集合预过滤，再做接收者分类；辅助函数调用点索引按需构建，而不是预先对全部包源码的每个调用求解签名。需求式索引对每个辅助函数逐一证明局部性——未导出、位于真正的 ES 模块文件中、且同文件所有引用都是直接调用位的辅助函数，按模块作用域规则其全部调用必在本文件内，此时只索引该文件。任一前提无法证明（带导出修饰符、位于全局 script 文件、存在别名化或无法归类的引用）即回退到原全部包源码索引，回退路径就是原语义本身：证明只影响开销，不影响结果。惰性单一全局索引方案被否决，因为当前源码树确实会走到辅助函数参数路径，该方案仍需支付几乎全额的 `getResolvedSignature` 扫描成本。
+
 每个已声明的 harness 事件都必须存在扫描得到的生产方。找不到生产方时，生成过程会将其视为无调用方的事件词汇或尚不支持的语义 dispatch 形态并明确失败；没有监听方的扩展点仍然合法。`internal/dispatch` 插桩不会被当作它所观察的每个事件的订阅，因此关系矩阵只记录直接的产品监听方，不再手工补充间接关系。
 
 ### B. 带作用域的事件路由生成一份强类型解析函数表

.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
-2026-07-22-evidence-based-larger-hosted-runners.md: 983d5520bd73fc3cf82c37bf0d4a9ff1c6e6f51c
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: a86dcf2c60d7b950e7557e84ef6993e712a2ce09
+2026-07-22-evidence-based-larger-hosted-runners.md: d46b8291ec05e997728da76354354f9e36bd2fb4
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: e05ad30a713258ed7bc3d8099f8d6fab3d7c0c5d

.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md

@@ -18,9 +18,9 @@ The required primary path depends on those enterprise pools. Standard GitHub-hos
 
 The former gate-level and coarse primary shard jobs are absent from the workflow. Their static, lint, coverage, snapshot, and scenario shard selectors are also absent from the repository, so an unused diagnostic path cannot preserve a second CI architecture.
 
-Linux primary work uses three independent 32-core jobs. Coverage runs alone with its own worker bound, and the static scheduler runs alone so its result has no post-build consumer tail. After static gates finish, that job publishes its emitted `apps/*/lib`, `packages/*/*/lib`, and `vendor/*/lib` tree as a run-scoped artifact. The third job restores that exact tree, then starts lint, Node 24 runtime compatibility, build-backed snapshots, and all artifact consumers without repeating the build. Generated NodeNext consumer directories are excluded from Oxlint discovery because the artifact check removes them while these processes overlap. The pnpm store is restored without putting cache uploads on the pull-request critical path; Oxlint has no repository-managed result cache. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
+Linux primary work uses three independent 32-core jobs. Coverage runs alone with its own worker bound, and the static scheduler owns source and documentation gates that do not consume emitted output. The third job owns the single Linux build, then starts lint, Node 24 runtime compatibility, build-backed snapshots, documentation typechecking, and all artifact consumers against that tree. This [independent consumer build](2026-07-30-independent-ci-consumer-build.md) lets all three jobs request runners immediately without duplicating compilation or transferring a run-scoped artifact. Generated NodeNext consumer directories are excluded from Oxlint discovery because the artifact check removes them while these processes overlap. The pnpm store is restored without putting cache uploads on the pull-request critical path; Oxlint has no repository-managed result cache. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
 
-The gate dependencies remain explicit. Coverage consumes source and does not wait for build. Documentation typechecking builds its complete project-reference graph once. Snapshot replay and publication consumers wait for emitted output, while Node-version compatibility jobs exercise runtime-sensitive source loading without repeating the primary source-graph typecheck. PTY and subprocess suites keep their bounded inner concurrency rather than inheriting the runner's core count.
+The gate dependencies remain explicit. Coverage consumes source and does not wait for build. Documentation typechecking consumes the consumer lane's complete project-reference output. Snapshot replay and publication consumers wait for emitted output, while Node-version compatibility jobs exercise runtime-sensitive source loading without repeating the primary source-graph typecheck. PTY and subprocess suites keep their bounded inner concurrency rather than inheriting the runner's core count.
 
 The artifact boundary remains explicit. `scripts/publint-all.ts` calls publint's supported API against an in-memory publication view formed from each manifest's declared files plus npm's mandatory metadata, avoiding one package-manager pack process per package. `scripts/verify-built-package-invariants.mjs` stages the declared `lib/` files below the real package and imports its compiled self-reference through plain Node and Cordis Loader normalization; a runtime chunk omitted from the publication contract still fails.
 
@@ -68,7 +68,7 @@ An additional serial Linux reference runs on the in-house self-hosted pool (`vm-
 
 **Keep build behind typecheck.** This orders independent compiler invocations and turns snapshot replay into a three-stage critical chain. Build output has its own success dependency, so only snapshot and publication consumers wait for it.
 
-**Keep static gates and post-build consumers on one runner.** Reusing one workspace avoids a setup wave and artifact transfer, but build-duration variance delays every consumer and leaves their lint and snapshot tails after the static result. A run-scoped built tree preserves one exact build while independent jobs keep both complete paths within the observed target.
+**Publish the static job's build to post-build consumers.** A run-scoped artifact preserves one exact build, but the workflow can only consume it by waiting for the entire static job and then requesting another runner. The [independent consumer build](2026-07-30-independent-ci-consumer-build.md) assigns the single Linux build to its actual consumers instead.
 
 **Keep the complete required path on standard GitHub-hosted capacity.** This avoids repository-external runner configuration, but exact-head standard-runner runs remain materially slower and can spend longer queued behind shared capacity. Standard-hosted compatibility and serial references preserve portable evidence without making that slower topology the ordinary primary path.
 
@@ -80,7 +80,7 @@ An additional serial Linux reference runs on the in-house self-hosted pool (`vm-
 
 The required topology pays one setup wave per 32-core lane and retains no shard selectors. Every ordinary pull request consumes paid enterprise Linux and Windows minutes; manual benchmarks add other sizes only when remeasurement is useful.
 
-GitHub rounds each larger-runner execution up to a whole minute, so complete-job measurement exposes both billed time and workflow complexity. Splitting Linux repeats setup twice and transfers one built tree, but isolates coverage, static gates, and post-build consumers from each other's critical paths without repeating the build; consolidating Windows avoids repeating its slower setup.
+GitHub rounds each larger-runner execution up to a whole minute, so complete-job measurement exposes both billed time and workflow complexity. Splitting Linux repeats setup twice, but the consumer lane owns the only built tree and coverage, static gates, and post-build consumers enter runner allocation independently; consolidating Windows avoids repeating its slower setup.
 
 Performance targets are observations, not cancellation deadlines or correctness requirements. Manual all-size and serial suites remain available when image, dependency, scheduler, or pricing changes need remeasurement.
 

.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md

@@ -18,11 +18,11 @@ Status: implemented
 
 原有的门禁级和粗粒度主流程分片作业已从工作流中移除。相应的静态、lint、覆盖率、快照和场景分片选择器也已从仓库中移除，因此未使用的诊断路径无法继续维系第二套 CI 架构。
 
-Linux 主流程使用 3 个相互独立的 32 核作业。覆盖率单独运行，并设有自己的工作线程上限；静态调度器也单独运行，因此构建后的消费方不会拖延其结果。静态门禁完成后，该作业将其生成的 `apps/*/lib`、`packages/*/*/lib` 和 `vendor/*/lib` 目录树作为仅供本次运行使用的产物发布。第三个作业恢复完全相同的目录树，再让 lint、Node 24 运行时兼容性、依赖构建产物的快照和所有产物消费方基于构建完成后的工作树启动，而不重复构建。生成的 NodeNext 消费方目录不会纳入 Oxlint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 会得到恢复，但缓存上传不会进入拉取请求关键路径；Oxlint 没有由仓库管理的结果缓存。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
+Linux 主流程使用 3 个相互独立的 32 核作业。覆盖率单独运行，并设有自己的工作线程上限；静态调度器负责不消费生成输出的源码和文档门禁。第三个作业负责唯一一次 Linux 构建，随后让 lint、Node 24 运行时兼容性、依赖构建产物的快照、文档类型检查和所有产物消费方基于该目录树启动。这种[消费方独立构建](2026-07-30-independent-ci-consumer-build.md)使 3 个作业都能立即请求运行器，而无需重复编译或传输仅供本次运行使用的产物。生成的 NodeNext 消费方目录不会纳入 Oxlint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 会得到恢复，但缓存上传不会进入拉取请求关键路径；Oxlint 没有由仓库管理的结果缓存。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
 
-门禁依赖关系保持显式。覆盖率消费源码，不等待构建。文档类型检查只构建一次完整的 project-reference 图。快照回放和发布消费方等待生成的输出，而 Node 版本兼容性作业会验证对运行时敏感的源码加载，且不重复主源码项目图的类型检查。PTY 和子进程套件继续使用自身有界的内部并发，不继承运行器的核心数。
+门禁依赖关系保持显式。覆盖率消费源码，不等待构建。文档类型检查以消费方通道的完整 project-reference 输出为输入。快照回放和发布消费方等待生成的输出，而 Node 版本兼容性作业会验证对运行时敏感的源码加载，且不重复主源码项目图的类型检查。PTY 和子进程套件继续使用自身有界的内部并发，不继承运行器的核心数。
 
-产物边界保持显式。`scripts/publint-all.ts` 对内存中的发布视图调用 publint 支持的 API；该视图由每个 manifest（元数据清单）声明的文件和 npm 强制要求的元数据组成，从而避免为每个包启动一次包管理器 pack 进程。`scripts/verify-built-package-invariants.mjs` 将已声明的 `lib/` 文件暂存到真实包下，并通过普通 Node 和 Cordis Loader 规范化导入其已编译的自身引用；发布契约只要遗漏一个运行时分片，检查仍会失败。
+产物边界保持显式。`scripts/publint-all.ts` 对内存中的发布视图调用 publint 支持的 API；该视图由每个 manifest（元数据清单）声明的文件和 npm 强制要求的元数据组成，从而避免为每个包（package）启动一次包管理器 pack 进程。`scripts/verify-built-package-invariants.mjs` 将已声明的 `lib/` 文件暂存到真实包下，并通过普通 Node 和 Cordis Loader 规范化导入其已编译的自身引用；发布契约只要遗漏一个运行时分片，检查仍会失败。
 
 Windows 以一次 32 核环境设置同时承载阻塞性构建、生产网站和观测性的构建产物契约。重复的 lint、覆盖率和快照清单由 Linux 承担，因为在 Windows 上运行这些观测性副本会延长付费关键路径，却不会新增任何阻塞性平台契约。
 
@@ -68,7 +68,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 **让构建继续等待类型检查。** 此方案会给相互独立的编译器调用排定先后顺序，并把快照回放变成 3 阶段关键链。构建输出本身有独立的成功依赖关系，因此只有快照和发布消费方需要等待它。
 
-**将静态门禁和构建后消费方保留在同一台运行器上。** 复用同一个工作区可以省去一轮设置和一次产物传输，但构建耗时的波动会延迟每个消费方，并使消费方的 lint 和快照尾段延续到静态结果之后。仅供本次运行使用的已构建目录树可以保留同一份构建结果，而相互独立的作业能让两条完整路径都保持在实测目标内。
+**将静态作业的构建发布给构建后消费方。** 仅供本次运行使用的产物能保留同一份构建结果，但工作流要消费它，只能先等待整个静态作业完成，再请求另一台运行器。[消费方独立构建](2026-07-30-independent-ci-consumer-build.md)则转而让实际消费方负责唯一一次 Linux 构建。
 
 **将完整必需路径保留在 GitHub 标准托管容量上。** 此方案可以避免依赖仓库外部的运行器配置，但标准运行器上的分支头精确运行仍明显更慢，也可能因共享容量而排队更久。标准托管兼容性作业和串行参考流程保留可移植证据，无需让这套较慢的拓扑成为普通主路径。
 
@@ -80,7 +80,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 必需拓扑中的每个 32 核通道只承担 1 轮设置开销，且不保留分片选择器。每个普通拉取请求都会消耗付费的企业级 Linux 和 Windows 运行器分钟数；只有在重新测量有价值时，手动基准测试才会加入其他规格。
 
-GitHub 会把每次大型运行器执行向上取整到整分钟计费，因此完整作业测量能同时呈现计费时长与工作流复杂度。拆分 Linux 会重复两轮设置并传输一份已构建目录树，但能使覆盖率、静态门禁与构建后消费方不再相互进入关键路径，且无需重复构建；合并 Windows 则避免重复其耗时更长的设置。
+GitHub 会把每次大型运行器执行向上取整到整分钟计费，因此完整作业测量能同时呈现计费时长与工作流复杂度。拆分 Linux 会重复两轮设置，但消费方通道拥有唯一一份已构建目录树，且覆盖率、静态门禁与构建后消费方分别进入运行器分配；合并 Windows 则避免重复其耗时更长的设置。
 
 性能目标是观测结果，而非取消截止时间或正确性要求。当映像、依赖、调度器或定价发生变化而需要重新测量时，仍可使用手动全规格和串行套件。
 

.agents/notes/implemented/process/2026-07-30-independent-ci-consumer-build.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-30-independent-ci-consumer-build.md
+2026-07-30-independent-ci-consumer-build.md: ea87d8051a30c282bdf57ddc3226072be8cb7f24
+2026-07-30-independent-ci-consumer-build.zh.md: 1b5faf73711bd522ddf0ae04f870e0401a481c0e

.agents/notes/implemented/process/2026-07-30-independent-ci-consumer-build.md

@@ -0,0 +1,35 @@
+# Agent Note: Independent CI consumer build
+
+Status: implemented
+
+English | [中文](2026-07-30-independent-ci-consumer-build.zh.md)
+
+## Problem
+
+The [larger-runner topology](2026-07-22-evidence-based-larger-hosted-runners.md) gave the static and built-consumer inventories separate jobs, but the static job owned their shared build. It uploaded the emitted tree only after every static gate completed, and the consumer job declared a job-level dependency before restoring that tree. Compiled-output snapshots and publication checks genuinely require a complete build; they do not require runtime-closure checks, documentation generation, module-graph verification, or Knip.
+
+That wider dependency made runner availability part of the required critical chain. In one failover run, static waited 8 minutes 1 second for a runner and ran for 1 minute 41 seconds; only then could consumers enter the same shared pool, where they waited another 10 minutes 34 seconds before running for 1 minute 58 seconds. Reusing the static build saved repository work but serialized two independent runner allocations.
+
+## Decision
+
+The three required Linux jobs enter runner allocation independently. Coverage remains source-only. Static owns source and documentation checks that do not consume emitted output. The consumer job owns the single Linux build together with documentation typechecking, compiled-output snapshots, publication checks, NodeNext checks, and built-bin smokes.
+
+The consumer's internal gate graph preserves the real dependency. Build and source-only Node compatibility start first; publint waits for build, built-package invariants validate that publication view, and every compiled-output consumer waits for that validation. Example and Web snapshots therefore continue to exercise current `lib/` output under plain Node, while no GitHub job waits for an unrelated job or transfers a built-tree artifact.
+
+Windows and serial reference aggregates retain their own build ownership. The change is confined to the required pull-request Linux topology; `all checks passed` still aggregates the same named jobs and fails for any unsuccessful dependency.
+
+## Alternatives considered
+
+**Keep publishing the static job's build.** This preserves one build but cannot express the actual step-level dependency: GitHub makes the consumer wait for the whole static job before it can request a runner. The saved build time is smaller than the repeated queue delay during failover saturation.
+
+**Build independently in both jobs.** Removing the job dependency while leaving build in static would restore parallel allocation, but every pull request would compile the same tree twice. Moving documentation typechecking and build ownership to the consumer preserves one build.
+
+**Add a dedicated build job.** A narrow producer would make the dependency name accurate, but it would add a fourth setup and runner-allocation stage before consumers. The consumer already owns every long-lived use of emitted output, so a separate producer has no second independent consumer.
+
+**Combine static and consumers only during failover.** One long job would avoid the second allocation, but conditional job inventories and result aggregation would create a second CI topology. Independent jobs preserve the same graph on hosted and failover pools.
+
+## Consequences
+
+Static and consumer queue delays overlap instead of accumulating. The consumer's active time includes the build, while the static job becomes shorter and artifact upload, download, compression, and extraction disappear. Total Linux build count remains one.
+
+A static failure no longer prevents the consumer inventory from producing its own evidence; the final verdict still fails. Build and documentation-typecheck failures appear under `node 24 / snapshots and artifacts` rather than `node 24 / static`, matching the job that owns their output dependency.

.agents/notes/implemented/process/2026-07-30-independent-ci-consumer-build.zh.md

@@ -0,0 +1,35 @@
+# Agent Note: 消费方独立构建
+
+Status: implemented
+
+[English](2026-07-30-independent-ci-consumer-build.md) | 中文
+
+## 问题
+
+[大型运行器拓扑](2026-07-22-evidence-based-larger-hosted-runners.md)将静态门禁清单和构建后消费方清单分配给不同作业，但二者共用的构建由静态作业负责。静态作业要等所有静态门禁完成后才上传生成的目录树，消费方作业则在恢复该目录树前声明了作业级依赖。基于编译输出的快照与发布校验确实需要完整构建，但不依赖运行时依赖闭包检查、文档生成、模块图验证或 Knip。
+
+这项过宽的依赖使运行器可用性成为必需关键链的一环。一次故障切换运行中，静态作业等待运行器 8 分 1 秒，随后运行 1 分 41 秒；直到此时，消费方作业才能进入同一个共享池，它又等待 10 分 34 秒，随后运行 1 分 58 秒。复用静态作业的构建省去了部分仓库工作，却让两次原本相互独立的运行器分配串行发生。
+
+## 决策
+
+3 个必需 Linux 作业分别进入运行器分配。覆盖率仍只消费源码。静态作业负责无需消费生成输出的源码检查与文档检查。消费方作业负责唯一一次 Linux 构建，以及文档类型检查、基于编译输出的快照、发布校验、NodeNext 检查和 built-bin 冒烟测试。
+
+消费方内部的门禁图保留实际依赖关系。构建和只消费源码的 Node 兼容性检查率先启动；publint 等待构建完成，已构建包不变式检查会验证该发布视图，所有编译输出消费方都等待这项验证完成。因此，示例和 Web 快照仍会在普通 Node 下验证当前 `lib/` 输出；同时，没有任何 GitHub 作业需要等待无关作业或传输已构建目录树产物。
+
+Windows 与串行参考聚合流程仍各自负责自身构建。本变更仅涉及拉取请求的必需 Linux 拓扑；`all checks passed` 仍聚合同一批具名作业，任一依赖未成功时都会失败。
+
+## 曾考虑的替代方案
+
+**继续发布静态作业的构建。** 此方案只需构建一次，却无法表达实际的步骤级依赖：GitHub 会让消费方等到整个静态作业结束后才可请求运行器。故障切换池饱和时，再次排队的延迟超过了省下的构建时间。
+
+**在两个作业中分别独立构建。** 在静态作业中保留构建、同时移除作业依赖，可以恢复并行分配，但每个拉取请求都会对同一目录树编译两次。将文档类型检查和构建职责移给消费方，则仍只需构建一次。
+
+**新增专用构建作业。** 职责单一的生产方能让依赖名称与实际关系相符，但会在消费方之前新增第 4 个需要设置和分配运行器的阶段。所有需要持续使用生成输出的任务都已由消费方作业负责，因此单独增设生产方也没有第二个相互独立的消费方。
+
+**仅在故障切换期间合并静态作业与消费方作业。** 单个长作业可以避免第二次分配，但带条件分支的作业清单与结果聚合会形成第二套 CI 拓扑。独立作业能让托管池与故障切换池使用同一作业图。
+
+## 后果
+
+静态作业与消费方作业的排队延迟会相互重叠，不再累加。消费方的活动耗时包含构建；静态作业则变短，产物上传、下载、压缩和解压步骤全部消失。Linux 构建总次数仍为 1 次。
+
+静态作业失败不再阻止消费方清单生成自身证据；最终判定仍会失败。构建与文档类型检查失败会归入 `node 24 / snapshots and artifacts` 而非 `node 24 / static`，这一归类与输出依赖的实际归属一致。

.agents/notes/implemented/simplification/2026-07-20-remove-stdio-and-echo-agents.i18n.yaml

@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-20-remove-stdio-and-echo-agents.md: d804f8e3c3de886fc48c87d6378c9f06d0e3c15a
-2026-07-20-remove-stdio-and-echo-agents.zh.md: d42e6d527231e1703153c02c66d9286fa1e73c95
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/simplification/2026-07-20-remove-stdio-and-echo-agents.md
+2026-07-20-remove-stdio-and-echo-agents.md: b578e49f004cabf0e37385ea06a769ced5388217
+2026-07-20-remove-stdio-and-echo-agents.zh.md: 3d8eecf73f17f3b62dc4b35b0f3c1af03f6f3a86

.agents/notes/implemented/simplification/2026-07-20-remove-stdio-and-echo-agents.md

@@ -18,7 +18,7 @@ The stdio and Echo agents are removed without compatibility packages, modes, com
 
 The remaining application roles are explicit:
 
-- [`@deepseek-ai/dsh-tui-demo`](../../../../packages/examples/tui-demo/README.md) owns terminal-interactive execution. It rejects non-TTY streams before Loader boot; `examples/tui-agent` owns the complete coding composition, Code Mode overlay, PTY coverage, and terminal snapshots.
+- [`@deepseek-ai/dsh-tui`](../../../../packages/ui/tui/README.md) owns terminal-interactive execution. It rejects non-TTY streams before Loader boot; `apps/cli/config/base.cordis.yml` plus the `tui.cordis.yml` overlay own the complete coding composition, with PTY plus terminal-snapshot coverage in `apps/cli/tests/`.
 - [`@deepseek-ai/dsh-cli-demo`](../../../../packages/examples/cli-demo/README.md) owns non-interactive execution, including pipes. `examples/headless-agent` owns the real-model one-shot composition, replay snapshots, generic real-agent suites, and test-only keyless Loader fixtures.
 - [`@deepseek-ai/dsh-acp-demo`](../../../../packages/examples/acp-demo/README.md) and `@deepseek-ai/dsh-jsonrpc` own their framed protocol integrations.
 
@@ -30,7 +30,7 @@ Keyless validation is test-owned. The Headless Loader smoke uses a fixture adapt
 
 TUI and Headless Loader coverage run the real app packages in source and built modes. PTY-driven subprocess coverage is reserved for the TUI lifecycle; other entry-point smokes use the one-shot pipe protocol. Headless proves its task/result and tool-call contracts. Generated graphs and repository searches reject stale package, command, leaf, SDK-interface, `createStdioChat`, and `StdioRuntime` references.
 
