docs: make AGENTS.md the canonical agent guide for Codex/Astra (#1742)

Codex reads AGENTS.md by default, not CLAUDE.md. Move project guidance to
AGENTS.md (trim/nest long validation notes under docs/AGENTS.md), leave
CLAUDE.md as a thin @AGENTS.md wrapper for Claude Code, and document the
Codex project_doc_max_bytes raise needed for the ~35KiB root file.

Co-authored-by: Colby McHenry <colbymchenry@users.noreply.github.com>

AGENTS.md

@@ -0,0 +1,283 @@
+# AGENTS.md
+
+Canonical project guidance for coding agents working in this repository (Codex/Astra, Claude Code via `@AGENTS.md`, Cursor, etc.).
+
+**Codex size note:** root `AGENTS.md` is intentionally kept near ~35 KiB (critical build/test/arch/retrieval rules retained). Longer validation methodology + the Excalidraw worked example live in `docs/AGENTS.md`. This environment sets `project_doc_max_bytes = 49152` so root (and root+`docs/` when cwd is under `docs/`) are not silently truncated at the 32 KiB default.
+
+## Project Overview
+
+CodeGraph is a local-first code intelligence library + CLI + MCP server. It parses any supported codebase with tree-sitter, stores symbols/edges/files in SQLite (FTS5), and exposes a knowledge graph to AI agents (Claude Code, Cursor, Codex CLI, opencode) over MCP. Per-project data lives in `.codegraph/`. Extraction is deterministic — derived from AST, not LLM-summarized.
+
+Distributed as `@colbymchenry/codegraph` on npm; same binary serves as installer, indexer, and MCP server.
+
+## Build, Test, Run
+
+```bash
+npm run build           # tsc + copy schema.sql and *.wasm + build the viewer into dist/; chmods dist/bin/codegraph.js
+npm run build:lib       # the viewer's components as @colbymchenry/codegraph-ui (ui/dist) — NOT part of `build`
+npm run dev             # tsc --watch
+npm run clean           # rm -rf dist
+
+npm test                # vitest run (all)
+npm run test:watch
+npm run test:eval       # only __tests__/evaluation/
+npm run eval            # build then run __tests__/evaluation/runner.ts via tsx
+
+npm run cli             # build then run the local dist binary
+
+# Single test file / pattern
+npx vitest run __tests__/installer-targets.test.ts
+npx vitest run __tests__/extraction.test.ts -t "TypeScript"
+```
+
+`copy-assets` (called from `build`) copies `src/db/schema.sql` and all `src/extraction/wasm/*.wasm` files into `dist/`. **Any new SQL or grammar wasm must be copied or it won't ship.**
+
+One other build step writes into `dist/` and is subject to the same rule: `build:ui` builds the
+browser viewer into `dist/viewer/` (never `dist/ui/` — that's the terminal ui).
+`scripts/check-ui-build.mjs` asserts both `dist/viewer/` and the copied grammars in
+`dist/extraction/wasm/` after every build and inside every release archive — the viewer's syntax
+highlighting reads a file with the same grammar the engine indexed it with, so a missing wasm is an
+unhighlighted screen as well as an extraction gap.
+
+`npm run build:lib` is separate and does NOT run as part of `npm run build`: it compiles the same
+`ui/src` tree a second way, with `svelte-package`, into `ui/dist` — the `@colbymchenry/codegraph-ui`
+component library the Pro app imports (task CG-61). `scripts/check-ui-package.mjs` then prunes the
+standalone app's shell out of it, resolves the extensionless import specifiers `svelte-package`
+leaves behind, and asserts the seam: nothing outside `lib/adapter.js` may reach the network. The
+package is **prepared, not published** — `ui/package.json` carries `"private": true` deliberately,
+and `scripts/pack-npm.sh` only packs a tarball when `CODEGRAPH_PACK_UI=1`.
+
+Tests run as **two vitest projects** (`vitest.workspace.mts`): `engine` (node) and `ui` (jsdom, the
+Svelte plugin, `resolve.conditions: ['browser']`) for the single `__tests__/ui-package.test.ts`.
+`npm test` still runs both. The split is not cosmetic — `browser` is a package-resolution
+condition, and applied globally it hands the engine's suites the browser builds of
+`web-tree-sitter` and friends. The root config (`vitest.config.mts`, `.mts` because the plugin is
+ESM-only and the repo is CJS) is the shared base; note that a workspace project **concatenates**
+the base's `include` with its own, which is why the `ui` project does not `extends` it.
+
+Node engines: `>=20.0.0 <25.0.0`. There is a hard exit on Node 25.x and below 20 (see `src/bin/node-version