-The TUI PTY smoke includes the Code Mode overlay composition, while `examples/cordis-agent/tests/keyless-smoke.e2e.ts` provides a minimal PTY boot over the real Cordis-agent Loader tree. The built `dsh` bin rejects a piped TUI launch before Loader boot and points at its one-shot `-p` mode; `apps/cli/tests/built-bin.e2e.ts` pins that path, while `cli-demo`'s built-bin suite runs text, JSON, and structurally parsed `stream-json` output under plain Node, persists fresh sessions, and rejects invalid arguments and missing config without contaminating stdout. Time-context integration uses the real Headless composition for two ordered turns, while its package tests own finer elapsed-time behavior.
+The built `dsh` bin rejects a piped TUI launch before Loader boot and points at its one-shot `-p` mode; `apps/cli/tests/built-bin.e2e.ts` pins that path, while `cli-demo`'s built-bin suite runs text, JSON, and structurally parsed `stream-json` output under plain Node, persists fresh sessions, and rejects invalid arguments and missing config without contaminating stdout. Code Mode has programmatic TUI snapshots and an ACP overlay demo. Time-context integration uses the real Headless composition for two ordered turns, while its package tests own finer elapsed-time behavior.
 
 ## Alternatives considered
 

.agents/notes/implemented/simplification/2026-07-20-remove-stdio-and-echo-agents.zh.md

@@ -18,7 +18,7 @@ DeepSeek Harness 在 TUI 和 Headless coding agent 之外，还提供了两个
 
 保留的应用角色均有明确归属：
 
-- [`@deepseek-ai/dsh-tui-demo`](../../../../packages/examples/tui-demo/README.md) 负责终端交互式执行。它会在 Loader 启动前拒绝非 TTY 流；`examples/tui-agent` 拥有完整 coding 组装、Code Mode 覆盖层、PTY 覆盖和终端快照。
+- [`@deepseek-ai/dsh-tui`](../../../../packages/ui/tui/README.md) 负责终端交互式执行。它会在 Loader 启动前拒绝非 TTY 流；`apps/cli/config/base.cordis.yml` 与 `tui.cordis.yml` overlay 拥有完整 coding 组装，PTY 与终端快照覆盖则位于 `apps/cli/tests/`。
 - [`@deepseek-ai/dsh-cli-demo`](../../../../packages/examples/cli-demo/README.md) 负责非交互式执行，包括管道方式。`examples/headless-agent` 拥有真实模型的单次任务组装、回放快照、通用真实 agent 测试套件，以及仅供测试使用的无密钥 Loader fixture。
 - [`@deepseek-ai/dsh-acp-demo`](../../../../packages/examples/acp-demo/README.md) 和 `@deepseek-ai/dsh-jsonrpc` 负责各自的分帧协议集成。
 
@@ -30,7 +30,7 @@ SDK 工程模型与 create/config 工作流将 `stdio` 运行接口选项替换
 
 TUI 与 Headless 的 Loader 覆盖以源码和构建产物两种模式运行真实 app 包。由 PTY 驱动的子进程覆盖仅用于 TUI 生命周期；其他入口冒烟测试使用单次管道协议。Headless 验证任务/结果契约和工具调用契约。生成图谱与仓库搜索会拒绝陈旧的包、命令、叶节点、SDK 接口、`createStdioChat` 和 `StdioRuntime` 引用。
 
-TUI PTY 冒烟测试包含 Code Mode 覆盖层组装，而 `examples/cordis-agent/tests/keyless-smoke.e2e.ts` 会基于真实 Cordis-agent Loader 目录树执行最小 PTY 启动。构建后的 `dsh` 可执行文件会在 Loader 启动前拒绝通过管道启动 TUI，并指向其单次 `-p` 模式；`apps/cli/tests/built-bin.e2e.ts` 固定了该执行路径，而 `cli-demo` 的 built-bin 套件在普通 Node 下运行文本、JSON 和经过结构化解析的 `stream-json` 输出，持久化新建会话，并在不污染 stdout 的情况下拒绝无效参数和缺失配置。时间上下文集成通过真实 Headless 组装执行两个有序轮次，而更细粒度的耗时行为由时间上下文的包级测试负责。
+构建后的 `dsh` 可执行文件会在 Loader 启动前拒绝通过管道启动 TUI，并指向其单次 `-p` 模式；`apps/cli/tests/built-bin.e2e.ts` 固定了该执行路径，而 `cli-demo` 的 built-bin 套件在普通 Node 下运行文本、JSON 和经过结构化解析的 `stream-json` 输出，持久化新建会话，并在不污染 stdout 的情况下拒绝无效参数和缺失配置。Code Mode 由程序化 TUI 快照与 ACP overlay demo 覆盖。时间上下文集成通过真实 Headless 组装执行两个有序轮次，而更细粒度的耗时行为由时间上下文的包级测试负责。
 
 ## 曾考虑的替代方案
 

.agents/notes/implemented/simplification/2026-07-29-shared-base-config-overlays.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/simplification/2026-07-29-shared-base-config-overlays.md
+2026-07-29-shared-base-config-overlays.md: ee642cbc786bef708791fb58e655c5a3f0e9c4e7
+2026-07-29-shared-base-config-overlays.zh.md: b7fc9c6b121b8d0eb94d734af6bda6df45e25b1d

.agents/notes/implemented/simplification/2026-07-29-shared-base-config-overlays.md

@@ -0,0 +1,53 @@
+# Agent Note: One shared base config with per-surface overlays
+
+Status: implemented
+
+English | [中文](2026-07-29-shared-base-config-overlays.zh.md)
+
+## Problem
+
+`dsh` shipped two full config trees that were 43 rows the same. `apps/cli/cordis.yml` composed the web surface as 74 flat rows, while the TUI booted `examples/tui-agent/cordis.yml`, whose single `@deepseek-ai/dsh-tui-demo` row mounted twelve plugins and re-declared their configuration as its own twenty-key pass-through `Config`.
+
+Neither file was what its location claimed. `examples/tui-agent` was not an example: `apps/cli/src/tui.ts` hardcoded it as the product's default config, and it owned the TUI PTY smoke, the eight terminal snapshot scenarios, and the PTY harness the `cordis-agent` leaf imported. `dsh-tui-demo` was not a demo either — it was the application, mounted by the shipped binary from `packages/examples/`.
+
+The duplication was the load-bearing problem. Of the 43 shared rows, 38 were byte-identical and 5 differed for a defensible per-surface reason, so every capability change had to be made twice and could silently drift. The bundle also inverted a default: `composeTuiApp` read `config.goals ?? {}`, so the shipped TUI mounted goals, `tool-goal`, `goal-session`, and `/goal` although no config key requested them.
+
+## Decision
+
+One shared base, one overlay per surface, composed as sibling patch lists.
+
+`apps/cli/config/base.cordis.yml` holds the 43 rows both surfaces mount. `apps/cli/config/tui.cordis.yml` and `apps/cli/config/web.cordis.yml` are **patch lists**, not trees: each states the handful of rows whose value is surface-specific and inserts its own rows. The launcher includes the base once and applies every overlay as a sibling patch list at **one** include level, because include patches never cross an include boundary — stacking overlays as nested includes would silently stop reaching base rows.
+
+Precedence is list order, last write winning per row: base, then the surface overlay, then either a `--config` overlay or the personal `~/.dsh/config.yaml`, then the launcher's own flag and profile patches.
+
+`--config <path>` now applies an overlay **instead of** the personal overlay, so a demo or test tree never inherits the user's provider and model. `--config-replace <path>` boots a file as the entire tree, bypassing base, surface overlay, and personal overlay alike; that is what the old `--config` did, so trees like `examples/web-cordis` moved to the new flag. Both flags survive the `/resume` execve handoff, or resuming would silently change the agent.
+
+A patch replaces its target row's whole `config` rather than merging, which shapes the split: a row whose value differs per surface lives in the overlays, never in the base, so no row is patched by three layers at once. Session identity therefore cannot ride a config key at all — it moved to `dsh-agent-loop`'s `CONFIGURED_AGENT_IDENTITIES_KEY`, as [the launcher-owned identity note](../architecture/2026-07-28-launcher-owned-resume-identity.md) now records.
+
+`examples/tui-agent`, `examples/cordis-agent`, `examples/code-mode`, and `packages/examples/tui-demo` are deleted. The TUI tests move to `apps/cli/tests/`, the cordis-toolset e2e to `packages/cordis/tool-cordis/tests/`, and the supported Code Mode demo remains the ACP overlay at `examples/acp-agent/code-mode.cordis.yml`.
+
+## Alternatives considered
+
+**Leave both trees flat and duplicated.** Rejected: 43 rows maintained twice is the defect, and a gate asserting they stay identical would freeze the duplication rather than remove it.
+
+**Nest the overlays as includes (`code-mode` → `tui` → `base`).** Rejected after testing the Loader: patches do not cross an include boundary, so the outer file's patches are dropped with only a warning. A three-level chain left `tools` unpatchable, and a base behind one include made every personal patch a silent no-op.
+
+**Put the union of all rows in the base and have each overlay disable what it does not want.** Rejected: the base stops meaning "shared", and each surface carries rows it exists only to switch off.
+
+**Keep the per-surface rows in the base and let overlays patch them.** Adopted only for the five rows that must exist in both trees, because a patch cannot create a row. Their base entries carry the plugin name and the config both surfaces share; each overlay states the rest.
+
+## Consequences
+
+An overlay or `--config` tree that named `@deepseek-ai/dsh-tui-demo`, or patched the `tui-agent` row, no longer resolves. Overlays now patch the row that owns each key: the model route on `agent-loop`, the persona on `system-prompt`, presentation on `tui`.
+
+A patch whose `id` matches no row stays a no-op rather than an error. That is deliberate: one personal overlay is shared across surfaces, and `insert` rows match nothing by design, so a row that exists only under `web` must not fail the TUI's boot.
+
+`dsh web` gains `--config`, threaded into `AppCLIEntry` as an extra overlay. Web keeps sandboxed Bash and filesystem providers plus approval, permission presets, directory picking, and browser permission UI; the overlay disables the shared local providers because patches can disable rows but cannot delete them. The TUI query index uses a unique process-local temporary database because the SQLite backend requires one writer owner. It is a disposable derived index rebuilt by each process; `/resume` lists the underlying corpus directly and does not depend on index reuse. `AppCLIEntry` reads both the base and its surface overlay when recovering row defaults for its own patch merge, since a flag override must preserve the overlay's other fields on the same row.
+
+## Verification
+
+Composition is checked by booting each tree through the real Loader and inspecting settled entries, not by reading YAML; both surfaces settle with zero unloaded rows, and Web starts its `httpServer` with sandboxed Bash and filesystem providers. Code Mode remains covered by the ACP overlay and programmatic TUI snapshots rather than a separate shipped TUI application.
+
+All eight terminal snapshot scenarios replay byte-identically after moving, and the 14-case PTY smoke passes, including two cases that assert a personal overlay reaches an **inserted** row — the behavior the vendored `plugin-include` fix enables ([`vendor/README.md`](../../../../vendor/README.md) local modification 8, covered by `packages/ui/app-boot/tests/config-reload.spec.ts`).
+
+Flattening surfaced three latent defects, each fixed here: the TUI captured the optional `sessionQuery` service once at construction and so could permanently disable `/resume` when it won the mount race; the shipped session-store root silently reverted to a project-local `./.sessions`; and `--config-replace` was dropped by the resume handoff.

.agents/notes/implemented/simplification/2026-07-29-shared-base-config-overlays.zh.md

@@ -0,0 +1,53 @@
+# Agent Note: 一份共享 base 配置加各 surface 的 overlay
+
+Status: implemented
+
+[English](2026-07-29-shared-base-config-overlays.md) | 中文
+
+## 问题
+
+`dsh` 交付了两棵完整的配置树，其中 43 个配置项完全相同。`apps/cli/cordis.yml` 以 74 个平铺配置项组合 web surface，而 TUI 启动的是 `examples/tui-agent/cordis.yml`——其中单独一行 `@deepseek-ai/dsh-tui-demo` 挂载了十二个插件，并把它们的配置重新声明为自己那份二十个键、仅作透传的 `Config`。
+
+这两份文件都名不副实。`examples/tui-agent` 并不是示例：`apps/cli/src/tui.ts` 把它硬编码为产品的默认配置；它还拥有 TUI 的 PTY 冒烟测试、八个终端快照场景，以及被 `cordis-agent` 叶节点 import 的 PTY harness。`dsh-tui-demo` 也不是 demo——它就是应用本身，由交付的二进制从 `packages/examples/` 中挂载。
+
+真正决定性的问题是重复。43 个共享配置项中，38 个逐字节相同，5 个因各 surface 的正当理由而不同；因此每次能力改动都必须改两处，而且可能无声漂移。该组合包还反转了一个默认值：`composeTuiApp` 读取 `config.goals ?? {}`，于是交付的 TUI 挂载了 goals、`tool-goal`、`goal-session` 和 `/goal`——尽管没有任何配置键要求它们。
+
+## 决策
+
+一份共享 base，每个 surface 一份 overlay，以平级 patch 列表的形式组合。
+
+`apps/cli/config/base.cordis.yml` 持有两个 surface 都会挂载的 43 个配置项。`apps/cli/config/tui.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 是 **patch 列表**，不是配置树：各自声明少数取值因 surface 而异的配置项，并 insert 自己的配置项。启动器只 include base 一次，并把每个 overlay 作为**同一** include 层级上的平级 patch 列表应用——因为 include patch 不会跨越 include 边界，把 overlay 堆叠成嵌套 include 会使其静默地无法触达 base 配置项。
+
+优先级即列表顺序，逐配置项后写者胜：base，然后是 surface overlay，接着是 `--config` overlay 或个人 `~/.dsh/config.yaml`，最后是启动器自身的 flag 与 profile patch。
+
+`--config <path>` 现在应用一个 overlay 来**取代**个人 overlay，因此 demo 或测试用的树绝不会继承用户的 provider 与 model。`--config-replace <path>` 则把某个文件作为整棵树启动，同时绕过 base、surface overlay 与个人 overlay；这正是旧 `--config` 的行为，所以像 `examples/web-cordis` 这样的树改用了新 flag。两个 flag 都会在 `/resume` 的 execve 交接中保留，否则 resume 会静默更换 agent。
+
+patch 会整体替换目标配置项的 `config` 而不合并，这决定了拆分方式：取值因 surface 而异的配置项住在 overlay 中，绝不住在 base 里，从而没有任何配置项会被三层同时 patch。因此会话身份根本不能经由配置键传递——它迁移到了 `dsh-agent-loop` 的 `CONFIGURED_AGENT_IDENTITIES_KEY`，如[启动器持有身份的 note](../architecture/2026-07-28-launcher-owned-resume-identity.md) 现在所记录。
+
+`examples/tui-agent`、`examples/cordis-agent`、`examples/code-mode` 与 `packages/examples/tui-demo` 均被删除。TUI 测试迁往 `apps/cli/tests/`，cordis 工具集的 e2e 迁入 `packages/cordis/tool-cordis/tests/`，受支持的 Code Mode demo 则保留为 `examples/acp-agent/code-mode.cordis.yml` 中的 ACP overlay。
+
+## 备选方案
+
+**保留两棵平铺且重复的树。** 拒绝：43 个配置项维护两份正是缺陷本身，而用一个门禁断言二者保持一致只会固化重复，而非消除它。
+
+**把 overlay 嵌套成 include（`code-mode` → `tui` → `base`）。** 在对 Loader 实测后拒绝：patch 不会跨越 include 边界，因此外层文件的 patch 只会伴随一条告警被丢弃。三层链条使 `tools` 无法被 patch，而位于一层 include 之后的 base，会让每个个人 patch 都变成静默的空操作。
+
+**把所有配置项的并集放进 base，由各 overlay 禁用自己不需要的部分。** 拒绝：base 将不再意味着「共享」，而每个 surface 都要携带仅为将其关闭而存在的配置项。
+
+**把因 surface 而异的配置项留在 base 中，由 overlay 去 patch。** 仅对必须同时存在于两棵树中的那五个配置项采用，因为 patch 无法创建配置项。它们在 base 中的条目携带插件名与两个 surface 共享的配置，其余部分由各 overlay 声明。
+
+## 影响
+
+指名 `@deepseek-ai/dsh-tui-demo`、或 patch `tui-agent` 配置项的 overlay 与 `--config` 树将不再可解析。overlay 现在要 patch 拥有对应键的那一行：模型路由在 `agent-loop`，人设在 `system-prompt`，呈现设置在 `tui`。
+
+若某个 patch 的 `id` 不匹配任何配置项，Loader 仍只告警而不报错。这是有意为之：同一份个人 overlay 会跨 surface 共用，而 `insert` 配置项按设计本就不匹配任何目标，因此仅在 `web` 下存在的配置项不能让 TUI 启动失败。
+
+`dsh web` 新增 `--config`，作为一份额外 overlay 传入 `AppCLIEntry`。Web 保留沙箱化 Bash 与文件系统提供方，以及审批、权限预设、目录选择和浏览器权限界面；覆盖层会禁用共享的本地提供方，因为补丁可以禁用条目但不能删除条目。TUI 查询索引使用进程唯一的临时数据库，因为 SQLite 后端要求单写入者所有权。该索引是每个进程重新构建的可丢弃派生数据；`/resume` 直接列出底层语料，不依赖索引复用。`AppCLIEntry` 在为自身 patch 合并恢复配置项默认值时会同时读取 base 与其 surface overlay，因为 flag 覆盖必须保留同一配置项上 overlay 的其他字段。
+
+## 验证
+
+组合的正确性通过用真实 Loader 启动每棵树并检查已就绪的条目来核对，而不是靠阅读 YAML：两个界面都能稳定完成且没有未加载项；Web 会以沙箱化 Bash 与文件系统提供方启动 `httpServer`。Code Mode 继续由 ACP overlay 与程序化 TUI 快照覆盖，而不再维护独立交付的 TUI 应用。
+
+全部八个终端快照场景在迁移后逐字节重放一致，14 个用例的 PTY 冒烟测试全部通过，其中两个用例断言个人 overlay 能触达一个 **insert 进来的**配置项——这正是 vendored `plugin-include` 修复所启用的行为（[`vendor/README.md`](../../../../vendor/README.md) 本地修改第 8 条，由 `packages/ui/app-boot/tests/config-reload.spec.ts` 覆盖）。
+
+平铺过程暴露出三处潜伏缺陷，均在此一并修复：TUI 曾在构造时一次性捕获可选的 `sessionQuery` 服务，因此在挂载竞争中胜出时会永久禁用 `/resume`；交付的会话存储根目录曾静默退回项目本地的 `./.sessions`；`--config-replace` 曾在 resume 交接中被丢弃。

.agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.md
-2026-07-18-tui-terminal-state-snapshots.md: 18c79bc2d0dabf4d78887354f30a2cdc083899e1
-2026-07-18-tui-terminal-state-snapshots.zh.md: d1d4a6ca859e94a153e0bf645a17f03c0dac234b
+2026-07-18-tui-terminal-state-snapshots.md: b8e6f77d96fbd4d077d55fefa3652813c737a37b
+2026-07-18-tui-terminal-state-snapshots.zh.md: 1b2f6b58e53d9ccd5f0935dab9e9812d2bfe3eb4

.agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.md

@@ -21,7 +21,7 @@ TUI coverage has four complementary layers:
 3. `examples/tui-agent/tests/tui.snapshot.ts` replays committed JSONL session logs through the production agent loop and real tools, then compares the resulting semantic terminal state.
 4. `examples/tui-agent/tests/tui-keyless-smoke.e2e.ts` boots the real Loader composition in a PTY, drives a scripted conversation through streaming and `ask_user_question`, and verifies startup, input, exit, failure reporting, and terminal restoration.
 
-The runnable TUI has its own `examples/tui-agent` leaf beside the Headless and ACP leaves. It owns the interactive coding backends and tools directly and loads `@deepseek-ai/dsh-tui-demo`; TUI snapshots and PTY tests live with that leaf. The [redundant-agent removal](../simplification/2026-07-20-remove-stdio-and-echo-agents.md) owns this consolidation.
+The runnable TUI is the shipped `apps/cli` composition: the shared `base.cordis.yml` plus the `tui.cordis.yml` overlay, which owns the interactive coding backends, tools, and front door. TUI snapshots and PTY tests live in `apps/cli/tests/`. The [redundant-agent removal](../simplification/2026-07-20-remove-stdio-and-echo-agents.md) owns this consolidation.
 
 ### Recorded-session replay
 

.agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.zh.md

@@ -21,7 +21,7 @@ TUI 覆盖分为四个互补层次：
 3. `examples/tui-agent/tests/tui.snapshot.ts` 通过生产 agent loop 和真实工具回放已提交的 JSONL 会话日志，再比较生成的语义终端状态。
 4. `examples/tui-agent/tests/tui-keyless-smoke.e2e.ts` 在 PTY 中启动真实 Loader 组合，驱动一段经过流式输出和 `ask_user_question` 的脚本化会话，并验证启动、输入、退出、失败报告和终端恢复。
 
-可运行 TUI 在 `examples/tui-agent` 中拥有独立叶节点，与 Headless 和 ACP 叶节点并列。它直接拥有交互式 coding 后端与工具，并加载 `@deepseek-ai/dsh-tui-demo`；TUI 快照和 PTY 测试也归属这个叶节点。[移除重复 agent 的决策](../simplification/2026-07-20-remove-stdio-and-echo-agents.md)负责此次整合。
+可运行 TUI 就是交付的 `apps/cli` 组合：共享的 `base.cordis.yml` 加 `tui.cordis.yml` overlay，后者拥有交互式 coding 后端、工具与前端入口。TUI 快照和 PTY 测试位于 `apps/cli/tests/`。[移除重复 agent 的决策](../simplification/2026-07-20-remove-stdio-and-echo-agents.md)负责此次整合。
 
 ### 已录制会话回放
 

.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md
-2026-07-24-web-gui-browser-e2e-lane.md: 97be6d5d70d12f783e2c80b1bcd546cfa4582ca8
-2026-07-24-web-gui-browser-e2e-lane.zh.md: f55b4f8011cc42aa4e0f8c11d9900525a565e1a7
+2026-07-24-web-gui-browser-e2e-lane.md: 898b8b5fe1b8d65b108b4afa95b782a1ce1e5c71
+2026-07-24-web-gui-browser-e2e-lane.zh.md: 966a9854aee8f3b63b2ae8f1362f91a40b9ba894

.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md

@@ -16,7 +16,7 @@ The web GUI ships as a real assembled chain — chromium page → client plugin
 
 A plain shared-fixture module (the [testing-policy sanctioned shape](../../../../docs/testing.md)), not a package: the gate-worthy logic — replay derivation, session parsing, log scrubbing, persistence — lives in the gated packages `dsh-llm-replay`, `dsh-acp-snapshot`, and `dsh-session-persistence-jsonl`; what remains is boot wiring and browser glue, and chromium-driving code cannot hold per-file 100% coverage on the browserless coverage runners.
 
-`launchWebScaffold()` boots the real web composition from the shipped `apps/cli/cordis.yml` through the vendored Loader's include mechanism — the same tree and mechanism `AppCLIEntry` drives for `dsh web`. Divergences ride include patches over that tree, the ACP `cordis.snapshot.yml` pattern expressed in-process: temp `persistenceRoot`, `workspace-context` disabled (recorded fixtures must not embed this repo's AGENTS.md), `session-title-llm` disabled (its fire-and-forget title call would race the loop for the session's replay cursor), the webserver row pinned to port 0 with the built dist, and in keyless modes `llm-deepseek` disabled. A patch id that stops matching a row fails the boot sweep loudly instead of drifting. The boot runs `chdir`'d to the temp workspace so the api-gateway's `process.cwd()` session default, tool cwds, and fixtures agree; the `dsh web` bin's own glue (argv, profile json, AppCLIEntry) stays held by the keyless CLI smokes in `smoke-real.e2e.ts`. Setup rollback and ordinary close both dispose the Cordis tree before removing the two owned temp roots, attempt every cleanup independently, and report cleanup failures without masking the setup failure.
+`launchWebScaffold()` boots the real web composition from the shipped `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml` through the vendored Loader's include mechanism — the same tree and mechanism `AppCLIEntry` drives for `dsh web`. Divergences ride include patches over that tree, the ACP `cordis.snapshot.yml` pattern expressed in-process: temp `persistenceRoot`, `workspace-context` disabled (recorded fixtures must not embed this repo's AGENTS.md), `session-title-llm` disabled (its fire-and-forget title call would race the loop for the session's replay cursor), the webserver row pinned to port 0 with the built dist, and in keyless modes `llm-deepseek` disabled. A patch id that stops matching a row fails the boot sweep loudly instead of drifting. The boot runs `chdir`'d to the temp workspace so the api-gateway's `process.cwd()` session default, tool cwds, and fixtures agree; the `dsh web` bin's own glue (argv, profile json, AppCLIEntry) stays held by the keyless CLI smokes in `smoke-real.e2e.ts`. Setup rollback and ordinary close both dispose the Cordis tree before removing the two owned temp roots, attempt every cleanup independently, and report cleanup failures without masking the setup failure.
 
 Keyless model displacement is the disabled adapter row plus `installLlmReplay` filling the open seam on the settled root ctx in providers-catalog mode — never catch-all: with the adapter row disabled no adapter exists, so catch-all would leave `resolveModelInfo` unroutable and `compact-basic`'s post-step pressure check would warn every step instead of being provably inert (the published 128k `contextWindow` keeps it inert for small fixtures). The direct install rather than an inserted replay plugin row is deliberate: it returns the `ReplayHandle` the teardown consumption check needs. A scenario with no fixture leaves the seam empty, so a stray stream fails loud with NO_ADAPTER.
 
@@ -46,7 +46,7 @@ The lane covers three behavior families. Live-turn scenarios pin ordinary tool e
 
 ### CI stance
 
-The lane is a required compare-only gate for Linux pull requests under the [browser snapshot CI decision](2026-07-30-web-browser-snapshot-ci-gate.md). The static job publishes `apps/web/dist` with the package build artifacts; the `node 24 / snapshots and artifacts` consumer job installs the lockfile-selected Chromium, restores its OS-and-lockfile-keyed cache, and runs the lane with `DSH_SNAPSHOT=replay`. This is an intentional plane split: the host and specs use the [tsx source-launch contract](../architecture/2026-07-29-dsh-source-launch-tsx-esm.md), while the browser consumes `apps/web/dist` and package `lib/client.js` artifacts, so the gate depends on `built-package-invariants` for those client artifacts. The hosted and self-hosted default-branch Linux serial jobs run the same gate; the hosted job produces the browser cache consumed by pull requests, while the persistent self-hosted pool needs no hosted cache. CI never records or refreshes goldens. Scenarios remain POSIX-oriented and stay outside the Windows and macOS matrices.
+The lane is a required compare-only gate for Linux pull requests under the [browser snapshot CI decision](2026-07-30-web-browser-snapshot-ci-gate.md). The `node 24 / snapshots and artifacts` consumer job owns the [single Linux build](../process/2026-07-30-independent-ci-consumer-build.md), installs the lockfile-selected Chromium, restores its OS-and-lockfile-keyed cache, and runs the lane with `DSH_SNAPSHOT=replay`. This is an intentional plane split: the host and specs use the [tsx source-launch contract](../architecture/2026-07-29-dsh-source-launch-tsx-esm.md), while the browser consumes `apps/web/dist` and package `lib/client.js` artifacts, so the gate depends on `built-package-invariants` for those client artifacts. The hosted and self-hosted default-branch Linux serial jobs run the same gate; the hosted job produces the browser cache consumed by pull requests, while the persistent self-hosted pool needs no hosted cache. CI never records or refreshes goldens. Scenarios remain POSIX-oriented and stay outside the Windows and macOS matrices.
 
 ## Prior art
 
@@ -66,7 +66,7 @@ Surveyed AI-chat/agent web UIs and mocking layers (LibreChat, vercel/ai-chatbot
 
 **A committed normalized-session-log golden as a second expected surface.** Rejected: the log surface is pinned by the ACP/headless/TUI suites through the same loop and persistence; here it would double refresh cost and re-test lower tiers. Inline world-state assertions on root-context events keep the world-verification duty.
 
-**Spawning the `dsh web` bin with a `DSH_SNAPSHOT` replay branch.** Rejected: it needs a test-only replay branch plus environment plumbing in the shipped CLI. The in-process scaffold already loads the same `apps/cli/cordis.yml`; only argv, profile JSON, and `AppCLIEntry` glue remain outside it, and the keyless CLI smokes cover those paths.
+**Spawning the `dsh web` bin with a `DSH_SNAPSHOT` replay branch.** Rejected: it needs a test-only replay branch plus environment plumbing in the shipped CLI. The in-process scaffold already loads the same `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml`; only argv, profile JSON, and `AppCLIEntry` glue remain outside it, and the keyless CLI smokes cover those paths.
 
 **Changing the wire protocol for testability.** Rejected: the contract already has a first-class keyless isomorphic seam (`InProcessApiClient(toFetchHandler(api))`), the per-event unbatched SSE is exactly what makes replay observable in a browser, and testing a wire we no longer ship would invert the tier's purpose.
 

.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.zh.md

@@ -16,7 +16,7 @@ Web GUI 以一条真实组装链交付——chromium 页面 → client 插件 bu
 
 一个普通的共享 fixture 模块（[测试政策认可的形态](../../../../docs/testing.md)），不是包：值得门禁把守的逻辑——回放推导、会话解析、日志脱敏、持久化——都在已受门禁的包 `dsh-llm-replay`、`dsh-acp-snapshot`、`dsh-session-persistence-jsonl` 中；剩下的只是启动接线和浏览器胶水，而驱动 chromium 的源码在无浏览器的覆盖率 runner 上无法诚实保持逐文件 100% 覆盖率。
 
-`launchWebScaffold()` 通过 vendored Loader 的 include 机制，从交付的 `apps/cli/cordis.yml` 启动真实 web 组合——与 `AppCLIEntry` 为 `dsh web` 驱动的是同一棵树、同一套机制。差异全部经 include patch 覆盖在这棵树上，即 ACP `cordis.snapshot.yml` 模式的进程内表达：临时 `persistenceRoot`；禁用 `workspace-context`（录制的 fixture 不得嵌入本仓库的 AGENTS.md）；禁用 `session-title-llm`（其发后不管的标题调用会与循环争抢会话的回放游标）；webserver 行钉到端口 0 加已构建 dist；无密钥模式下禁用 `llm-deepseek`。patch 的 id 一旦不再匹配任何行，boot 扫描会大声失败而不是漂移。boot 在临时工作区 `chdir` 下运行，使 api-gateway 的 `process.cwd()` 会话默认值、工具 cwd 与 fixture 一致；`dsh web` bin 自身的胶水（argv、profile json、AppCLIEntry）仍由 `smoke-real.e2e.ts` 中的无密钥 CLI 冒烟把守。初始化回滚和正常关闭都会先对 Cordis 树执行 dispose（资源释放），再删除 scaffold 持有的两个临时根目录；每项清理都会独立尝试，并会报告清理失败而不掩盖初始化失败。
+`launchWebScaffold()` 通过 vendored Loader 的 include 机制，从交付的 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 启动真实 web 组合——与 `AppCLIEntry` 为 `dsh web` 驱动的是同一棵树、同一套机制。差异全部经 include patch 覆盖在这棵树上，即 ACP `cordis.snapshot.yml` 模式的进程内表达：临时 `persistenceRoot`；禁用 `workspace-context`（录制的 fixture 不得嵌入本仓库的 AGENTS.md）；禁用 `session-title-llm`（其发后不管的标题调用会与循环争抢会话的回放游标）；webserver 行钉到端口 0 加已构建 dist；无密钥模式下禁用 `llm-deepseek`。patch 的 id 一旦不再匹配任何行，boot 扫描会大声失败而不是漂移。boot 在临时工作区 `chdir` 下运行，使 api-gateway 的 `process.cwd()` 会话默认值、工具 cwd 与 fixture 一致；`dsh web` bin 自身的胶水（argv、profile json、AppCLIEntry）仍由 `smoke-real.e2e.ts` 中的无密钥 CLI 冒烟把守。初始化回滚和正常关闭都会先对 Cordis 树执行 dispose（资源释放），再删除 scaffold 持有的两个临时根目录；每项清理都会独立尝试，并会报告清理失败而不掩盖初始化失败。
 
 无密钥的模型替换 = 禁用适配器行的 patch 加 `installLlmReplay` 在停稳的根 ctx 上以提供方目录（providers-catalog）模式填充开放的 seam——绝不用 catch-all：适配器行被禁用后不存在任何适配器，catch-all 会让 `resolveModelInfo` 无路由可走，`compact-basic` 的步后压力检查将步步告警，而不是被可证明地闲置（发布的 128k `contextWindow` 使该路径对小 fixture 保持闲置）。选择直接安装而非插入回放插件行是刻意的：直接安装返回收尾消费检查所需的 `ReplayHandle`。没有 fixture 的场景让 seam 保持空置，任何离群的流式调用都会以 NO_ADAPTER 大声失败。
 
@@ -46,7 +46,7 @@ Web GUI 以一条真实组装链交付——chromium 页面 → client 插件 bu
 
 ### CI 立场
 
-根据[浏览器快照 CI 决策](2026-07-30-web-browser-snapshot-ci-gate.md)，该车道是 Linux 拉取请求必需的只比较门禁。static 任务会把 `apps/web/dist` 与包构建产物一同发布；`node 24 / snapshots and artifacts` 消费方任务安装锁文件选定的 Chromium，恢复以操作系统和锁文件为键的缓存，并用 `DSH_SNAPSHOT=replay` 运行该车道。这是有意的平面切分：host 与 spec 使用 [tsx 源码启动契约](../architecture/2026-07-29-dsh-source-launch-tsx-esm.md)，浏览器则消费 `apps/web/dist` 和包的 `lib/client.js` 产物，因此门禁依赖 `built-package-invariants` 提供这些客户端产物。托管和自托管的默认分支 Linux 串行任务运行同一门禁；托管任务生成供 PR 消费的浏览器缓存，持久化自托管池则不需要托管侧缓存。CI 从不录制或刷新预期输出。场景仍面向 POSIX，并继续置于 Windows 和 macOS 矩阵之外。
+根据[浏览器快照 CI 决策](2026-07-30-web-browser-snapshot-ci-gate.md)，该车道是 Linux 拉取请求必需的只比较门禁。`node 24 / snapshots and artifacts` 消费方任务在[消费方独立构建](../process/2026-07-30-independent-ci-consumer-build.md)中负责唯一一次 Linux 构建，安装锁文件选定的 Chromium，恢复以操作系统和锁文件为键的缓存，并用 `DSH_SNAPSHOT=replay` 运行该车道。这是有意的平面切分：host 与 spec 使用 [tsx 源码启动契约](../architecture/2026-07-29-dsh-source-launch-tsx-esm.md)，浏览器则消费 `apps/web/dist` 和包的 `lib/client.js` 产物，因此门禁依赖 `built-package-invariants` 提供这些客户端产物。托管和自托管的默认分支 Linux 串行任务运行同一门禁；托管任务生成供 PR 消费的浏览器缓存，持久化自托管池则不需要托管侧缓存。CI 从不录制或刷新预期输出。场景仍面向 POSIX，并继续置于 Windows 和 macOS 矩阵之外。
 
 ## 业界先例
 
@@ -66,7 +66,7 @@ Web GUI 以一条真实组装链交付——chromium 页面 → client 插件 bu
 
 **第二份提交的规范化会话日志预期输出。** 已否决：日志表面已由 ACP/headless/TUI 套件经同一循环与持久化钉住；在此只会翻倍刷新成本并重复测试下层。内联在根上下文事件上的世界状态断言保住了验证世界的义务。
 
-**以 `DSH_SNAPSHOT` 回放分支拉起 `dsh web` bin。** 已否决：它需要在交付的 CLI 中增加测试专用回放分支和环境变量管道。进程内 scaffold 已加载同一份 `apps/cli/cordis.yml`；只剩 argv、profile JSON 和 `AppCLIEntry` 胶水不在其覆盖范围内，而这些路径已由无密钥 CLI 冒烟覆盖。
+**以 `DSH_SNAPSHOT` 回放分支拉起 `dsh web` bin。** 已否决：它需要在交付的 CLI 中增加测试专用回放分支和环境变量管道。进程内 scaffold 已加载同一份 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml`；只剩 argv、profile JSON 和 `AppCLIEntry` 胶水不在其覆盖范围内，而这些路径已由无密钥 CLI 冒烟覆盖。
 
 **为可测试性改 wire 协议。** 已否决：契约已有第一等的无密钥同构 seam（`InProcessApiClient(toFetchHandler(api))`），逐事件不合批的 SSE 恰是回放在浏览器中可观测的原因，测试一条不再交付的 wire 会颠倒该层的存在意义。
 

.agents/notes/implemented/testing/2026-07-30-web-browser-snapshot-ci-gate.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-07-30-web-browser-snapshot-ci-gate.md
-2026-07-30-web-browser-snapshot-ci-gate.md: 3f87bb0f3d936bcee7ba7c3d84ae808c6ede1a97
-2026-07-30-web-browser-snapshot-ci-gate.zh.md: af563f0e2a1c20f7b53d371e97e41b7ffa52a1d1
+2026-07-30-web-browser-snapshot-ci-gate.md: 14402485034cd85ec5781477ce67481165d47e62
+2026-07-30-web-browser-snapshot-ci-gate.zh.md: f214c524253d2ad8a43e8543dc65ddfcfd7c065c

.agents/notes/implemented/testing/2026-07-30-web-browser-snapshot-ci-gate.md

@@ -12,7 +12,7 @@ The [keyless web browser e2e lane](2026-07-24-web-gui-browser-e2e-lane.md) runs
 
 For Linux PRs, the `node 24 / snapshots and artifacts` job must run the full web browser replay/compare suite. `scripts/run-gates.ts` registers `test:web:built` as a `ci-consumers` gate and explicitly injects `DSH_SNAPSHOT=replay`; CI never runs in `record` or `refresh` mode, so when the committed goldens disagree with the currently assembled application, the tests fail directly instead of silently rewriting them on the runner and then passing.
 
-The static CI job already builds all publishable artifacts; it puts `apps/web/dist` and the package `lib/` directories in the built-tree artifact, which the consumer job reuses without rebuilding the entire repository. On hosted runners, CI installs Chromium and its system dependencies at the Playwright version in the lockfile. On the persistent failover VM, the image owns the Linux system packages and CI installs only Chromium, avoiding per-run `apt` mutation. The hosted default-branch Linux serial job runs the suite and produces the operating-system-and-lockfile-keyed browser cache; pull requests restore it without paying compression and upload on the required path, with an operating-system prefix fallback across lockfile changes. The self-hosted standby runs the same comparison without hosted cache actions.
+The consumer job owns the [single Linux build](../process/2026-07-30-independent-ci-consumer-build.md), so `apps/web/dist` and the package `lib/` directories remain in its workspace for the browser suite. On hosted runners, CI installs Chromium and its system dependencies at the Playwright version in the lockfile. On the persistent failover VM, the image owns the Linux system packages and CI installs only Chromium, avoiding per-run `apt` mutation. The hosted default-branch Linux serial job runs the suite and produces the operating-system-and-lockfile-keyed browser cache; pull requests restore it without paying compression and upload on the required path, with an operating-system prefix fallback across lockfile changes. The self-hosted standby runs the same comparison without hosted cache actions.
 
 Local `pnpm run test:web` continues to build first and then run the full browser suite; `test:web:built` is the entry point for existing build artifacts. Developers explicitly run `DSH_SNAPSHOT=refresh pnpm run test:web` only after confirming that user-visible output changed intentionally, review every expected-output diff, and then verify again in replay mode that no files are written.
 
@@ -26,10 +26,10 @@ An observed self-hosted consumer run measured `web-snapshot` at 112.15 seconds a
 
 **Run CI in `refresh` mode and then check the working tree.** Rejected: checking after writing turns the assertion mechanism into a generator; if the working-tree check is wired incorrectly, it can turn a regression into a passing expected-output update. Replay compares the existing goldens directly and has a smaller failure surface.
 
-**Create a standalone browser job and rebuild the entire repository.** Rejected: it would duplicate dependency installation and the publishable build. The existing Linux consumer job already consumes the same built-tree artifact and is part of the unified required verdict.
+**Create a standalone browser job and rebuild the entire repository.** Rejected: it would duplicate dependency installation and the publishable build. The existing Linux consumer job already owns that build and is part of the unified required verdict.
 
 **Replace real Chromium with jsdom snapshots.** Rejected: jsdom does not cover the browser, HTTP/SSE carriage, or the composition of real client plugin bundles. It remains useful for fast lower-layer feedback, but cannot replace the assembled browser chain.
 
 ## Consequences
 
-Before merge, every PR proves that the current web assembly matches all committed browser expected outputs, turning a missed refresh from an “unrelated change in a later PR” into a failure in the PR that introduced it. The cost is Chromium provisioning and one serial pass through the browser scenarios in the consumer job; built-artifact reuse and the browser cache avoid duplicate builds and downloads on reruns. The gate still makes no claim of cross-platform browser consistency, and if a Playwright/Chromium upgrade changes the ARIA format, the upgrade PR must explicitly refresh the expected outputs and review the churn.
+Before merge, every PR proves that the current web assembly matches all committed browser expected outputs, turning a missed refresh from an “unrelated change in a later PR” into a failure in the PR that introduced it. The cost is Chromium provisioning and one serial pass through the browser scenarios in the consumer job; the consumer-owned build and browser cache avoid duplicate builds and downloads on reruns. The gate still makes no claim of cross-platform browser consistency, and if a Playwright/Chromium upgrade changes the ARIA format, the upgrade PR must explicitly refresh the expected outputs and review the churn.

.agents/notes/implemented/testing/2026-07-30-web-browser-snapshot-ci-gate.zh.md

@@ -12,7 +12,7 @@ Status: implemented
 
 Linux PR 的 `node 24 / snapshots and artifacts` 必须运行完整 Web 浏览器 replay/compare。`scripts/run-gates.ts` 把 `test:web:built` 作为 `ci-consumers` 的一个 gate，并显式注入 `DSH_SNAPSHOT=replay`；CI 永不以 `record` 或 `refresh` 模式运行，因此提交的 golden 与当前组装应用不一致时测试直接失败，不会在 runner 内静默改写后通过。
 
-静态 CI job 已经构建全部发布产物；它把 `apps/web/dist` 和包的 `lib/` 目录放进 built-tree 产物，消费方 job 复用该产物而不重复全仓构建。在托管运行器上，CI 按锁文件中的 Playwright 版本安装 Chromium 及其系统依赖。在持久化故障切换 VM 上，镜像负责预装 Linux 系统软件包，CI 只安装 Chromium，避免每次运行都通过 `apt` 改动系统。托管的默认分支 Linux 串行 job 运行该套件，并生成以操作系统和锁文件为键的浏览器缓存；PR 恢复该缓存，使必需路径无需承担压缩和上传开销，并可在锁文件变化时按操作系统前缀回退。自托管热备运行相同的比较，但不执行托管缓存操作。
+消费方 job 在[消费方独立构建](../process/2026-07-30-independent-ci-consumer-build.md)中负责唯一一次 Linux 构建，因此 `apps/web/dist` 和包的 `lib/` 目录会保留在其工作区中，供浏览器套件使用。在托管运行器上，CI 按锁文件中的 Playwright 版本安装 Chromium 及其系统依赖。在持久化故障切换 VM 上，镜像负责预装 Linux 系统软件包，CI 只安装 Chromium，避免每次运行都通过 `apt` 改动系统。托管的默认分支 Linux 串行 job 运行该套件，并生成以操作系统和锁文件为键的浏览器缓存；PR 恢复该缓存，使必需路径无需承担压缩和上传开销，并可在锁文件变化时按操作系统前缀回退。自托管热备运行相同的比较，但不执行托管缓存操作。
 
 本地 `pnpm run test:web` 仍先构建再运行浏览器全集；`test:web:built` 是已有构建产物的执行入口。开发者只在确认用户可见输出有意变化后显式运行 `DSH_SNAPSHOT=refresh pnpm run test:web`，评审每一处 expected diff，再以 replay 模式复验不再写文件。
 
@@ -26,10 +26,10 @@ Linux PR 的 `node 24 / snapshots and artifacts` 必须运行完整 Web 浏览
 
 **让 CI 以 `refresh` 模式运行后检查工作树。** 已否决：写后比较把断言机制变成生成器，若工作树检查接线失效就会把回归更新成绿色；replay 直接比较已有 golden，失败面更小。
 
-**新建独立 browser job 并重新构建全仓。** 已否决：它会重复依赖安装和发布构建。现有 Linux consumer job 已消费同一 built-tree artifact，并已被统一的 required verdict 聚合。
+**新建独立 browser job 并重新构建全仓。** 已否决：它会重复依赖安装和发布构建。现有 Linux 消费方 job 已负责该构建，并已被统一的 required verdict 聚合。
 
 **用 jsdom 快照代替真实 Chromium。** 已否决：jsdom 不覆盖浏览器、HTTP/SSE 承载及真实 client plugin bundle 组合；它保留为快速的下层反馈，不能替代 assembled browser chain。
 
 ## 后果
 
-每个 PR 都在合并前证明当前 Web 组装与所有已提交的浏览器 expected 一致，漏刷从“后续 PR 的无关变化”变成引入 PR 自己的失败。成本是消费方 job 需要供给 Chromium，并串行运行一轮浏览器场景；built artifact 复用与浏览器缓存避免重跑时重复构建和下载。门禁仍不声称跨平台浏览器一致性，Playwright/Chromium 升级若改变 aria 格式，升级 PR 必须显式 refresh 并评审 churn。
+每个 PR 都在合并前证明当前 Web 组装与所有已提交的浏览器 expected 一致，漏刷从“后续 PR 的无关变化”变成引入 PR 自己的失败。成本是消费方 job 需要供给 Chromium，并串行运行一轮浏览器场景；消费方独立构建与浏览器缓存避免重跑时重复构建和下载。门禁仍不声称跨平台浏览器一致性，Playwright/Chromium 升级若改变 aria 格式，升级 PR 必须显式 refresh 并评审 churn。

.agents/notes/proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.md
-2026-07-28-storage-root-and-derived-medium-recovery.md: 06fa98b10dc5ac3164d8905e7005a42d9e99ae92
-2026-07-28-storage-root-and-derived-medium-recovery.zh.md: b7bd18ffdbfaf412d9a91940cf1770e273f5b847
+2026-07-28-storage-root-and-derived-medium-recovery.md: 3937b17502d0bf640625e73822b758b1862b5391
+2026-07-28-storage-root-and-derived-medium-recovery.zh.md: ff2f196a911986d23b7a102e6336f353c6aef278

.agents/notes/proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.md

@@ -8,7 +8,7 @@ English | [中文](2026-07-28-storage-root-and-derived-medium-recovery.zh.md)
 
 The persisted projection cache ([RFC](2026-07-27-session-projection-and-command-log.md), shipped as `dsh-session-projection-cache`) surfaced two gaps in the storage substrate it landed on. Both are properties of the domain-KV stack ([design](2026-07-24-domain-kv-storage-and-workspace.md)), not of the cache itself, and both bite the cache first because it is the first *derived* medium on that stack.
 
-**Where the files actually live.** The shipped composition gives the json backend a relative root — `root: './.storages'` (apps/cli/cordis.yml) — and `AppCLIEntry.composePatches` patches only the session store's root to the global harness home (`$DSH_HOME/sessions`, default `~/.dsh/sessions`, profile-overridable via `persistenceRoot`); no equivalent patch or profile key exists for `storage-json`. `JsonStorageBackend` never resolves its root either — each unit open joins the still-relative path against whatever `process.cwd()` is at that moment (packages/storage/storage-json/src/index.ts) — the exact hazard the JSONL session backend resolves-once to prevent ("later process.cwd() changes cannot split one backend across roots", packages/session-persistence/session-persistence-jsonl/src/index.ts). Net effect: session logs are global across launch directories, but `workspace.json` and `session_projcache.json` land under `<launch dir>/.storages/`. Two launches from different directories share their sessions yet see different workspace registries and different projection caches — and the cache exists precisely to serve the cross-session cold listing, which now misses for every session last cached under another launch directory.
+**Where the files actually live.** The shipped Web overlay gives the json backend a relative root — `root: './.storages'` (`apps/cli/config/web.cordis.yml`) — while the shared base defaults the session store to the global harness home (`$DSH_HOME/sessions`, default `~/.dsh/sessions`); no equivalent global root exists for `storage-json`. `JsonStorageBackend` never resolves its root either — each unit open joins the still-relative path against whatever `process.cwd()` is at that moment (packages/storage/storage-json/src/index.ts) — the exact hazard the JSONL session backend resolves-once to prevent ("later process.cwd() changes cannot split one backend across roots", packages/session-persistence/session-persistence-jsonl/src/index.ts). Net effect: session logs are global across launch directories, but `workspace.json` and `session_projcache.json` land under `<launch dir>/.storages/`. Two launches from different directories share their sessions yet see different workspace registries and different projection caches — and the cache exists precisely to serve the cross-session cold listing, which now misses for every session last cached under another launch directory.
 
 **How recovery works today.** Inside a healthy medium the cache is fully self-healing by design: a `stateVersion`-mismatched row is discarded and refolded, a log shrunk below a row's watermark is detected by the anchored restore floor and answered with one full re-read, and every background write is fail-soft. But at the *medium* level there is no recovery at all: a truncated, hand-edited, or version-bumped `session_projcache.json` fails `openJsonUnit` with `malformed-medium`/`version-mismatch` (packages/storage/storage-json/src/format.ts), a schema-drifted record fails domain open with `invalid-record` (packages/storage/storage-domain/src/index.ts), the rejection propagates through `SessionProjectionCache[Service.init]`, and under the CLI's fail-loud boot the assembly refuses to start. A file whose entire content is rebuildable from session logs can brick boot. This contradicts the cache package's own stated stance ("a stale or unreadable cache costs a longer tail replay, never a wrong value") and the cache domain spec's JSDoc ("version bumps discard the whole medium"), which today describes an aspiration, not the implementation. The same fail-loud path is *correct* for `workspace.json` — workspace records are authoritative, not derivable — so the missing concept is a per-domain declaration of authority, not a global behavior change.
 

.agents/notes/proposed/architecture/2026-07-28-storage-root-and-derived-medium-recovery.zh.md

@@ -8,7 +8,7 @@ Status: proposed
 
 持久投影缓存（[RFC](2026-07-27-session-projection-and-command-log.md)，已作为 `dsh-session-projection-cache` 落地）暴露了它所依托的存储基座的两个缺口。二者都是 domain-KV 栈（[设计](2026-07-24-domain-kv-storage-and-workspace.md)）的属性而非缓存自身的问题，且都首先咬到缓存——因为它是这条栈上第一个*派生*介质。
 
-**文件到底存在哪。** 出厂组合给 json 后端的是相对根目录——`root: './.storages'`（apps/cli/cordis.yml）——而 `AppCLIEntry.composePatches` 只把会话存储的根 patch 到全局 harness home（`$DSH_HOME/sessions`，默认 `~/.dsh/sessions`，可经 profile 键 `persistenceRoot` 覆盖）；`storage-json` 没有对应的 patch 也没有 profile 键。`JsonStorageBackend` 自己也从不 resolve 根——每次打开 unit 都把仍然相对的路径 join 到当时的 `process.cwd()` 上（packages/storage/storage-json/src/index.ts）——这正是 JSONL 会话后端用「构造时 resolve 一次」防住的那个隐患（"later process.cwd() changes cannot split one backend across roots"，packages/session-persistence/session-persistence-jsonl/src/index.ts）。净效果：会话日志跨启动目录全局共享，但 `workspace.json` 和 `session_projcache.json` 落在 `<启动目录>/.storages/` 下。从两个不同目录启动，会话相同，工作区注册表和投影缓存却各是一份——而缓存存在的意义恰恰是跨会话冷列表，如今凡是上次在别的启动目录下缓存过的会话全部 miss。
+**文件到底存在哪。** 出厂 Web overlay 给 json 后端的是相对根目录——`root: './.storages'`（`apps/cli/config/web.cordis.yml`）——而共享 base 将会话存储默认为全局 harness home（`$DSH_HOME/sessions`，默认 `~/.dsh/sessions`）；`storage-json` 没有对应的全局根目录。`JsonStorageBackend` 自己也从不 resolve 根——每次打开 unit 都把仍然相对的路径 join 到当时的 `process.cwd()` 上（packages/storage/storage-json/src/index.ts）——这正是 JSONL 会话后端用「构造时 resolve 一次」防住的那个隐患（"later process.cwd() changes cannot split one backend across roots"，packages/session-persistence/session-persistence-jsonl/src/index.ts）。净效果：会话日志跨启动目录全局共享，但 `workspace.json` 和 `session_projcache.json` 落在 `<启动目录>/.storages/` 下。从两个不同目录启动，会话相同，工作区注册表和投影缓存却各是一份——而缓存存在的意义恰恰是跨会话冷列表，如今凡是上次在别的启动目录下缓存过的会话全部 miss。
 
 **现在是怎么恢复的。** 在健康介质内部，缓存按设计完全自愈：`stateVersion` 不匹配的行被丢弃重折，日志缩短到行水位以下由带锚的 restore floor 检出并以一次全量重读回答，每次后台写都是 fail-soft。但在*介质*层面完全没有恢复：被截断、被手改或版本被 bump 的 `session_projcache.json` 会让 `openJsonUnit` 以 `malformed-medium`/`version-mismatch` 失败（packages/storage/storage-json/src/format.ts），schema 漂移的记录让域 open 以 `invalid-record` 失败（packages/storage/storage-domain/src/index.ts），拒绝一路穿过 `SessionProjectionCache[Service.init]`，在 CLI 的 fail-loud 启动下整个组装拒绝启动。一个内容完全可从会话日志重建的文件能把启动搞死。这与缓存包自己声明的立场（"a stale or unreadable cache costs a longer tail replay, never a wrong value"）和缓存域 spec 的 JSDoc（"version bumps discard the whole medium"）相矛盾——后者今天描述的是愿望而非实现。同一条 fail-loud 路径对 `workspace.json` 却是*正确*的——工作区记录是权威数据，不可派生——所以缺的概念是按域声明权威性，而不是全局改行为。
 

.github/workflows/ci.yml

@@ -27,9 +27,12 @@ env:
 
 jobs:
 
+  # FIXME: Re-enable the three hosted serial reference jobs before cutting a release.
+  # The self-hosted standby remains active on every master push.
+
   # Three enterprise jobs isolate coverage, static analysis, and the
-  # build-backed consumer tail. The static job publishes its exact build so
-  # consumers do not repeat the longest part of their critical path.
+  # build-backed consumer tail. The consumer job owns the only Linux build so
+  # all three jobs enter runner allocation independently.
   #
   # FAILOVER: each Linux enterprise job resolves its pool through the
   # DSH_CI_FAILOVER repository variable. Unset (normal), the expressions
@@ -93,19 +96,6 @@ jobs:
           DSH_ARCHIVE_BASE_REF: ${{ github.event.pull_request.base.sha }}
         run: pnpm run check:ci:static
 
-      - name: Pack built tree
-        run: >-
-          tar -czf "$RUNNER_TEMP/node-24-built-tree.tar.gz"
-          apps/*/lib apps/web/dist packages/*/*/lib vendor/*/lib
-
-      - uses: actions/upload-artifact@v7
-        with:
-          name: node-24-built-tree
-          path: ${{ runner.temp }}/node-24-built-tree.tar.gz
-          if-no-files-found: error
-          retention-days: 1
-          compression-level: 0
-
   node-24-coverage:
     if: github.event_name == 'pull_request'
     runs-on: >-
@@ -120,8 +110,9 @@ jobs:
       # across six always-on runner instances, and the timing-sensitive
       # process suites have documented aggregate-contention failures.
       # 8 × 6 instances = 48 workers worst case on 64 cores.
-      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && github.event.pull_request.user.login != 'dependabot[bot]' && '8' || '24' }}
-      DSH_GATE_CONCURRENCY: '8'
+      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && github.event.pull_request.user.login != 'dependabot[bot]' && '8' || '8' }}
+      DSH_GATE_CONCURRENCY: '3'
+      NODE_OPTIONS: '--max-old-space-size=8192'
     steps:
       - uses: actions/checkout@v6
         with:
@@ -171,7 +162,6 @@ jobs:
         run: pnpm run check:ci:coverage
 
   node-24-consumers:
-    needs: node-24
     if: github.event_name == 'pull_request'
     runs-on: >-
       ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
@@ -191,14 +181,6 @@ jobs:
         with:
           persist-credentials: false
 
-      - uses: actions/download-artifact@v8
-        with:
-          name: node-24-built-tree
-          path: ${{ runner.temp }}
-
-      - name: Restore built tree
-        run: tar -xzf "$RUNNER_TEMP/node-24-built-tree.tar.gz"
-
       - uses: pnpm/action-setup@v4
         with:
           dest: ${{ runner.temp }}/setup-pnpm
@@ -224,8 +206,8 @@ jobs:
           restore-keys: |
             ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-
 
-      # Pull requests restore the cache produced by serial-linux on master;
-      # they do not pay compression and upload on the required path.
+      # Pull requests restore the cache normally produced by serial-linux on
+      # master; they do not pay compression and upload on the required path.
       - uses: actions/cache/restore@v4
         if: vars.DSH_CI_FAILOVER != 'selfhosted' || github.event.pull_request.user.login == 'dependabot[bot]'
         with:
@@ -320,8 +302,9 @@ jobs:
   # The required pull-request Windows signal: the two blocking win32 surfaces
   # (workspace build, production site) execute with real, checksum-verified
   # Windows Node under Wine on standard hosted Linux. The master
-  # serial-windows job below keeps the complete native-kernel inventory —
-  # including the observational portability gates this lane does not run —
+  # serial-windows job below normally keeps the complete native-kernel
+  # inventory — including the observational portability gates this lane does
+  # not run —
   # on real windows-2025. This job only provisions runner state (caches,
   # apt); scripts/wine-windows-gates.sh owns the gate logic and is the same
   # script the optional local gate `pnpm run check:windows-wine` runs.
@@ -426,12 +409,12 @@ jobs:
           cp /var/cache/apt/archives/*.deb "$HOME/wine-debs/"
           du -sh "$HOME/wine-debs"
 
-  # Master pushes run only the serial reference jobs below.
-  # Each host executes the complete, unsharded primary Node aggregate with one
-  # gate worker, giving reviewers a simple cross-platform oracle for completeness
-  # and timing.
+  # The hosted reference jobs below are temporarily disabled; the self-hosted
+  # standby remains active. Each enabled host executes the complete, unsharded
+  # primary Node aggregate with one gate worker, giving reviewers a simple
+  # cross-platform oracle for completeness and timing.
   serial-linux:
-    if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+    if: false
     name: serial / linux
     runs-on: ubuntu-latest
     steps:
@@ -556,7 +539,7 @@ jobs:
         run: pnpm run check:ci:linux-primary
 
   serial-macos:
-    if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+    if: false
     name: serial / macos
     runs-on: macos-latest
     steps:
@@ -583,7 +566,7 @@ jobs:
         run: pnpm run check:ci
 
   serial-windows:
-    if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+    if: false
     name: serial / windows
     runs-on: windows-2025
     steps:

AGENTS.md

@@ -31,6 +31,7 @@ packages/    @deepseek-ai/dsh-<pkg> workspaces at packages/<group>/<pkg>/
   cordis/      self-referential toolset: the agent inspects/mounts plugins in its own runtime
   hooks/       Claude Code/Codex hook bridges + shared wire-protocol library
   session-persistence/  persistence seam + JSONL/SQLite backends
+  settings/    user-settings seam + file-backed provider
   acp/         automation-only Agent Client Protocol server
   ui/          TUI/JSON-RPC bridges; boot, approval, interaction plugins
   examples/    demo bundles (agent-spine + TUI/CLI/ACP/JSON-RPC bins) leaves load

apps/cli/README.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write apps/cli/README.md
-README.md: 93c36d18abd06bbd7a80c918f520b92489180395
-README.zh.md: 85f4624a592eaf2ae44dc31fb4e18fb5657e62fd
+README.md: 2bc36cce6205a4bfc3ba1d7ee15f0e0b2feab215
+README.zh.md: 0e0771658cecb4bee0f3eadd0639ad531e64ea3c

apps/cli/README.md

@@ -2,19 +2,23 @@
 
 English | [中文](README.zh.md)
 
-The `dsh` command-line entry follows the `apps/` assembly tier: `apps/*` are product assemblies over `packages/*` libraries. Plain `dsh` boots the interactive TUI coding agent, `dsh -p "task"` runs one headless turn, and `dsh web` serves the browser UI.
 
-Argv is parsed once through a [Commander](https://github.com/tj/commander.js) adapter ([`src/args.ts`](src/args.ts)): one program whose default (no subcommand) is the TUI/headless surface (`--config`, `-p`/`--prompt`, `--resume`) and whose `web` subcommand is the browser UI. `src/bin.ts` switches on the resolved mode and dynamic-imports only that mode's module. `dsh --help` lists every mode and `dsh web --help` renders the web usage, `dsh --version` prints this app's version, and an unknown option or a mistyped `--resume` fails loud (stderr, exit 1) instead of misrouting. `dsh web`'s `--host`/`--port` are unvalidated pass-through overrides: the `dsh-host-webserver` schema is the single source of both the default (the shipped `cordis.yml` value when a flag is absent) and validity, and rejects a bad value at boot. `--trusted-host` appends named authorities for the /api browser-trust fence; an all-interfaces bind additionally derives the machine's LAN IP literals itself ([`src/app-cli-entry.ts`](src/app-cli-entry.ts)), so the printed LAN URL works without flags.
+Argv is parsed once through a [Commander](https://github.com/tj/commander.js) adapter ([`src/args.ts`](src/args.ts)): one program whose default (no subcommand) is the TUI/headless surface (`--config`, `-p`/`--prompt`, `--resume`), whose `meta` subcommand is the same TUI over this checkout, whose `upgrade` subcommand is an option-less guided-session entry, and whose `web` subcommand is the browser UI. `src/bin.ts` switches on the resolved mode and dynamic-imports only that mode's module. `dsh --help` lists every mode and `dsh web --help` renders the web usage, `dsh --version` prints this app's version, and an unknown option or a mistyped `--resume` fails loud (stderr, exit 1) instead of misrouting. Every subcommand that shares no option with the default surface — `upgrade`, `web`, `meta` — rejects a leaked `--config`/`-p`/`--resume` rather than running and dropping it. `dsh web`'s `--host`/`--port` are unvalidated pass-through overrides: the `dsh-host-webserver` schema is the single source of both the default (the shipped Web overlay value when a flag is absent) and validity, and rejects a bad value at boot. `--trusted-host` appends named authorities for the /api browser-trust fence; an all-interfaces bind additionally derives the machine's LAN IP literals itself ([`src/app-cli-entry.ts`](src/app-cli-entry.ts)), so the printed LAN URL works without flags.
 
 The TUI surface:
 
-- boots the shipped default config (`examples/tui-agent/cordis.yml`), or the tree named by `--config <path>` (the demo/test escape for booting an alternate example tree), through [`dsh-app-boot`](../../packages/ui/app-boot/README.md);
-- resumes a persisted session with `dsh --resume <session-id>` and, when the Node host exposes `process.execve`, supplies the TUI's in-place handoff host: after selector preflight and current-session flush, the host disposes the app and replaces the process with a normalized `dsh --resume <id>`; runtimes without process replacement keep the displayed command fallback. The flag provides the id on the boot context under `RESUME_SESSION_ID_KEY` (no environment variable), which the shipped config reads through `!!js`, and a missing or unreadable id fails loud instead of creating a fresh session;
-- treats the **invoking directory** as the workspace — sessions, relative paths, and workspace instructions resolve from the cwd;
+- boots `base.cordis.yml` plus `tui.cordis.yml` through [`dsh-app-boot`](../../packages/ui/app-boot/README.md); `--config <path>` applies a patch-list overlay instead of the personal overlay, while `--config-replace <path>` boots that file as the complete tree;
+- resumes a persisted session with `dsh --resume <session-id>` and, when the Node host exposes `process.execve`, supplies the TUI's in-place handoff host: after selector preflight and current-session flush, the host disposes the app and replaces the process with a normalized resume invocation; runtimes without process replacement leave the session running and say so. This CLI owns session identity and the exit line rather than the config: it mints or selects the `main` session id and provides it, plus the exact command that reproduces this invocation, on the boot context ([`MAIN_SESSION_ID_KEY`](../../packages/ui/tui/README.md) and `TUI_GOODBYE_MESSAGE_KEY`). No `cordis.yml` key can drop resume, and a missing or unreadable id fails loud instead of creating a fresh session;
+- treats the **invoking directory** as the workspace — sessions, relative paths, and workspace instructions resolve from the cwd (`dsh meta` is the sole exception, below);
 - tells the agent where its own source lives: after boot it adds a prompt section naming this harness checkout, resolved from the launcher's real path so it holds under a PATH symlink and an arbitrary cwd, so the self-referential `cordis` toolset can read and modify it;
 - applies the personal overlay from `~/.dsh` (see [app-boot's Personal config](../../packages/ui/app-boot/README.md#personal-config)): `.env` fills environment gaps (ambient > project `.env` > personal `.env`), `config.yaml` patches the booted tree.
 
-The Web and headless surfaces boot one shared composition (`cordis.yml`): both treat the invoking directory as the default project and Workspace root, create named Workspaces beneath that root unless `--workspace-root <path>` overrides it, load applicable `AGENTS.md`/`CLAUDE.md` instructions into each agent-loop request prefix with a 65,536-byte render budget, and opt into first-message model titles. Headless differs only in listening on an OS-assigned port (parallel `dsh -p` runs never collide; the stderr-printed URL opens the live session in a browser). Both need the frontend dist and client bundles built (`pnpm run build && pnpm run build:web`).
+`dsh meta` is that same TUI with this harness checkout as the workspace, so working on dsh itself needs no `cd`. It chdirs to the checkout root — resolved from the launcher's real path, the same root the source-path prompt section names — after both `.env` layers are loaded, so environment precedence is unchanged while the session cwd and HMR watch root move together. Meta always starts a fresh session and accepts no default-surface options; use ordinary `dsh --resume <id>` to resume a persisted session.
+
+`dsh upgrade` is a guided fresh-session entry over the default TUI surface: it mints a fresh session in the invoking directory and seeds its first turn with the bundled `dsh-upgrade` skill, exactly as if the user typed `/skill:<name>`. The launcher passes the skill name on the boot context ([`INITIAL_SKILL_KEY`](../../packages/ui/tui/README.md)), which the TUI auto-invokes once the chat is live. Both take no options — `--config`, `-p`, and `--resume` fail loud — and seed only on this first launch, so a later `dsh --resume <id>` of the session is an ordinary TUI session with no re-injection.
+
+
+The Web and headless surfaces boot `base.cordis.yml` plus `web.cordis.yml`, then apply `$DSH_HOME/config.yaml`; an explicit `--config <path>` replaces that personal overlay. Both surfaces otherwise share the same composition: both treat the invoking directory as the default project and Workspace root, create named Workspaces beneath that root unless `--workspace-root <path>` overrides it, load applicable `AGENTS.md`/`CLAUDE.md` instructions into each agent-loop request prefix with a 65,536-byte render budget, and opt into first-message model titles. Headless differs only in listening on an OS-assigned port (parallel `dsh -p` runs never collide; the stderr-printed URL opens the live session in a browser). Both need the frontend dist and client bundles built (`pnpm run build && pnpm run build:web`).
 
 The shipped TUI and Web compositions register the native DeepSeek adapter plus pi-ai OpenAI and Anthropic profiles. Credentials and endpoint overrides come from the provider-standard `DEEPSEEK_API_KEY` / `DEEPSEEK_BASE_URL`, `OPENAI_API_KEY` / `OPENAI_BASE_URL`, and `ANTHROPIC_API_KEY` / `ANTHROPIC_BASE_URL` pairs in the boot's layered environment.
 

apps/cli/README.zh.md

@@ -2,19 +2,23 @@
 
 [English](README.md) | 中文
 
-`dsh` 命令行入口遵循 `apps/` 组装层：`apps/*` 是位于 `packages/*` 库之上的产品组装。直接运行 `dsh` 会启动交互式 TUI 编码 agent（智能体），`dsh -p "task"` 运行一个无头轮次，`dsh web` 则提供浏览器 UI。
 
-Argv 只会通过 [Commander](https://github.com/tj/commander.js) 适配器（[`src/args.ts`](src/args.ts)）解析一次：同一个程序的默认形式（无子命令）是 TUI／无头界面（`--config`、`-p`/`--prompt`、`--resume`），`web` 子命令则是浏览器 UI。`src/bin.ts` 按解析后的 mode 分支，仅动态导入该 mode 的模块。`dsh --help` 列出所有 mode，`dsh web --help` 渲染 Web 用法，`dsh --version` 打印此应用的版本；未知选项或拼错的 `--resume` 会明确报错（stderr，退出码 1），而不会被错路由。`dsh web` 的 `--host`/`--port` 是未验证的直通覆盖：`dsh-host-webserver` schema 是默认值（标志缺失时使用已交付的 `cordis.yml` 值）和有效性的唯一真源，并在启动时拒绝错误值。`--trusted-host` 为 /api 浏览器信任栅栏追加具名权威；全接口绑定还会自行推导本机的 LAN IP 字面量（[`src/app-cli-entry.ts`](src/app-cli-entry.ts)），因此打印出的 LAN URL 无需任何标志即可使用。
+Argv 只会通过 [Commander](https://github.com/tj/commander.js) 适配器（[`src/args.ts`](src/args.ts)）解析一次：同一个程序的默认形式（无子命令）是 TUI／无头界面（`--config`、`-p`/`--prompt`、`--resume`），`meta` 子命令是以本 checkout 为 workspace 的同一个 TUI，`upgrade` 子命令是无选项的引导会话入口，`web` 子命令则是浏览器 UI。`src/bin.ts` 按解析后的 mode 分支，仅动态导入该 mode 的模块。`dsh --help` 列出所有 mode，`dsh web --help` 渲染 Web 用法，`dsh --version` 打印此应用的版本；未知选项或拼错的 `--resume` 会明确报错（stderr，退出码 1），而不会被错路由。凡与默认界面不共享任何选项的子命令（`upgrade`、`web`、`meta`）都会拒绝泄漏进来的 `--config`/`-p`/`--resume`，而不会照常运行并丢弃它。`dsh web` 的 `--host`/`--port` 是未验证的直通覆盖：`dsh-host-webserver` schema 是默认值（标志缺失时使用已交付的 Web 覆盖层值）和有效性的唯一真源，并在启动时拒绝错误值。`--trusted-host` 为 /api 浏览器信任栅栏追加具名权威；全接口绑定还会自行推导本机的 LAN IP 字面量（[`src/app-cli-entry.ts`](src/app-cli-entry.ts)），因此打印出的 LAN URL 无需任何标志即可使用。
 
 TUI 界面：
 
-- 启动已交付的默认配置（`examples/tui-agent/cordis.yml`），或由 `--config <path>` 指定的树（演示／测试用于启动其他示例树的逃生口），并通过 [`dsh-app-boot`](../../packages/ui/app-boot/README.md) 完成启动；
-- 使用 `dsh --resume <session-id>` 恢复已持久化会话。当 Node 宿主公开 `process.execve` 时，还会提供 TUI 的原地移交宿主：选择器预检并刷新当前会话后，宿主会释放应用，并以规范化的 `dsh --resume <id>` 替换进程；不支持进程替换的运行时保留屏幕上显示的命令回退。该标志通过 `RESUME_SESSION_ID_KEY` 在启动上下文中提供 id（不使用环境变量），已交付的配置通过 `!!js` 读取它；缺失或无法读取的 id 会明确报错，而不会创建新会话；
-- 将 **调用目录** 视为 workspace：会话、相对路径和 workspace 指令都从 cwd 解析；
+- 通过 [`dsh-app-boot`](../../packages/ui/app-boot/README.md) 启动 `base.cordis.yml` 与 `tui.cordis.yml`；`--config <path>` 应用一个补丁列表覆盖并替代个人覆盖，而 `--config-replace <path>` 将指定文件作为完整配置树启动；
+- 使用 `dsh --resume <session-id>` 恢复已持久化会话。当 Node 宿主公开 `process.execve` 时，还会提供 TUI 的原地移交宿主：选择器预检并刷新当前会话后，宿主会释放应用，并以规范化的恢复调用替换进程；不支持进程替换的运行时会让会话继续运行并给出提示。会话身份与退出行由本 CLI 拥有，而非由配置指定：它创建或选定 `main` 会话 id，并把该 id 以及可复现本次调用的确切命令一起提供到启动上下文（[`MAIN_SESSION_ID_KEY`](../../packages/ui/tui/README.md) 与 `TUI_GOODBYE_MESSAGE_KEY`）。任何 `cordis.yml` 键都无法移除恢复能力；缺失或无法读取的 id 会明确报错，而不会创建新会话；
+- 将 **调用目录** 视为 workspace：会话、相对路径和 workspace 指令都从 cwd 解析（`dsh meta` 是唯一例外，见下文）；
 - 告知 agent 自身源码所在位置：启动后添加一个命名此 harness checkout 的提示词段。该路径从启动器的真实路径解析，因此在 PATH 符号链接和任意 cwd 下仍然有效，使自指的 `cordis` 工具集可以读取并修改它；
 - 应用 `~/.dsh` 中的个人覆盖（参见 [app-boot 的个人配置](../../packages/ui/app-boot/README.md#personal-config)）：`.env` 填补环境缺口（环境中已有的值 > 项目 `.env` > 个人 `.env`），`config.yaml` 则修补已启动的树。
 
-Web 和无头界面启动同一个共享组合（`cordis.yml`）：两者都将调用目录视为默认项目和 Workspace 根目录，除非通过 `--workspace-root <path>` 覆盖，否则会在该根目录下创建具名 Workspace；它们会把适用的 `AGENTS.md`/`CLAUDE.md` 指令加载到每个 agent-loop 请求前缀中，渲染预算为 65,536 字节，并选用首条消息模型标题。无头界面唯一的差异是监听操作系统分配的端口（并行 `dsh -p` 运行绝不冲突；stderr 打印的 URL 会在浏览器中打开实时会话）。两者都需要先构建前端 dist 和客户端 bundle（`pnpm run build && pnpm run build:web`）。
+`dsh meta` 是以本 harness checkout 为 workspace 的同一个 TUI，因此开发 dsh 自身无需 `cd`。它在两层 `.env` 都加载之后才 chdir 到 checkout 根目录（从启动器的真实路径解析，与源码路径提示词段所指的根目录相同），因此环境优先级不变，而会话 cwd 与 HMR 监视根目录会一并移动。Meta 始终创建新会话，不接受默认界面的任何选项；恢复已持久化会话应使用普通的 `dsh --resume <id>`。
+
+`dsh upgrade` 是默认 TUI 界面之上的引导式全新会话入口：它在调用目录中创建一个全新会话，并以内置 `dsh-upgrade` skill 播种其首轮，效果等同于用户手动键入 `/skill:<name>`。启动器将 skill 名称提供到启动上下文（[`INITIAL_SKILL_KEY`](../../packages/ui/tui/README.md)），TUI 在聊天就绪后自动调用它。两者都不接受任何选项——`--config`、`-p`、`--resume` 都会明确报错——且仅在首次启动时播种，因此之后 `dsh --resume <id>` 恢复该会话时是普通 TUI 会话，不会重复注入。
+
+
+Web 和无头界面启动 `base.cordis.yml` 与 `web.cordis.yml`，随后应用 `$DSH_HOME/config.yaml`；显式的 `--config <path>` 会替代该个人覆盖。除此之外，两者共享同一套组合：两者都将调用目录视为默认项目和 Workspace 根目录，除非通过 `--workspace-root <path>` 覆盖，否则会在该根目录下创建具名 Workspace；它们会把适用的 `AGENTS.md`/`CLAUDE.md` 指令加载到每个 agent-loop 请求前缀中，渲染预算为 65,536 字节，并选用首条消息模型标题。无头界面唯一的差异是监听操作系统分配的端口（并行 `dsh -p` 运行绝不冲突；stderr 打印的 URL 会在浏览器中打开实时会话）。两者都需要先构建前端 dist 和客户端 bundle（`pnpm run build && pnpm run build:web`）。
 
 已交付的 TUI 和 Web 组合会注册原生 DeepSeek 适配器，以及 pi-ai 的 OpenAI 和 Anthropic 提供方配置。凭据和端点覆盖来自启动分层环境中的提供方标准变量对：`DEEPSEEK_API_KEY` / `DEEPSEEK_BASE_URL`、`OPENAI_API_KEY` / `OPENAI_BASE_URL` 和 `ANTHROPIC_API_KEY` / `ANTHROPIC_BASE_URL`。
 

examples/tui-agent/composition.md → apps/cli/composition.md

@@ -3,36 +3,71 @@
 
 # TUI Agent App Composition
 
-The TUI agent combines the real DeepSeek adapter, coding tools, compaction, subagents, and workflows with the full-screen terminal app package.
+The TUI surface combines the shared CLI base with its surface overlay and full-screen terminal package.
 
 ```mermaid
 flowchart LR
-  cfg["examples/tui-agent<br/>cordis.yml"]
+  cfg["apps/cli/config<br/>cordis.yml"]
+  plugin_tui_timer["timer<br/>@cordisjs/plugin-timer"]
+  cfg --> plugin_tui_timer
   plugin_tui_hmr["hmr<br/>@cordisjs/plugin-hmr"]
   cfg --> plugin_tui_hmr
-  plugin_tui_llm_deepseek["llm-deepseek<br/>@deepseek-ai/dsh-llm-deepseek"]
-  cfg --> plugin_tui_llm_deepseek
+  plugin_tui_llm["llm<br/>@deepseek-ai/dsh-llm"]
+  cfg --> plugin_tui_llm
+  plugin_tui_session["session<br/>@deepseek-ai/dsh-session"]
+  cfg --> plugin_tui_session
+  plugin_tui_session_title["session-title<br/>@deepseek-ai/dsh-session-title"]
+  cfg --> plugin_tui_session_title
+  plugin_tui_session_title_llm["session-title-llm<br/>@deepseek-ai/dsh-session-title-first-message-llm"]
+  cfg --> plugin_tui_session_title_llm
+  plugin_tui_user_interaction["user-interaction<br/>@deepseek-ai/dsh-user-interaction"]
+  cfg --> plugin_tui_user_interaction
+  plugin_tui_agent["agent<br/>@deepseek-ai/dsh-agent"]
+  cfg --> plugin_tui_agent
+  plugin_tui_tasks["tasks<br/>@deepseek-ai/dsh-tasks-local"]
+  cfg --> plugin_tui_tasks
+  plugin_tui_llm_retry["llm-retry<br/>@deepseek-ai/dsh-llm-retry"]
+  cfg --> plugin_tui_llm_retry
   plugin_tui_llm_pi_ai["llm-pi-ai<br/>@deepseek-ai/dsh-llm-pi-ai"]
   cfg --> plugin_tui_llm_pi_ai
+  plugin_tui_session_persistence_jsonl["session-persistence-jsonl<br/>@deepseek-ai/dsh-session-persistence-jsonl"]
+  cfg --> plugin_tui_session_persistence_jsonl
+  plugin_tui_session_query_sqlite["session-query-sqlite<br/>@deepseek-ai/dsh-session-query-sqlite"]
+  cfg --> plugin_tui_session_query_sqlite
   plugin_tui_subprocess["subprocess<br/>@deepseek-ai/dsh-subprocess-local"]
   cfg --> plugin_tui_subprocess
-  plugin_tui_bash["bash<br/>@deepseek-ai/dsh-bash-local"]
-  cfg --> plugin_tui_bash
-  plugin_tui_tui_agent["tui-agent<br/>@deepseek-ai/dsh-tui-demo"]
-  cfg --> plugin_tui_tui_agent
-  plugin_tui_tui_agent --> bundle_agent_core["@deepseek-ai/dsh-agent-spine-demo"]
-  plugin_tui_tui_agent --> bundle_jsonl["@deepseek-ai/dsh-session-persistence-jsonl"]
-  plugin_tui_tui_agent --> frontdoor_tui["@deepseek-ai/dsh-tui<br/>pre-created main agent"]
-  bundle_agent_core --> spine_llm["ctx.llm"]
-  bundle_agent_core --> spine_sessions["ctx.sessions"]
-  bundle_agent_core --> spine_tools["ctx.tools + tool-bash"]
-  bundle_agent_core --> spine_loop["ctx.agents + ctx.agentLoop"]
-  plugin_tui_session_title_llm["session-title-llm<br/>@deepseek-ai/dsh-session-title-first-message-llm"]
-  cfg --> plugin_tui_session_title_llm
+  plugin_tui_bash_local["bash-local<br/>@deepseek-ai/dsh-bash-local"]
+  cfg --> plugin_tui_bash_local
+  plugin_tui_tool_bash["tool-bash<br/>@deepseek-ai/dsh-tool-bash"]
+  cfg --> plugin_tui_tool_bash
+  plugin_tui_tool_tasks["tool-tasks<br/>@deepseek-ai/dsh-tool-tasks"]
+  cfg --> plugin_tui_tool_tasks
+  plugin_tui_fs_policy["fs-policy<br/>@deepseek-ai/dsh-fs-policy"]
+  cfg --> plugin_tui_fs_policy
+  plugin_tui_tool_fs["tool-fs<br/>@deepseek-ai/dsh-tool-fs"]
+  cfg --> plugin_tui_tool_fs
+  plugin_tui_tool_fs_search["tool-fs-search<br/>@deepseek-ai/dsh-tool-fs-search"]
+  cfg --> plugin_tui_tool_fs_search
+  plugin_tui_workspace_context["workspace-context<br/>@deepseek-ai/dsh-workspace-context"]
+  cfg --> plugin_tui_workspace_context
+  plugin_tui_skill["skill<br/>@deepseek-ai/dsh-skill"]
+  cfg --> plugin_tui_skill
+  plugin_tui_skill_local["skill-local<br/>@deepseek-ai/dsh-skill-local"]
+  cfg --> plugin_tui_skill_local
+  plugin_tui_tool_skill["tool-skill<br/>@deepseek-ai/dsh-tool-skill"]
+  cfg --> plugin_tui_tool_skill
+  plugin_tui_commands["commands<br/>@deepseek-ai/dsh-commands"]
+  cfg --> plugin_tui_commands
+  plugin_tui_goal["goal<br/>@deepseek-ai/dsh-goal"]
+  cfg --> plugin_tui_goal
+  plugin_tui_goal_session["goal-session<br/>@deepseek-ai/dsh-goal-session"]
+  cfg --> plugin_tui_goal_session
+  plugin_tui_command_goal["command-goal<br/>@deepseek-ai/dsh-command-goal"]
+  cfg --> plugin_tui_command_goal
+  plugin_tui_plan_mode["plan-mode<br/>@deepseek-ai/dsh-plan-mode"]
+  cfg --> plugin_tui_plan_mode
   plugin_tui_token_meter["token-meter<br/>@deepseek-ai/dsh-token-meter"]
   cfg --> plugin_tui_token_meter
-  plugin_tui_tool_result_prune["tool-result-prune<br/>@deepseek-ai/dsh-compact-tool-result-prune"]
-  cfg --> plugin_tui_tool_result_prune
   plugin_tui_compact_basic["compact-basic<br/>@deepseek-ai/dsh-compact-basic"]
   cfg --> plugin_tui_compact_basic
   plugin_tui_subagent["subagent<br/>@deepseek-ai/dsh-subagent"]
@@ -49,37 +84,56 @@ flowchart LR
   cfg --> plugin_tui_workflow_workerthread
   plugin_tui_tool_workflow["tool-workflow<br/>@deepseek-ai/dsh-tool-workflow"]
   cfg --> plugin_tui_tool_workflow
-  plugin_tui_tool_ralph["tool-ralph<br/>@deepseek-ai/dsh-tool-ralph"]
-  cfg --> plugin_tui_tool_ralph
-  plugin_tui_plan_mode["plan-mode<br/>@deepseek-ai/dsh-plan-mode"]
-  cfg --> plugin_tui_plan_mode
-  plugin_tui_fs_local["fs-local<br/>@deepseek-ai/dsh-fs-local"]
-  cfg --> plugin_tui_fs_local
-  plugin_tui_fs_policy["fs-policy<br/>@deepseek-ai/dsh-fs-policy"]
-  cfg --> plugin_tui_fs_policy
-  plugin_tui_tool_fs["tool-fs<br/>@deepseek-ai/dsh-tool-fs"]
-  cfg --> plugin_tui_tool_fs
-  plugin_tui_tool_fs_search["tool-fs-search<br/>@deepseek-ai/dsh-tool-fs-search"]
-  cfg --> plugin_tui_tool_fs_search
   plugin_tui_timeout_policy["timeout-policy<br/>@deepseek-ai/dsh-timeout-policy"]
   cfg --> plugin_tui_timeout_policy
   plugin_tui_spill_local["spill-local<br/>@deepseek-ai/dsh-spill-local"]
   cfg --> plugin_tui_spill_local
   plugin_tui_spill_policy["spill-policy<br/>@deepseek-ai/dsh-spill-policy"]
   cfg --> plugin_tui_spill_policy
+  plugin_tui_tools["tools<br/>@deepseek-ai/dsh-tools"]
+  cfg --> plugin_tui_tools
+  plugin_tui_system_prompt["system-prompt<br/>@deepseek-ai/dsh-system-prompt"]
+  cfg --> plugin_tui_system_prompt
+  plugin_tui_agent_loop["agent-loop<br/>@deepseek-ai/dsh-agent-loop"]
+  cfg --> plugin_tui_agent_loop
+  plugin_tui_fs_local["fs-local<br/>@deepseek-ai/dsh-fs-local"]
+  cfg --> plugin_tui_fs_local
+  plugin_tui_llm_deepseek["llm-deepseek<br/>@deepseek-ai/dsh-llm-deepseek"]
+  cfg --> plugin_tui_llm_deepseek
 ```
 
 | Plugin id | Package / module |
 | --- | --- |
+| `timer` | `@cordisjs/plugin-timer` |
 | `hmr` | `@cordisjs/plugin-hmr` |
-| `llm-deepseek` | `@deepseek-ai/dsh-llm-deepseek` |
+| `llm` | `@deepseek-ai/dsh-llm` |
+| `session` | `@deepseek-ai/dsh-session` |
+| `session-title` | `@deepseek-ai/dsh-session-title` |
+| `session-title-llm` | `@deepseek-ai/dsh-session-title-first-message-llm` |
+| `user-interaction` | `@deepseek-ai/dsh-user-interaction` |
+| `agent` | `@deepseek-ai/dsh-agent` |
+| `tasks` | `@deepseek-ai/dsh-tasks-local` |
+| `llm-retry` | `@deepseek-ai/dsh-llm-retry` |
 | `llm-pi-ai` | `@deepseek-ai/dsh-llm-pi-ai` |
+| `session-persistence-jsonl` | `@deepseek-ai/dsh-session-persistence-jsonl` |
+| `session-query-sqlite` | `@deepseek-ai/dsh-session-query-sqlite` |
 | `subprocess` | `@deepseek-ai/dsh-subprocess-local` |
-| `bash` | `@deepseek-ai/dsh-bash-local` |
-| `tui-agent` | `@deepseek-ai/dsh-tui-demo` |
-| `session-title-llm` | `@deepseek-ai/dsh-session-title-first-message-llm` |
+| `bash-local` | `@deepseek-ai/dsh-bash-local` |
+| `tool-bash` | `@deepseek-ai/dsh-tool-bash` |
+| `tool-tasks` | `@deepseek-ai/dsh-tool-tasks` |
+| `fs-policy` | `@deepseek-ai/dsh-fs-policy` |
+| `tool-fs` | `@deepseek-ai/dsh-tool-fs` |
+| `tool-fs-search` | `@deepseek-ai/dsh-tool-fs-search` |
+| `workspace-context` | `@deepseek-ai/dsh-workspace-context` |
+| `skill` | `@deepseek-ai/dsh-skill` |
+| `skill-local` | `@deepseek-ai/dsh-skill-local` |
+| `tool-skill` | `@deepseek-ai/dsh-tool-skill` |
+| `commands` | `@deepseek-ai/dsh-commands` |
+| `goal` | `@deepseek-ai/dsh-goal` |
+| `goal-session` | `@deepseek-ai/dsh-goal-session` |
+| `command-goal` | `@deepseek-ai/dsh-command-goal` |
+| `plan-mode` | `@deepseek-ai/dsh-plan-mode` |
 | `token-meter` | `@deepseek-ai/dsh-token-meter` |
-| `tool-result-prune` | `@deepseek-ai/dsh-compact-tool-result-prune` |
 | `compact-basic` | `@deepseek-ai/dsh-compact-basic` |
 | `subagent` | `@deepseek-ai/dsh-subagent` |
 | `subagent-spawn` | `@deepseek-ai/dsh-subagent-spawn` |
@@ -88,16 +142,15 @@ flowchart LR
 | `tool-subagent-fork` | `@deepseek-ai/dsh-tool-subagent` |
 | `workflow-workerthread` | `@deepseek-ai/dsh-workflow-workerthread` |
 | `tool-workflow` | `@deepseek-ai/dsh-tool-workflow` |
-| `tool-ralph` | `@deepseek-ai/dsh-tool-ralph` |
-| `plan-mode` | `@deepseek-ai/dsh-plan-mode` |
-| `fs-local` | `@deepseek-ai/dsh-fs-local` |
-| `fs-policy` | `@deepseek-ai/dsh-fs-policy` |
-| `tool-fs` | `@deepseek-ai/dsh-tool-fs` |
-| `tool-fs-search` | `@deepseek-ai/dsh-tool-fs-search` |
 | `timeout-policy` | `@deepseek-ai/dsh-timeout-policy` |
 | `spill-local` | `@deepseek-ai/dsh-spill-local` |
 | `spill-policy` | `@deepseek-ai/dsh-spill-policy` |
+| `tools` | `@deepseek-ai/dsh-tools` |
+| `system-prompt` | `@deepseek-ai/dsh-system-prompt` |
+| `agent-loop` | `@deepseek-ai/dsh-agent-loop` |
+| `fs-local` | `@deepseek-ai/dsh-fs-local` |
+| `llm-deepseek` | `@deepseek-ai/dsh-llm-deepseek` |
 
-Source config: [`examples/tui-agent/cordis.yml`](cordis.yml).
+Source config: [`apps/cli/config/base.cordis.yml`](config/base.cordis.yml).
 
 Maintenance mode: hybrid: the leaf plugin list is parsed from its `cordis.yml`; app package expansion is curated from package source.

examples/tui-agent/cordis.yml → apps/cli/config/base.cordis.yml

@@ -1,24 +1,60 @@
-# Full-screen TUI coding agent with swappable model and local-bash backends.
-# `dsh-tui-demo` supplies the agent spine, workspace instructions, generic
-# task controls, JSONL persistence, the pi-tui front door, and `main`.
-# HMR remains a leaf because it depends on Loader internals. The app bin loads
-# the gitignored root `.env`; this file reads `DEEPSEEK_API_KEY` and optional
-# `DEEPSEEK_BASE_URL` through `!!js`.
+# The shared `dsh` core: every row both the TUI (`tui.cordis.yml`) and the web
+# surface (`web.cordis.yml`) mount identically. Neither surface includes the
+# other — each is a patch list applied over THIS file at one include level, so a
+# surface overlay, a `--config` overlay, and the personal `~/.dsh/config.yaml`
+# all address these rows by id. Patch lists stack in that order, last write
+# winning per row.
+#
+# A patch replaces the targeted row's whole `config` rather than merging into
+# it, so a row whose value differs per surface does NOT live here: it belongs to
+# each overlay, keeping any single row down to one overlay layer plus the user's.
+# Rows with surface-specific values appear below only with shared plugin identity
+# and neutral defaults; each overlay restates the complete surface configuration.
+#
+# Row order carries no load semantics (activation is service-availability
+# driven); the grouping is for readers.
+
+- id: timer
+  name: '@cordisjs/plugin-timer'
 
 - id: hmr
   name: '@cordisjs/plugin-hmr'
   config:
     root: ['.']
 
-# The native DeepSeek adapter. Shipped default: full thinking at max effort on
-# every request (wire-only defaults; they never enter the request header).
-- id: llm-deepseek
-  name: '@deepseek-ai/dsh-llm-deepseek'
+- id: llm
+  name: '@deepseek-ai/dsh-llm'
+
+- id: session
+  name: '@deepseek-ai/dsh-session'
+
+- id: session-title
+  name: '@deepseek-ai/dsh-session-title'
   config:
-    apiKey: !!js process.env.DEEPSEEK_API_KEY
-    baseURL: !!js process.env.DEEPSEEK_BASE_URL
-    thinking: enabled
-    reasoningEffort: max
+    fallbackMaxWords: 5
+    fallbackMaxBytes: 40
+    maxTitleBytes: 80
+
+- id: session-title-llm
+  name: '@deepseek-ai/dsh-session-title-first-message-llm'
+  config:
+    targetWords: 5
+    targetCjkCharacters: 10
+    maxInputBytes: 4096
+    maxOutputTokens: 64
+    timeoutMs: 60000
+
+- id: user-interaction
+  name: '@deepseek-ai/dsh-user-interaction'
+
+- id: agent
+  name: '@deepseek-ai/dsh-agent'
+
+- id: tasks
+  name: '@deepseek-ai/dsh-tasks-local'
+
+- id: llm-retry
+  name: '@deepseek-ai/dsh-llm-retry'
 
 - id: llm-pi-ai
   name: '@deepseek-ai/dsh-llm-pi-ai'
@@ -31,72 +67,90 @@
         apiKey: !!js process.env.ANTHROPIC_API_KEY
         baseURL: !!js process.env.ANTHROPIC_BASE_URL
 
-# Local executor for the app bundle's bash tool.
-# Managed child-process groups for the bash executor (spawn/kill/output plumbing).
+- id: session-persistence-jsonl
+  name: '@deepseek-ai/dsh-session-persistence-jsonl'
+  config:
+    root: !!js >-
+      (() => { const path = process.getBuiltinModule('node:path'); const home = process.getBuiltinModule('node:os').homedir(); const configured = process.env.DSH_HOME; const selected = configured !== undefined && configured.trim().length > 0 ? configured : path.join(home, '.dsh'); const expanded = selected === '~' ? home : selected.startsWith('~/') || selected.startsWith('~\\') ? path.join(home, selected.slice(2)) : selected; return path.join(path.resolve(expanded), 'sessions') })()
+
+# TUI consumes this shared session capability. Its launcher supplies a unique
+# process-local path; non-TUI surfaces disable the row in their overlay.
+- id: session-query-sqlite
+  name: '@deepseek-ai/dsh-session-query-sqlite'
+  config:
+    path: !!js launcherSessionQueryPath ?? './.sessions/session-query.db'
+
 - id: subprocess
   name: '@deepseek-ai/dsh-subprocess-local'
 
-- id: bash
+- id: bash-local
   name: '@deepseek-ai/dsh-bash-local'
   config:
     timeoutMs: 60000
 
-# The app bundle pre-creates the TUI's `main` agent.
-- id: tui-agent
-  name: '@deepseek-ai/dsh-tui-demo'
-  config:
-    provider: deepseek
-    model: deepseek-v4-pro
-    # `dsh --resume <id>` provides the session id on the boot context (the ids
-    # live under ./.sessions); with no flag the identifier is undefined and a
-    # fresh session starts each run. The typeof guard tolerates a launcher that
-    # never provides the slot, reading undefined rather than throwing.
-    resumeSessionId: !!js "typeof resumeSessionId === 'string' ? resumeSessionId : undefined"
-    persistenceRoot: './.sessions'
-    # Printed on exit and listed by `/resume`; `{session}` fills the live id.
-    # `dsh --resume <id>` resumes that session, so run it from this cwd.
-    resumeCommand: 'dsh --resume {session}'
-    workspaceContext:
-      maxBytes: 65536
-    ui:
-      showReasoning: true
-      maxToolOutputLines: 6
-    # Keep the persona to identity and behavior; tool plugins own tool guidance.
-    # The loop resolves {{model}} from this agent's configuration.
-    persona: |
-      You are a coding agent powered by the {{model}} model.
-
-      Verify your work by running the code or tests. Keep answers brief and
-      factual.
-
-# Model-made session titles on the first-message cadence: replaces the spine's
-# deterministic fallback title with a short model summary. The TUI renders the
-# logged `session/title` as the banner subtitle and the terminal window title.
-# Omitting provider/model inherits the main request's exact route.
-- id: session-title-llm
-  name: '@deepseek-ai/dsh-session-title-first-message-llm'
+- id: tool-bash
+  name: '@deepseek-ai/dsh-tool-bash'
+
+- id: tool-tasks
+  name: '@deepseek-ai/dsh-tool-tasks'
+
+- id: fs-policy
+  name: '@deepseek-ai/dsh-fs-policy'
+
+- id: tool-fs
+  name: '@deepseek-ai/dsh-tool-fs'
+
+- id: tool-fs-search
+  name: '@deepseek-ai/dsh-tool-fs-search'
+
+- id: workspace-context
+  name: '@deepseek-ai/dsh-workspace-context'
   config:
-    targetWords: 5
-    targetCjkCharacters: 10
-    maxInputBytes: 4096
-    maxOutputTokens: 64
-    timeoutMs: 60000
+    maxBytes: 65536
+
+- id: skill
+  name: '@deepseek-ai/dsh-skill'
+
+- id: skill-local
+  name: '@deepseek-ai/dsh-skill-local'
+
+- id: tool-skill
+  name: '@deepseek-ai/dsh-tool-skill'
+
+- id: commands
+  name: '@deepseek-ai/dsh-commands'
+
+- id: goal
+  name: '@deepseek-ai/dsh-goal'
+
+- id: goal-session
+  name: '@deepseek-ai/dsh-goal-session'
+
+- id: command-goal
+  name: '@deepseek-ai/dsh-command-goal'
+
+- id: plan-mode
+  name: '@deepseek-ai/dsh-plan-mode'
+  config:
+    section: |
+          You are in plan mode. Stay in plan mode until exit_plan_mode succeeds or the user switches the session mode. Imperative language to implement changes means plan the implementation, not execute it. A user's conversational agreement — including an answer confirming something you asked — approves nothing and does not end plan mode; fold the confirmed decision into the plan and submit it through exit_plan_mode.
+
+          Explore first. Use non-mutating reads, searches, static analysis, and checks to ground the plan in the actual repository. Do not edit or write files, change configuration, run formatters or code generation that rewrites tracked files, commit, or otherwise carry out the plan. Prefer existing functions and patterns over new machinery.
+
+          The tool catalog stays the same across modes for request-cache stability. These plan-mode rules override any later tool description or guidance that suggests using mutation tools; those tools remain listed only to keep the request shape stable. Do not use todo_write to track this planning phase: it tracks implementation after an approved plan, while the plan itself belongs in exit_plan_mode.
+
+          Resolve discoverable facts by inspection. Use ask_user_question only for user-owned choices or material ambiguity that inspection cannot answer. Do not ask the user where code lives or how current behavior works when you can find out.
+
+          Make the plan decision-complete: state the goal and success criteria; group implementation changes by subsystem; identify public API, schema, and data-flow changes; cover edge cases, failure modes, tests, acceptance criteria, and explicit assumptions. Keep it concise enough to review but detailed enough that another engineer can implement it without making design decisions.
+
+          When ready, call exit_plan_mode with the complete plan markdown, starting with a # title. Make exit_plan_mode the only and final tool call in that assistant response: it presents the plan for approval, and implementation begins only in a later step after approval. Do not paste the final plan as a plain reply or ask "should I proceed?" through prose or ask_user_question. If review rejects it, incorporate the feedback and present again. If the review channel is unavailable or aborted, stay in plan mode and ask the user to switch modes manually; do not proceed with implementation.
 
-# Replay-aware request pressure with one service-wide context window.
 - id: token-meter
   name: '@deepseek-ai/dsh-token-meter'
 
-- id: tool-result-prune
-  name: '@deepseek-ai/dsh-compact-tool-result-prune'
-
-# Summarize an older range after measured pressure or a canonical provider overflow.
-# Service-wide policy provides pressure, retention, and one overflow-retry default.
 - id: compact-basic
   name: '@deepseek-ai/dsh-compact-basic'
 
-# Expose fresh-child `spawn` and completed-prefix `fork` through independent
-# in-process backends. Each tool instance needs a distinct `toolName`; the registry
-# rejects duplicates. These leaves follow the app because it provides `ctx.agents` and `ctx.tools`.
 - id: subagent
   name: '@deepseek-ai/dsh-subagent'
 
@@ -122,9 +176,6 @@
     provider: fork
     toolName: subagent_fork
 
-
-# The worker-thread workflow engine fans a model-written JavaScript script's
-# `agent()` calls out through the spawn backend; the adjacent tool exposes it to the model.
 - id: workflow-workerthread
   name: '@deepseek-ai/dsh-workflow-workerthread'
   config:
@@ -133,64 +184,47 @@
 - id: tool-workflow
   name: '@deepseek-ai/dsh-tool-workflow'
 
-# A separate fixed consumer demonstrates fresh-agent Ralph iteration without
-# changing the workflow tool or same-session goal behavior.
-- id: tool-ralph
-  name: '@deepseek-ai/dsh-tool-ralph'
+- id: timeout-policy
+  name: '@deepseek-ai/dsh-timeout-policy'
 
-# Plan mode gives the TUI plugin-owned /plan [message] entry and /plan off exit
-# commands; the reviewed exit rides the TUI's user-interaction provider.
-- id: plan-mode
-  name: '@deepseek-ai/dsh-plan-mode'
-  config:
-    section: |
-          You are in plan mode. Stay in plan mode until exit_plan_mode succeeds or the user switches the session mode. Imperative language to implement changes means plan the implementation, not execute it. A user's conversational agreement — including an answer confirming something you asked — approves nothing and does not end plan mode; fold the confirmed decision into the plan and submit it through exit_plan_mode.
+- id: spill-local
+  name: '@deepseek-ai/dsh-spill-local'
 
-          Explore first. Use non-mutating reads, searches, static analysis, and checks to ground the plan in the actual repository. Do not edit or write files, change configuration, run formatters or code generation that rewrites tracked files, commit, or otherwise carry out the plan. Prefer existing functions and patterns over new machinery.
+- id: spill-policy
+  name: '@deepseek-ai/dsh-spill-policy'
+  config:
+    maxInlineBytes: 50000
 
-          The tool catalog stays the same across modes for request-cache stability. These plan-mode rules override any later tool description or guidance that suggests using mutation tools; those tools remain listed only to keep the request shape stable. Do not use todo_write to track this planning phase: it tracks implementation after an approved plan, while the plan itself belongs in exit_plan_mode.
+# ── rows every surface mounts, whose values each overlay states ──────────────
 
-          Resolve discoverable facts by inspection. Use ask_user_question only for user-owned choices or material ambiguity that inspection cannot answer. Do not ask the user where code lives or how current behavior works when you can find out.
+# The tool registry. Presentation mode is a surface choice, so each overlay
+# states it; omitting it here keeps the schema default (native).
+- id: tools
+  name: '@deepseek-ai/dsh-tools'
 
-          Make the plan decision-complete: state the goal and success criteria; group implementation changes by subsystem; identify public API, schema, and data-flow changes; cover edge cases, failure modes, tests, acceptance criteria, and explicit assumptions. Keep it concise enough to review but detailed enough that another engineer can implement it without making design decisions.
+# The deployment persona is a surface choice; plan-mode and tool plugins own
+# their own prompt sections.
+- id: system-prompt
+  name: '@deepseek-ai/dsh-system-prompt'
+  config:
+    persona: ''
 
-          When ready, call exit_plan_mode with the complete plan markdown, starting with a # title. Make exit_plan_mode the only and final tool call in that assistant response: it presents the plan for approval, and implementation begins only in a later step after approval. Do not paste the final plan as a plain reply or ask "should I proceed?" through prose or ask_user_question. If review rejects it, incorporate the feedback and present again. If the review channel is unavailable or aborted, stay in plan mode and ask the user to switch modes manually; do not proceed with implementation.
+# Agents created at startup. The TUI pre-creates `main`; the web surface creates
+# sessions on client request, so its overlay keeps this empty.
+- id: agent-loop
+  name: '@deepseek-ai/dsh-agent-loop'
+  config:
+    agents: []
 
-# Policy loads before the model-facing filesystem tools so writes and edits require
-# an observed file. This single-session app resolves relative paths from the process cwd.
+# The filesystem provider. `cwd` defaults to the package's `process.cwd()`; the
+# TUI states it explicitly because that value is also the session workspace.
 - id: fs-local
   name: '@deepseek-ai/dsh-fs-local'
-  config:
-    cwd: !!js process.cwd()
-
-- id: fs-policy
-  name: '@deepseek-ai/dsh-fs-policy'
-
-- id: tool-fs
-  name: '@deepseek-ai/dsh-tool-fs'
-
-# Bash-backed discovery tools (glob/grep): fixed ripgrep commands through the
-# local bash executor above — not ctx.fs. Capped results save the complete
-# formatted list through the spill backend below (ctx.spillStore, optional).
-- id: tool-fs-search
-  name: '@deepseek-ai/dsh-tool-fs-search'
 
-# The tool-call timeout enforcer: arms each declared ToolDefinition.timeoutMs
-# (the search tools above declare 30s) as a deadline on exec.signal. Without
-# it a declared budget is advisory and only the bash executor's own timeout
-# backstop applies.
-- id: timeout-policy
-  name: '@deepseek-ai/dsh-timeout-policy'
-
-# Tool-output spill stack: a local backend that saves oversized tool text under
-# a private session-scoped dir, and the tools/post-execute policy that replaces
-# an over-budget plain-text result with a preview + the spill locator/retrieval
-# hint. A leaf pair after the app (needs ctx.tools). The policy is a no-op until
-# a tool returns more than maxInlineBytes of plain text.
-- id: spill-local
-  name: '@deepseek-ai/dsh-spill-local'
-
-- id: spill-policy
-  name: '@deepseek-ai/dsh-spill-policy'
+# The native DeepSeek adapter; reads the key/base-url the boot's layered .env
+# loading left in the environment. Thinking defaults are a surface choice.
+- id: llm-deepseek
+  name: '@deepseek-ai/dsh-llm-deepseek'
   config:
-    maxInlineBytes: 50000
+    apiKey: !!js process.env.DEEPSEEK_API_KEY
+    baseURL: !!js process.env.DEEPSEEK_BASE_URL

apps/cli/config/tui.cordis.yml

@@ -0,0 +1,113 @@
+# `dsh` (the default surface) — the full-screen TUI, as a patch list over
+# `base.cordis.yml`. The launcher includes the base and applies this file, then
+# any `--config` overlay, then the personal `~/.dsh/config.yaml`, as sibling
+# patch lists at ONE include level: patches never cross an include boundary, so
+# stacking overlays as nested includes would silently stop reaching base rows.
+#
+# A patch replaces the targeted row's whole `config`, so each row below restates
+# every key it owns. A patch whose `id` matches no row is skipped with a Loader
+# warning, which is deliberate: one personal overlay is shared across surfaces,
+# so a row that exists only under `web` must not fail the TUI's boot.
+#
+# The launcher owns session identity and the exit line, and provides both on the
+# boot context rather than through config, so no key here — and no overlay
+# replacing one — can drop `--resume`.
+
+# ── surface-specific values the base deliberately omits ─────────────────────
+
+# `main` is the agent the TUI drives. `provider`/`model` are the route `dsh
+# login` rewrites and a personal overlay repoints; `cwd` anchors the session to
+# the invoking directory, which is also what scopes `/resume` to this workspace.
+- id: agent-loop
+  config:
+    agents:
+      - id: main
+        provider: deepseek
+        model: deepseek-v4-pro
+        cwd: !!js process.cwd()
+
+# Keep the persona to identity and behavior; tool plugins own tool guidance.
+# The loop resolves {{model}} from this agent's configuration.
+- id: system-prompt
+  config:
+    persona: |
+      You are a coding agent powered by the {{model}} model.
+
+      Verify your work by running the code or tests. Keep answers brief and
+      factual.
+
+# Shipped default: full thinking at max effort on every request (wire-only
+# defaults; they never enter the request header).
+- id: llm-deepseek
+  config:
+    apiKey: !!js process.env.DEEPSEEK_API_KEY
+    baseURL: !!js process.env.DEEPSEEK_BASE_URL
+    thinking: enabled
+    reasoningEffort: max
+
+# This single-session app resolves relative paths from the process cwd.
+- id: fs-local
+  config:
+    cwd: !!js process.cwd()
+
+# The shipped TUI presents the native tool registry.
+- id: tools
+  config:
+    mode: native
+
+# ── TUI-only rows ───────────────────────────────────────────────────────────
+
+- insert:
+    # Relational runtime checks over the authoritative event streams; each
+    # companion registers the assertions its own package owns.
+    - id: invariants
+      name: '@deepseek-ai/dsh-invariants'
+    - id: session-invariant
+      name: '@deepseek-ai/dsh-session/invariant'
+    - id: agent-invariant
+      name: '@deepseek-ai/dsh-agent/invariant'
+    - id: scope-invariant
+      name: '@deepseek-ai/dsh-scope/invariant'
+    - id: agent-loop-invariant
+      name: '@deepseek-ai/dsh-agent-loop/invariant'
+
+    - id: session-checkpoint-policy
+      name: '@deepseek-ai/dsh-session-checkpoint-policy'
+
+    # The derived query index behind `/resume`. The launcher provides a unique
+    # process-local path because this SQLite backend has one writer owner; the
+    # project-local fallback applies when no launcher sets the typed slot.
+    - id: session-reference
+      name: '@deepseek-ai/dsh-session-reference'
+
+    # Compacts oversized tool results before the broader conversation compactor
+    # runs, preserving the model-visible result within the configured budget.
+    - id: tool-result-prune
+      name: '@deepseek-ai/dsh-compact-tool-result-prune'
+
+    # Persisted same-session goals reach the model and the slash menu here; the
+    # domain, driver, and `/goal` command are in the base.
+    - id: tool-goal
+      name: '@deepseek-ai/dsh-tool-goal'
+
+    # A separate fixed consumer demonstrates fresh-agent Ralph iteration without
+    # changing the workflow tool or same-session goal behavior.
+    - id: tool-ralph
+      name: '@deepseek-ai/dsh-tool-ralph'
+
+    # The keyboard-backed provider behind ask_user_question and the plan-mode
+    # review, and the front door it renders inside.
+    - id: tui-prompt
+      name: '@deepseek-ai/dsh-tui/prompt'
+
+    # The TUI renders exactly the agent the agent-loop row bound, so it reads the
+    # same launcher-owned identity rather than restating one.
+    - id: tui
+      name: '@deepseek-ai/dsh-tui'
+      config:
+        sessionId: !!js configuredAgentIdentities?.main?.id ?? 'main'
+        showReasoning: true
+        maxToolOutputLines: 6
+
+    - id: tool-ask-user
+      name: '@deepseek-ai/dsh-tool-ask-user'

apps/cli/config/web.cordis.yml

@@ -0,0 +1,215 @@
+# `dsh web` — the browser surface, as a patch list over `base.cordis.yml`.
+# The launcher includes the base and applies this file, then any `--config`
+# overlay, then AppCLIEntry's profile-json and CLI-flag patches, as sibling patch
+# lists at ONE include level: patches never cross an include boundary, so
+# stacking overlays as nested includes would silently stop reaching base rows.
+#
+# A patch replaces the targeted row's whole `config`, so each row below restates
+# every key it owns. `--dev` appends the dsh-client-hmr row in code
+# (AppCLIEntry).
+
+# ── surface-specific values the base deliberately omits ─────────────────────
+
+# TODO: Re-enable shared HMR for Web after its reload lifecycle is tested.
+- id: hmr
+  disabled: true
+
+# Session query is a TUI capability; Web owns its own session presentation.
+- id: session-query-sqlite
+  disabled: true
+
+- id: tools
+  config:
+    # TEMPORARY workaround: DSH_TOOLS_MODE (native|code|both) opts a whole dsh
+    # process into Code Mode while per-session tool-mode selection is being
+    # designed; unset keeps the schema default (native). Remove the env seam
+    # once the web UI owns the choice per session.
+    mode: !!js process.env.DSH_TOOLS_MODE
+
+- id: llm-deepseek
+  config:
+    apiKey: !!js process.env.DEEPSEEK_API_KEY
+    baseURL: !!js process.env.DEEPSEEK_BASE_URL
+
+# The web surface replaces the unrestricted local executors with the shared
+# sandbox policy. Its default preserves the previous unrestricted behavior;
+# DSH_PERMISSION_MODE and the browser permission picker can confine a session.
+- insert:
+    - id: sandbox
+      name: '@deepseek-ai/dsh-sandbox-local'
+
+    - id: sandbox-policy
+      name: '@deepseek-ai/dsh-sandbox-policy'
+      config:
+        mode: !!js process.env.DSH_PERMISSION_MODE ?? 'danger-full-access'
+        workspaceRoot: !!js process.cwd()
+
+    - id: bash-sandbox
+      name: '@deepseek-ai/dsh-bash-sandbox'
+
+    - id: approval
+      name: '@deepseek-ai/dsh-user-approval'
+      config:
+        policy: !!js "(process.env.DSH_PERMISSION_MODE ?? 'danger-full-access') === 'danger-full-access' ? 'never' : 'ask'"
+
+    - id: permission
+      name: '@deepseek-ai/dsh-permission'
+      config:
+        presets:
+          read-only:
+            sandbox: read-only
+            approval: ask
+          workspace-write:
+            sandbox: workspace-write
+            approval: ask
+          danger-full-access:
+            sandbox: danger-full-access
+            approval: never
+
+    - id: fs-sandbox
+      name: '@deepseek-ai/dsh-fs-sandbox'
+
+- id: bash-local
+  disabled: true
+
+- id: fs-local
+  disabled: true
+
+# ── web-only host rows, the transport layer, and the browser roster ─────────
+
+# `dshClient` rows are the browser roster the modules node half scans into
+# window.__DSH_BOOT__; the modules row is simultaneously a host row.
+- insert:
+    - id: session-projection
+      name: '@deepseek-ai/dsh-session-projection'
+
+    - id: code-runtime
+      name: '@deepseek-ai/dsh-code-runtime-worker'
+
+    - id: storage
+      name: '@deepseek-ai/dsh-storage'
+
+    - id: storage-json
+      name: '@deepseek-ai/dsh-storage-json'
+      config:
+        root: './.storages'
+
+    - id: storage-domain
+      name: '@deepseek-ai/dsh-storage-domain'
+      config:
+        backend: json
+
+    - id: workspace
+      name: '@deepseek-ai/dsh-workspace'
+
+    - id: session-projection-cache
+      name: '@deepseek-ai/dsh-session-projection-cache'
+      config:
+        writeEveryEvents: 200
+        writeIntervalMs: 5000
+
+    - id: tool-todo
+      name: '@deepseek-ai/dsh-tool-todo'
+
+    # The API gateway: the transport-agnostic dispatch face every client shape
+    # shares. provider/model are the host default routing — the profile json's
+    # mapping target (user config overrides these engineering defaults).
+    - id: directory-picker
+      name: '@deepseek-ai/dsh-host-directory-picker-browse'
+
+    - id: api-gateway
+      name: '@deepseek-ai/dsh-host-apiproxy'
+      config:
+        provider: deepseek
+        model: deepseek-v4-flash
+
+    # ── layer 2: transport/service ──────────────────────────────────────────────
+
+    # Plain route-registration carrier. distIndex is an assembly fact, not user
+    # config — AppCLIEntry resolves the frontend dist and patches it in; host and
+    # port arrive as CLI-flag patches over these defaults.
+    - id: webserver
+      name: '@deepseek-ai/dsh-host-webserver'
+      config:
+        host: 127.0.0.1
+        port: 3080
+
+    # ── browser plugin roster (dshClient rows; node halves are layer-2 hosts) ──
+
+    # Dual-face: node half scans this very tree for dshClient rows, composes
+    # window.__DSH_BOOT__, serves /plugins/<id>/client.js; browser half is the
+    # module table the shell kernel constructs before cordis exists (§4.7 —
+    # adopted as a plugin entry by the kernel, never fetched).
+    - id: modules
+      name: '@deepseek-ai/dsh-client-modules'
+
+    # Owns both ends of the web transport: node half binds the gateway to the
+    # webserver under /api; browser half is the fetch/SSE client.
+    - id: connection
+      name: '@deepseek-ai/dsh-client-connection'
+
+    - id: client-runtime
+      name: '@deepseek-ai/dsh-client-runtime'
+
+    - id: ui-theme
+      name: '@deepseek-ai/dsh-client-ui-theme'
+
+    - id: locale
+      name: '@deepseek-ai/dsh-client-locale'
+
+    - id: ui-layout
+      name: '@deepseek-ai/dsh-client-ui-layout'
+
+    - id: ui-sidebar
+      name: '@deepseek-ai/dsh-client-ui-sidebar'
+
+    - id: ui-settings
+      name: '@deepseek-ai/dsh-client-ui-settings'
+
+    - id: ui-settings-general
+      name: '@deepseek-ai/dsh-client-ui-settings-general'
+
+    - id: ui-models
+      name: '@deepseek-ai/dsh-client-ui-models'
+
+    - id: ui-conversation
+      name: '@deepseek-ai/dsh-client-ui-conversation'
+
+
+    - id: ui-workspace
+      name: '@deepseek-ai/dsh-client-ui-workspace'
+
+    # Input triggers: the '/' | '@' pipeline (ui-slash), the command surface over
+    # it (ui-command), and the two reference sources (ui-skill / ui-subagent).
+    - id: ui-slash
+      name: '@deepseek-ai/dsh-client-ui-slash'
+
+    - id: ui-command
+      name: '@deepseek-ai/dsh-client-ui-command'
+
+    - id: ui-skill
+      name: '@deepseek-ai/dsh-client-ui-skill'
+
+    - id: ui-subagent
+      name: '@deepseek-ai/dsh-client-ui-subagent'
+
+    # Goal surface: GoalBar in the input dock over the goal session projection.
+    - id: ui-goal
+      name: '@deepseek-ai/dsh-client-ui-goal'
+
+    # Model selection: the /model popupSelect + composer seat over session.models.
+    - id: ui-model
+      name: '@deepseek-ai/dsh-client-ui-model'
+
+    - id: ui-permission
+      name: '@deepseek-ai/dsh-client-ui-permission'
+
+    # Plan control: the composer plan seat over the plan projection + /plan channel.
+    - id: ui-plan
+      name: '@deepseek-ai/dsh-client-ui-plan'
+
+    - id: ui-question
+      name: '@deepseek-ai/dsh-client-ui-question'
+
+    - id: ui-trajectory
+      name: '@deepseek-ai/dsh-client-ui-trajectory'

apps/cli/cordis.yml

@@ -1,420 +0,0 @@
-# dsh web — the full web-shape composition: host runtime (layer 1), the
-# transport/service layer (layer 2), and the browser plugin roster (dshClient
-# rows the modules node half scans into window.__DSH_BOOT__). Row order
-# carries no load semantics (activation is service-availability driven); the
-# grouping below is for readers. `--dev` appends the dsh-client-hmr row in
-# code (AppCLIEntry) — prod and dev differ by exactly that one row.
-# AppCLIEntry patches this tree before boot: profile json + CLI flags +
-# distIndex land as config patches over the rows below (yaml = engineering
-# defaults, json = user config, user wins per field).
-
-# ── layer 1: runtime ────────────────────────────────────────────────────────
-
-- id: timer
-  name: '@cordisjs/plugin-timer'
-
-- id: llm
-  name: '@deepseek-ai/dsh-llm'
-
-- id: session
-  name: '@deepseek-ai/dsh-session'
-
-# Projection registry: drives every registered domain unit over committed
-# session events and serves finished values (history-tail projections block +
-# session/projection frames). Without this row every domain's optional unit
-# injection stays silent — no block, no frames, no titles/todos on the web.
-- id: session-projection
-  name: '@deepseek-ai/dsh-session-projection'
-
-- id: session-title
-  name: '@deepseek-ai/dsh-session-title'
-  config:
-    fallbackMaxWords: 5
-    fallbackMaxBytes: 40
-    maxTitleBytes: 80
-
-# Model-made titles on the first-message cadence (the web sidebar renders
-# session/title). Same values as the TUI composition.
-- id: session-title-llm
-  name: '@deepseek-ai/dsh-session-title-first-message-llm'
-  config:
-    targetWords: 5
-    targetCjkCharacters: 10
-    maxInputBytes: 4096
-    maxOutputTokens: 64
-    timeoutMs: 60000
-
-- id: system-prompt
-  name: '@deepseek-ai/dsh-system-prompt'
-  config:
-    persona: ''
-
-- id: tools
-  name: '@deepseek-ai/dsh-tools'
-  config:
-    # TEMPORARY workaround: DSH_TOOLS_MODE (native|code|both) opts a whole dsh
-    # process into Code Mode while per-session tool-mode selection is being
-    # designed; unset keeps the schema default (native). Remove the env seam
-    # once the web UI owns the choice per session.
-    mode: !!js process.env.DSH_TOOLS_MODE
-
-# Code Mode substrate for the row above. Mounted unconditionally because
-# Loader metadata is static (no conditional rows): a native-mode boot only
-# registers the service — a worker thread spawns per run_code execution.
-- id: code-runtime
-  name: '@deepseek-ai/dsh-code-runtime-worker'
-
-- id: user-interaction
-  name: '@deepseek-ai/dsh-user-interaction'
-
-- id: agent
-  name: '@deepseek-ai/dsh-agent'
-
-- id: tasks
-  name: '@deepseek-ai/dsh-tasks-local'
-
-- id: agent-loop
-  name: '@deepseek-ai/dsh-agent-loop'
-  config:
-    agents: []
-
-# The native DeepSeek adapter; reads the key/base-url the boot's layered
-# .env loading (cwd then $DSH_HOME) left in the environment.
-- id: llm-deepseek
-  name: '@deepseek-ai/dsh-llm-deepseek'
-  config:
-    apiKey: !!js process.env.DEEPSEEK_API_KEY
-    baseURL: !!js process.env.DEEPSEEK_BASE_URL
-
-# Common pi-ai provider routes read credentials and endpoint overrides from the
-# boot's layered environment.
-- id: llm-pi-ai
-  name: '@deepseek-ai/dsh-llm-pi-ai'
-  config:
-    providers:
-      - provider: openai
-        apiKey: !!js process.env.OPENAI_API_KEY
-        baseURL: !!js process.env.OPENAI_BASE_URL
-      - provider: anthropic
-        apiKey: !!js process.env.ANTHROPIC_API_KEY
-        baseURL: !!js process.env.ANTHROPIC_BASE_URL
-
-# Transient-failure recovery around the loop's model calls (same policy as
-# the TUI's agent-spine composition; defaults: 2 retries, 500ms→10s backoff).
-- id: llm-retry
-  name: '@deepseek-ai/dsh-llm-retry'
-
-# Session store root. AppCLIEntry resolves the engineering default to a
-# global dir under the Harness home ($DSH_HOME, else ~/.dsh): sessions live
-# in one place across every cwd, not a project-local ./.sessions. The
-# persistenceRoot profile key (user config) still overrides this per field.
-- id: session-persistence-jsonl
-  name: '@deepseek-ai/dsh-session-persistence-jsonl'
-  config:
-    root: './.sessions'
-
-- id: storage
-  name: '@deepseek-ai/dsh-storage'
-
-- id: storage-json
-  name: '@deepseek-ai/dsh-storage-json'
-  config:
-    root: './.storages'
-
-- id: storage-domain
-  name: '@deepseek-ai/dsh-storage-domain'
-  config:
-    backend: json
-
-- id: workspace
-  name: '@deepseek-ai/dsh-workspace'
-
-# Persisted projection cache: durable per-session checkpoints of every
-# registered projection unit (json backend → ./.storages/session_projcache.json,
-# beside workspace.json), throttled between the two mandatory points
-# (turn/end + detach), serving cold listings without full-log loads.
-- id: session-projection-cache
-  name: '@deepseek-ai/dsh-session-projection-cache'
-  config:
-    writeEveryEvents: 200
-    writeIntervalMs: 5000
-
-# Managed child-process groups for the bash executor (spawn/kill/output plumbing).
-- id: subprocess
-  name: '@deepseek-ai/dsh-subprocess-local'
-
-# The sandboxed product path (the acp-agent composition): per-platform
-# runner provider, the shared policy home, the confined bash executor, and
-# the approval seam its escalation asks through. The web deployment default
-# is danger-full-access + never (same behavior as the former bash-local
-# rows); DSH_PERMISSION_MODE opts a process into a confined default, and
-# per-session switches ride the /permission command's knob events.
-- id: sandbox
-  name: '@deepseek-ai/dsh-sandbox-local'
-
-- id: sandbox-policy
-  name: '@deepseek-ai/dsh-sandbox-policy'
-  config:
-    mode: !!js process.env.DSH_PERMISSION_MODE ?? 'danger-full-access'
-    workspaceRoot: !!js process.cwd()
-
-- id: bash-sandbox
-  name: '@deepseek-ai/dsh-bash-sandbox'
-
-- id: approval
-  name: '@deepseek-ai/dsh-user-approval'
-  config:
-    policy: !!js "(process.env.DSH_PERMISSION_MODE ?? 'danger-full-access') === 'danger-full-access' ? 'never' : 'ask'"
-
-# Presets over the two knobs (requires the confining executor + approval):
-# the web permission chip's table, served through the permissions projection
-# and switched through /permission.
-- id: permission
-  name: '@deepseek-ai/dsh-permission'
-  config:
-    presets:
-      read-only:
-        sandbox: read-only
-        approval: ask
-      workspace-write:
-        sandbox: workspace-write
-        approval: ask
-      danger-full-access:
-        sandbox: danger-full-access
-        approval: never
-
-- id: tool-bash
-  name: '@deepseek-ai/dsh-tool-bash'
-
-- id: tool-todo
-  name: '@deepseek-ai/dsh-tool-todo'
-
-- id: tool-tasks
-  name: '@deepseek-ai/dsh-tool-tasks'
-
-# fs cwd stays the package default (process.cwd()) — the same value the
-# gateway injects into session.cwd, so paths and sessions agree. The
-# sandboxed backend rides the SAME policy as bash: write/edit fence by the
-# effective mode, so read/write/edit stay available under every mode.
-- id: fs-sandbox
-  name: '@deepseek-ai/dsh-fs-sandbox'
-
-- id: fs-policy
-  name: '@deepseek-ai/dsh-fs-policy'
-
-- id: tool-fs
-  name: '@deepseek-ai/dsh-tool-fs'
-
-- id: tool-fs-search
-  name: '@deepseek-ai/dsh-tool-fs-search'
-
-- id: workspace-context
-  name: '@deepseek-ai/dsh-workspace-context'
-  config:
-    maxBytes: 65536
-
-- id: skill
-  name: '@deepseek-ai/dsh-skill'
-
-- id: skill-local
-  name: '@deepseek-ai/dsh-skill-local'
-
-- id: tool-skill
-  name: '@deepseek-ai/dsh-tool-skill'
-
-# Host command registry: the single source of truth behind command.list /
-# command.execute; the web '/' menu is a pure projection of this registry.
-- id: commands
-  name: '@deepseek-ai/dsh-commands'
-
-# Goal service + automatic same-session continuation + the /goal command.
-# The GoalService registers the 'goal' session projection unit; the web
-# GoalBar reads it through useProjection.
-- id: goal
-  name: '@deepseek-ai/dsh-goal'
-
-- id: goal-session
-  name: '@deepseek-ai/dsh-goal-session'
-
-- id: command-goal
-  name: '@deepseek-ai/dsh-command-goal'
-
-# Plan mode registers /plan (the first real command on the web surface).
-# Section text mirrors examples/tui-agent/cordis.yml (the reference
-# deployment); plan-mode throws at load on an empty section.
-- id: plan-mode
-  name: '@deepseek-ai/dsh-plan-mode'
-  config:
-    section: |
-          You are in plan mode. Stay in plan mode until exit_plan_mode succeeds or the user switches the session mode. Imperative language to implement changes means plan the implementation, not execute it. A user's conversational agreement — including an answer confirming something you asked — approves nothing and does not end plan mode; fold the confirmed decision into the plan and submit it through exit_plan_mode.
-
-          Explore first. Use non-mutating reads, searches, static analysis, and checks to ground the plan in the actual repository. Do not edit or write files, change configuration, run formatters or code generation that rewrites tracked files, commit, or otherwise carry out the plan. Prefer existing functions and patterns over new machinery.
-
-          The tool catalog stays the same across modes for request-cache stability. These plan-mode rules override any later tool description or guidance that suggests using mutation tools; those tools remain listed only to keep the request shape stable. Do not use todo_write to track this planning phase: it tracks implementation after an approved plan, while the plan itself belongs in exit_plan_mode.
-
-          Resolve discoverable facts by inspection. Use ask_user_question only for user-owned choices or material ambiguity that inspection cannot answer. Do not ask the user where code lives or how current behavior works when you can find out.
-
-          Make the plan decision-complete: state the goal and success criteria; group implementation changes by subsystem; identify public API, schema, and data-flow changes; cover edge cases, failure modes, tests, acceptance criteria, and explicit assumptions. Keep it concise enough to review but detailed enough that another engineer can implement it without making design decisions.
-
-          When ready, call exit_plan_mode with the complete plan markdown, starting with a # title. Make exit_plan_mode the only and final tool call in that assistant response: it presents the plan for approval, and implementation begins only in a later step after approval. Do not paste the final plan as a plain reply or ask "should I proceed?" through prose or ask_user_question. If review rejects it, incorporate the feedback and present again. If the review channel is unavailable or aborted, stay in plan mode and ask the user to switch modes manually; do not proceed with implementation.
-
-# token-meter rejects unknown config keys — keep this row bare.
-- id: token-meter
-  name: '@deepseek-ai/dsh-token-meter'
-
-- id: compact-basic
-  name: '@deepseek-ai/dsh-compact-basic'
-
-- id: subagent
-  name: '@deepseek-ai/dsh-subagent'
-
-- id: subagent-spawn
-  name: '@deepseek-ai/dsh-subagent-spawn'
-  config:
-    providerName: spawn
-
-- id: subagent-fork
-  name: '@deepseek-ai/dsh-subagent-fork'
-  config:
-    providerName: fork
-
-- id: tool-subagent
-  name: '@deepseek-ai/dsh-tool-subagent'
-  config:
-    provider: spawn
-    toolName: subagent
-
-- id: tool-subagent-fork
-  name: '@deepseek-ai/dsh-tool-subagent'
-  config:
-    provider: fork
-    toolName: subagent_fork
-
-- id: workflow-workerthread
-  name: '@deepseek-ai/dsh-workflow-workerthread'
-  config:
-    provider: spawn
-
-- id: tool-workflow
-  name: '@deepseek-ai/dsh-tool-workflow'
-
-- id: timeout-policy
-  name: '@deepseek-ai/dsh-timeout-policy'
-
-- id: spill-local
-  name: '@deepseek-ai/dsh-spill-local'
-
-# Omitting maxInlineBytes makes the whole policy a silent no-op — always
-# state it explicitly.
-- id: spill-policy
-  name: '@deepseek-ai/dsh-spill-policy'
-  config:
-    maxInlineBytes: 50000
-
-# The API gateway: the transport-agnostic dispatch face every client shape
-# shares. provider/model are the host default routing — the profile json's
-# mapping target (user config overrides these engineering defaults).
-# Directory-picking package, dual-face: the node half serves the gateway's
-# host.* picker RPCs, the browser half fills ui-workspace's directory-flow
-# slots — one row composes the whole interaction. Swap point: mount
-# '-native' instead for the host-display OS chooser.
-- id: directory-picker
-  name: '@deepseek-ai/dsh-host-directory-picker-browse'
-
-- id: api-gateway
-  name: '@deepseek-ai/dsh-host-apiproxy'
-  config:
-    provider: deepseek
-    model: deepseek-v4-flash
-
-# ── layer 2: transport/service ──────────────────────────────────────────────
-
-# Plain route-registration carrier. distIndex is an assembly fact, not user
-# config — AppCLIEntry resolves the frontend dist and patches it in; host and
-# port arrive as CLI-flag patches over these defaults.
-- id: webserver
-  name: '@deepseek-ai/dsh-host-webserver'
-  config:
-    host: 127.0.0.1
-    port: 3080
-
-# ── browser plugin roster (dshClient rows; node halves are layer-2 hosts) ──
-
-# Dual-face: node half scans this very tree for dshClient rows, composes
-# window.__DSH_BOOT__, serves /plugins/<id>/client.js; browser half is the
-# module table the shell kernel constructs before cordis exists (§4.7 —
-# adopted as a plugin entry by the kernel, never fetched).
-- id: modules
-  name: '@deepseek-ai/dsh-client-modules'
-
-# Owns both ends of the web transport: node half binds the gateway to the
-# webserver under /api; browser half is the fetch/SSE client.
-- id: connection
-  name: '@deepseek-ai/dsh-client-connection'
-
-- id: client-runtime
-  name: '@deepseek-ai/dsh-client-runtime'
-
-- id: ui-theme
-  name: '@deepseek-ai/dsh-client-ui-theme'
-
-- id: locale
-  name: '@deepseek-ai/dsh-client-locale'
-
-- id: ui-layout
-  name: '@deepseek-ai/dsh-client-ui-layout'
-
-- id: ui-sidebar
-  name: '@deepseek-ai/dsh-client-ui-sidebar'
-
-- id: ui-settings
-  name: '@deepseek-ai/dsh-client-ui-settings'
-
-- id: ui-settings-general
-  name: '@deepseek-ai/dsh-client-ui-settings-general'
-
-- id: ui-models
-  name: '@deepseek-ai/dsh-client-ui-models'
-
-- id: ui-conversation
-  name: '@deepseek-ai/dsh-client-ui-conversation'
-
-
-- id: ui-workspace
-  name: '@deepseek-ai/dsh-client-ui-workspace'
-
-# Input triggers: the '/' | '@' pipeline (ui-slash), the command surface over
-# it (ui-command), and the two reference sources (ui-skill / ui-subagent).
-- id: ui-slash
-  name: '@deepseek-ai/dsh-client-ui-slash'
-
-- id: ui-command
-  name: '@deepseek-ai/dsh-client-ui-command'
-
-- id: ui-skill
-  name: '@deepseek-ai/dsh-client-ui-skill'
-
-- id: ui-subagent
-  name: '@deepseek-ai/dsh-client-ui-subagent'
-
-# Goal surface: GoalBar in the input dock over the goal session projection.
-- id: ui-goal
-  name: '@deepseek-ai/dsh-client-ui-goal'
-
-# Model selection: the /model popupSelect + composer seat over session.models.
-- id: ui-model
-  name: '@deepseek-ai/dsh-client-ui-model'
-
-# The /permission popup picker (hostBacked over the host /permission command).
-- id: ui-permission
-  name: '@deepseek-ai/dsh-client-ui-permission'
-
-# Plan control: the composer plan seat over the plan projection + /plan channel.
-- id: ui-plan
-  name: '@deepseek-ai/dsh-client-ui-plan'
-
-- id: ui-question
-  name: '@deepseek-ai/dsh-client-ui-question'
-
-- id: ui-trajectory
-  name: '@deepseek-ai/dsh-client-ui-trajectory'

apps/cli/package.json

@@ -9,17 +9,19 @@
   },
   "files": [
     "lib/bin.js",
-    "cordis.yml",
+    "config",
     "src"
   ],
   "license": "BSD-3-Clause",
   "dependencies": {
+    "@cordisjs/plugin-hmr": "workspace:*",
     "@cordisjs/plugin-include": "workspace:*",
     "@cordisjs/plugin-loader": "workspace:*",
     "@cordisjs/plugin-timer": "workspace:*",
     "@deepseek-ai/dsh-agent": "workspace:^",
     "@deepseek-ai/dsh-agent-loop": "workspace:^",
     "@deepseek-ai/dsh-app-boot": "workspace:^",
+    "@deepseek-ai/dsh-bash-local": "workspace:^",
     "@deepseek-ai/dsh-bash-sandbox": "workspace:^",
     "@deepseek-ai/dsh-client-connection": "workspace:^",
     "@deepseek-ai/dsh-client-hmr": "workspace:^",
@@ -48,15 +50,17 @@
     "@deepseek-ai/dsh-command-goal": "workspace:^",
     "@deepseek-ai/dsh-commands": "workspace:^",
     "@deepseek-ai/dsh-compact-basic": "workspace:^",
+    "@deepseek-ai/dsh-compact-tool-result-prune": "workspace:^",
     "@deepseek-ai/dsh-frontend": "workspace:^",
+    "@deepseek-ai/dsh-fs-local": "workspace:^",
     "@deepseek-ai/dsh-fs-policy": "workspace:^",
     "@deepseek-ai/dsh-fs-sandbox": "workspace:^",
     "@deepseek-ai/dsh-goal": "workspace:^",
     "@deepseek-ai/dsh-goal-session": "workspace:^",
     "@deepseek-ai/dsh-host-apiproxy": "workspace:^",
     "@deepseek-ai/dsh-host-directory-picker-browse": "workspace:^",
-    "@deepseek-ai/dsh-host-directory-picker-native": "workspace:^",
     "@deepseek-ai/dsh-host-webserver": "workspace:^",
+    "@deepseek-ai/dsh-invariants": "workspace:^",
     "@deepseek-ai/dsh-llm": "workspace:^",
     "@deepseek-ai/dsh-llm-deepseek": "workspace:^",
     "@deepseek-ai/dsh-llm-pi-ai": "workspace:^",
@@ -66,10 +70,15 @@
     "@deepseek-ai/dsh-plan-mode": "workspace:^",
     "@deepseek-ai/dsh-sandbox-local": "workspace:^",
     "@deepseek-ai/dsh-sandbox-policy": "workspace:^",
+    "@deepseek-ai/dsh-scope": "workspace:^",
+    "@deepseek-ai/dsh-tool-cordis": "workspace:^",
     "@deepseek-ai/dsh-session": "workspace:^",
+    "@deepseek-ai/dsh-session-checkpoint-policy": "workspace:^",
     "@deepseek-ai/dsh-session-persistence-jsonl": "workspace:^",
     "@deepseek-ai/dsh-session-projection": "workspace:^",
     "@deepseek-ai/dsh-session-projection-cache": "workspace:^",
+    "@deepseek-ai/dsh-session-query-sqlite": "workspace:^",
+    "@deepseek-ai/dsh-session-reference": "workspace:^",
     "@deepseek-ai/dsh-session-title": "workspace:^",
     "@deepseek-ai/dsh-session-title-first-message-llm": "workspace:^",
     "@deepseek-ai/dsh-skill": "workspace:^",
@@ -87,9 +96,12 @@
     "@deepseek-ai/dsh-tasks-local": "workspace:^",
     "@deepseek-ai/dsh-timeout-policy": "workspace:^",
     "@deepseek-ai/dsh-token-meter": "workspace:^",
+    "@deepseek-ai/dsh-tool-ask-user": "workspace:^",
     "@deepseek-ai/dsh-tool-bash": "workspace:^",
     "@deepseek-ai/dsh-tool-fs": "workspace:^",
     "@deepseek-ai/dsh-tool-fs-search": "workspace:^",
+    "@deepseek-ai/dsh-tool-goal": "workspace:^",
+    "@deepseek-ai/dsh-tool-ralph": "workspace:^",
     "@deepseek-ai/dsh-tool-skill": "workspace:^",
     "@deepseek-ai/dsh-tool-subagent": "workspace:^",
     "@deepseek-ai/dsh-tool-tasks": "workspace:^",
@@ -107,6 +119,7 @@
     "js-yaml": "^4.2.0"
   },
   "devDependencies": {
-    "@types/js-yaml": "^4.0.9"
+    "@types/js-yaml": "^4.0.9",
+    "node-pty": "1.1.0"
   }
 }

apps/cli/src/app-cli-entry.ts

@@ -1,8 +1,8 @@
 /**
  * AppCLIEntry — the pre-cordis boot glue the config-tree dsh surfaces share
- * (`dsh web` and `dsh -p` boot the one composition; TUI migrates later).
+ * for the Web/headless surface.
  * Everything here is what must exist before the Loader runs: layered env,
- * the patch composition over the shipped cordis.yml (profile json + CLI
+ * the patch composition over the shipped base and surface overlay (profile json + CLI
  * flags + the resolved frontend dist), and the fail-loud triple after the
  * tree settles.
  */
@@ -11,13 +11,10 @@ import { readFileSync } from 'node:fs'
 import { createRequire } from 'node:module'
 import { networkInterfaces } from 'node:os'
 import { join, resolve } from 'node:path'
-import { pathToFileURL } from 'node:url'
 import { Context } from 'cordis'
-import type { FiberState } from 'cordis'
-import Loader from '@cordisjs/plugin-loader'
-import Include, { type PatchOptions } from '@cordisjs/plugin-include'
+import type { PatchOptions } from '@cordisjs/plugin-include'
 import yaml from 'js-yaml'
-import { assertEntriesLoaded, installFailLoud, loadEnv } from '@deepseek-ai/dsh-app-boot'
+import { boot, installFailLoud, loadEnv, loadOverlayPatches, loadPersonalPatches } from '@deepseek-ai/dsh-app-boot'
 import { resolveDshHome } from '@deepseek-ai/dsh-paths'
 // Empty type import carries the httpServer Context merge for the port read below.
 import type {} from '@deepseek-ai/dsh-host-webserver'
@@ -90,18 +87,23 @@ const jsExprType = new yaml.Type('tag:yaml.org,2002:js', {
 })
 const includeYamlSchema = yaml.JSON_SCHEMA.extend(jsExprType)
 
-/**
- * Value mirror of cordis's `FiberState` const enum members the sweep needs
- * (a const enum has no runtime object to import; same rationale as the
- * client-side mirror in dsh-client-web).
- */
-const FIBER_ACTIVE = 2 as FiberState.ACTIVE
-const FIBER_PENDING = 0 as FiberState.PENDING
-
 /** Constructor facts for one dsh invocation over the shared composition (argv already parsed by the surface bin). */
 export interface AppCLIEntryOptions {
-  /** Absolute path of the shipped cordis.yml. */
+  /** Absolute path of the shared base config the Loader includes. */
   configPath: string
+  /**
+   * Absolute path of this surface's overlay: a patch list applied over
+   * {@link configPath} before this entry's own profile/flag patches. Its rows
+   * are also merge inputs, so a flag override preserves the overlay's other
+   * fields on the same row.
+   */
+  overlayPath: string
+  /**
+   * Optional explicit overlay applied after {@link overlayPath} and before
+   * this entry's own profile/flag patches. When absent, the personal
+   * `$DSH_HOME/config.yaml` overlay is applied instead.
+   */
+  extraOverlayPath?: string
   /** Whether to append the HMR row (the whole prod/dev difference; web surface only). */
   dev: boolean
   /** --host when explicitly passed; undefined keeps the yml engineering default. */
@@ -163,10 +165,8 @@ export class AppCLIEntry {
   }
 
   /**
-   * Compose the patch set from the non-yml config sources: computed
-   * engineering defaults (the global session root), profile json (user
-   * config, overriding those defaults), CLI flags, and the resolved frontend
-   * dist. Patches replace a row's config wholesale, so each patched row's yml
+   * Compose the patch set from profile json, CLI flags, and the resolved
+   * frontend dist. Patches replace a row's config wholesale, so each patched row's yml
    * static values are re-read here (bypass parse) and merged under the overrides.
    */
   private composePatches(): void {
@@ -178,12 +178,6 @@ export class AppCLIEntry {
       overrides.set(entryId, bag)
     }
 
-    // Source 0: computed engineering defaults. The session store defaults to
-    // a global dir under the Harness home ($DSH_HOME, else ~/.dsh) so history
-    // is shared across every cwd, not a project-local ./.sessions. The profile
-    // (Source 1) overwrites this same field via last-write-wins in put().
-    put('session-persistence-jsonl', 'root', join(resolveDshHome(), 'sessions'))
-
     // Source 1: profile json (missing file = empty; unmapped key = loud).
     for (const [key, value] of Object.entries(this.readProfile())) {
       const mapping = PROFILE_MAPPINGS.find(m => m.jsonPath === key)
@@ -216,62 +210,62 @@ export class AppCLIEntry {
     })
   }
 
-  /** Loader include boot; the dev HMR row mounts before await so the fail-loud triple covers it. */
+  /** Shared Loader boot; the dev HMR row mounts before await so the fail-loud sweep covers it. */
   private async bootTree(): Promise<void> {
-    const ctx = new Context()
-    ctx.baseUrl = pathToFileURL(join(resolve(this.options.configPath), '..')).href + '/'
-    await ctx.plugin(Loader)
-    ctx.loader.builtins.include = Include
-    await ctx.loader.create({
-      name: 'cordis:include',
-      config: {
-        path: pathToFileURL(resolve(this.options.configPath)).href,
-        ...this.patches.length > 0 ? { patches: this.patches } : {},
-      },
+    // One include of the shared base with every overlay as a sibling patch
+    // list: patches never cross an include boundary, so nesting them would
+    // silently stop reaching base rows. The surface overlay applies first, then
+    // this entry's profile-json and CLI-flag patches, which therefore win.
+    const patches = [
+      ...loadOverlayPatches('dsh', this.options.overlayPath),
+      ...this.options.extraOverlayPath === undefined
+        ? loadPersonalPatches('dsh') ?? []
+        : loadOverlayPatches('dsh', this.options.extraOverlayPath),
+      ...this.patches,
+    ]
+    this.ctx = await boot('dsh', resolve(this.options.configPath), patches, async (ctx) => {
+      if (this.options.dev) await ctx.loader.create({ name: '@deepseek-ai/dsh-client-hmr' })
     })
-    if (this.options.dev) {
-      await ctx.loader.create({ name: '@deepseek-ai/dsh-client-hmr' })
-    }
-    this.ctx = ctx
-    await ctx.loader.await()
   }
 
-  /**
-   * Fail-loud triple: assertEntriesLoaded catches import failures,
-   * installFailLoud catches late apply rejections, and the all-ACTIVE sweep
-   * below catches PENDING fibers (cordis inject waiting has no timeout).
-   */
+  /** Install the diagnostic for plugin rejections that happen after settled boot. */
   private assertBoot(): void {
     installFailLoud('dsh')
-    assertEntriesLoaded(this.ctx, 'dsh')
-    const failures: string[] = []
-    for (const entry of this.ctx.loader.entries()) {
-      if (entry.fiber === undefined || entry.disabled) continue
-      const state = entry.fiber.state
-      if (state === FIBER_ACTIVE) continue
-      if (state === FIBER_PENDING) {
-        const missing = Object.keys(entry.fiber.inject).filter(service => this.ctx.get(service) === undefined)
-        failures.push(`${entry.options.name}: pending (waiting for service${missing.length === 1 ? '' : 's'}: ${missing.join(', ') || 'unknown'})`)
-      } else {
-        failures.push(`${entry.options.name}: fiber state ${String(state)}`)
-      }
-    }
-    if (failures.length > 0) {
-      throw new Error(`dsh: ${String(failures.length)} entr${failures.length === 1 ? 'y' : 'ies'} did not activate\n${failures.join('\n')}`)
-    }
   }
 
-  /** Bypass parse of the shipped yml (id → row) for patch-merge inputs; Loader still reads the file itself. */
+  /**
+   * Bypass parse of the base and this surface's overlay (id → row) for
+   * patch-merge inputs; the Loader still reads both files itself. The overlay
+   * wins per row, matching the order its patches are applied in, and its
+   * `insert` rows are indexed too because a flag may target one of them.
+   */
   private parseYmlRows(): Map<string, { config?: unknown }> {
-    const doc = yaml.load(readFileSync(this.options.configPath, 'utf8'), { schema: includeYamlSchema })
-    if (!Array.isArray(doc)) throw new Error(`dsh: ${this.options.configPath} is not a top-level entry list`)
     const rows = new Map<string, { config?: unknown }>()
-    for (const row of doc as { id?: string; config?: unknown }[]) {
-      if (typeof row.id === 'string') rows.set(row.id, row)
+    const files = [this.options.configPath, this.options.overlayPath]
+    if (this.options.extraOverlayPath !== undefined) files.push(this.options.extraOverlayPath)
+    for (const file of files) {
+      for (const row of this.parseRowList(file)) {
+        if (typeof row.id === 'string') rows.set(row.id, row)
+        for (const inserted of row.insert ?? []) {
+          if (typeof inserted.id === 'string') rows.set(inserted.id, inserted)
+        }
+      }
     }
     return rows
   }
 
+  /**
+   * Parse one entry or patch list, rejecting anything that is not a top-level
+   * array so a malformed file fails here rather than at row lookup.
+   * @param file - absolute path of the config or overlay file.
+   * @returns the parsed top-level entries.
+   */
+  private parseRowList(file: string): { id?: string; config?: unknown; insert?: { id?: string; config?: unknown }[] }[] {
+    const doc = yaml.load(readFileSync(file, 'utf8'), { schema: includeYamlSchema })
+    if (!Array.isArray(doc)) throw new Error(`dsh: ${file} is not a top-level entry list`)
+    return doc as { id?: string; config?: unknown; insert?: { id?: string; config?: unknown }[] }[]
+  }
+
   /** Profile json under cwd; read-only — never created here, absent = no user config. */
   private readProfile(): Record<string, unknown> {
     let raw: string

apps/cli/src/args.ts

@@ -2,19 +2,24 @@
  * Commander adapter for the `dsh` command-line entry: the one place argv is
  * parsed and routed to a mode. `bin.ts` switches on the returned discriminant
  * and dynamic-imports that mode's module. One program: the default (no
- * subcommand) is the TUI/headless surface with option-only flags; `web` is a
- * real subcommand. Commander owns `--help`/`--version` and parse errors — it
- * prints and exits at the point of failure (a domain failure routes through
+ * subcommand) is the TUI/headless surface with option-only flags; `meta` and
+ * `web` are real subcommands. Commander owns `--help`/`--version` and parse
+ * errors — it prints and exits at the point of failure (a domain failure routes through
  * `command.error`), so this returns only a resolved mode.
  * @module @deepseek-ai/dsh/args
  */
 
 import { Command, CommanderError } from 'commander'
 
-/** Interactive TUI: the default mode. `--config` swaps the tree; `--resume <id>` rehydrates a session. */
+/**
+ * Interactive TUI: the default mode. `--config` applies an overlay over the
+ * shipped composition in place of the personal one, `--config-replace` boots a
+ * file as the whole tree instead, and `--resume <id>` rehydrates a session.
+ */
 interface TuiInvocation {
   mode: 'tui'
   config?: string
+  configReplace?: string
   resume?: string
 }
 
@@ -24,18 +29,35 @@ interface HeadlessInvocation {
   prompt: string
 }
 
+/** Interactive fresh TUI over this harness checkout; accepts no default-surface options. */
+interface MetaInvocation {
+  mode: 'meta'
+}
+
+/**
+ * Guided fresh-session entry: `dsh upgrade` seeds the first turn with the
+ * `dsh-upgrade` skill. It always mints a
+ * fresh session in the invoking directory and takes no options — `--resume`,
+ * `--config`, and `-p` are rejected as mistyped, so there is nothing to carry.
+ */
+interface SkillSessionInvocation {
+  mode: 'upgrade'
+}
+
 /**
  * Browser UI: `dsh web`. `host`/`port` are present only when the flag was
  * passed — pass-through overrides with no CLI default and no CLI validation:
  * the `dsh-host-webserver` schema (`host` a loopback/all-interfaces literal,
  * `port` a natural ≤ 65535) is the single source of both the default (the
- * shipped `cordis.yml` value stands when a flag is absent) and validity (a bad
+ * shipped Web overlay value stands when a flag is absent) and validity (a bad
  * value fails loud at boot). `port` is `Number`-coerced only because the schema
  * wants a number, not a string. `dev` mounts the client HMR driver;
  * `workspaceRoot` is the parent directory for name-created workspaces.
  */
 interface WebInvocation {
   mode: 'web'
+  /** Overlay of loader patches applied over the shipped web composition. */
+  config?: string
   host?: string
   port?: number
   dev: boolean
@@ -45,10 +67,16 @@ interface WebInvocation {
 }
 
 /** The resolved `dsh` invocation: exactly one mode. `--help`/`--version`/errors exit inside {@link parseDshArgs}. */
-export type DshInvocation = TuiInvocation | HeadlessInvocation | WebInvocation
+export type DshInvocation =
+  | TuiInvocation
+  | HeadlessInvocation
+  | MetaInvocation
+  | SkillSessionInvocation
+  | WebInvocation
 
 /** Raw web-subcommand options straight from Commander. */
 interface WebOptions {
+  config?: string
   host?: string
   port?: string
   dev?: boolean
@@ -65,6 +93,7 @@ interface WebOptions {
 function resolveWeb(options: WebOptions): WebInvocation {
   return {
     mode: 'web',
+    ...options.config !== undefined && { config: options.config },
     ...options.host !== undefined && { host: options.host },
     ...options.port !== undefined && { port: Number(options.port) },
     dev: options.dev === true,
@@ -86,21 +115,34 @@ export function parseDshArgs(argv: readonly string[], version: string): DshInvoc
   const program = new Command()
     .name('dsh')
     .version(version, '-V, --version', 'output the version number')
-    .description('dsh: interactive TUI (default), headless task, and browser UI')
+    .description('dsh: DeepSeek Harness — an interactive coding agent for your terminal.\nRun `dsh` with no arguments to start a session in the current directory.')
+    // The default surface takes no positional task, so `dsh "task"` fails
+    // commander's arity check with no hint; these examples are where a first
+    // reader learns the entry points and that a one-shot task rides `-p`.
+    .addHelpText('after', `
+Examples:
+  dsh                     start an interactive session in this directory
+  dsh -p "run the tests"  answer one task, print the result, and exit
+  dsh --resume <id>       continue a past session
+`)
     .exitOverride()
+    // Stop parent options at a subcommand boundary so `web --config` belongs to
+    // Web while `--config ... web` remains a leaked default-surface option.
+    .enablePositionalOptions()
     // Default surface: option-only (no positional), so `web` can be a real
     // subcommand without a positional collision.
-    .option('--config <path>', 'boot an alternate cordis.yml instead of the shipped tree (TUI mode)')
-    .option('-p, --prompt <task>', 'run one headless turn for this task, print the result, and exit')
-    .option('--resume <id>', 'resume the persisted session with this id (TUI mode)')
-    .action((options: { config?: string; prompt?: string; resume?: string }) => {
+    .option('-p, --prompt <task>', 'answer this task without the interactive UI, then exit')
+    .option('--resume <id>', 'continue a past session by id')
+    .option('--config <path>', 'apply this overlay of loader patches instead of the personal one')
+    .option('--config-replace <path>', 'boot this file as the entire tree, ignoring the shipped and personal configuration')
+    .action((options: { config?: string; configReplace?: string; prompt?: string; resume?: string }) => {
       if (options.prompt !== undefined) {
         // A headless prompt owns the invocation; an empty task has nothing to
         // run, and --config/--resume are TUI inputs that must not silently
         // vanish from a headless run.
         if (options.prompt === '') program.error('error: --prompt needs a task')
-        if (options.config !== undefined || options.resume !== undefined) {
-          program.error('error: --prompt takes no --config or --resume')
+        if (options.config !== undefined || options.configReplace !== undefined || options.resume !== undefined) {
+          program.error('error: --prompt takes no --config, --config-replace, or --resume')
         }
         resolved = { mode: 'headless', prompt: options.prompt }
         return
@@ -108,32 +150,67 @@ export function parseDshArgs(argv: readonly string[], version: string): DshInvoc
       // An empty --resume= id would silently start a fresh session downstream
       // (agent-loop treats '' as no-resume), so a mistyped resume must fail loud.
       if (options.resume === '') program.error('error: --resume needs a session id')
+      // The two config flags are mutually exclusive: one layers over the shipped
+      // tree, the other discards it, so accepting both would silently drop one.
+      if (options.config !== undefined && options.configReplace !== undefined) {
+        program.error('error: --config and --config-replace are mutually exclusive')
+      }
       resolved = {
         mode: 'tui',
         ...options.config !== undefined && { config: options.config },
+        ...options.configReplace !== undefined && { configReplace: options.configReplace },
         ...options.resume !== undefined && { resume: options.resume },
       }
     })
 
-  const web = program.command('web').description('serve the browser UI (host/port default to the shipped config)')
+  // Commander parses the parent (default-surface) options on either side of a
+  // subcommand into `program.opts()`. For a subcommand that shares none of them,
+  // a leaked config/prompt/resume option is a mistyped invocation that must fail
+  // loud rather than silently run and drop the input.
+  const rejectParentOptions = (command: string): void => {
+    const parent = program.opts<{ config?: string; configReplace?: string; prompt?: string; resume?: string }>()
+    if (parent.config !== undefined || parent.configReplace !== undefined
+      || parent.prompt !== undefined || parent.resume !== undefined) {
+      program.error(`error: ${command} takes none of --config, --config-replace, -p/--prompt, or --resume`)
+    }
+  }
+
+  // Registration order is the rendered help order, so daily use comes first
+  // and the harness-development surfaces (`web --dev`, `meta`) come last.
+  // `upgrade` is a guided fresh-session entry: it takes no options and always
+  // mints a fresh session, so nothing is left to carry.
+  program
+    .command('upgrade')
+    .description('update this dsh installation to the latest version')
+    .action(() => {
+      rejectParentOptions('upgrade')
+      resolved = { mode: 'upgrade' }
+    })
+
+  // Host and port name no default: the CLI passes neither through when the flag
+  // is absent, so the shipped Web overlay value stands and restating it here
+  // would duplicate a fact this file does not own.
+  const web = program.command('web').description('serve the browser UI on the configured host and port')
   web
-    .option('--host <host>', 'override the config bind host (127.0.0.1 or 0.0.0.0)')
-    .option('--port <port>', 'override the config listen port (0 requests an OS-assigned port)')
-    .option('--dev', 'mount the client HMR driver and watch plugin bundles for rebuilds')
-    .option('--workspace-root <path>', 'parent directory for name-created workspaces')
+    .option('--config <path>', 'apply this overlay of loader patches over the shipped configuration')
+    .option('--host <host>', 'bind host; pass 0.0.0.0 to reach it from another machine')
+    .option('--port <port>', 'listen port; pass 0 to let the OS pick a free one')
+    .option('--dev', 'developer mode: hot-reload the browser client')
+    .option('--workspace-root <path>', 'parent directory for workspaces created from the browser UI')
     .option('--trusted-host <authority...>', 'extra authority the /api browser-trust fence accepts (host or host:port; repeatable)')
     .action((options: WebOptions) => {
-      // Commander parses the parent (default-surface) options on either side of
-      // the subcommand into `program.opts()`. `web` shares none of them, so a
-      // leaked `--config`/`-p`/`--resume` is a mistyped invocation that must
-      // fail loud rather than silently start the web server and drop it.
-      const parent = program.opts<{ config?: string; prompt?: string; resume?: string }>()
-      if (parent.config !== undefined || parent.prompt !== undefined || parent.resume !== undefined) {
-        program.error('error: web takes none of --config, -p/--prompt, or --resume')
-      }
+      rejectParentOptions('web')
       resolved = resolveWeb(options)
     })
 
+  program
+    .command('meta')
+    .description('work on the dsh source that runs this command, from any directory')
+    .action(() => {
+      rejectParentOptions('meta')
+      resolved = { mode: 'meta' }
+    })
+
   try {
     program.parse(argv, { from: 'user' })
   } catch (error) {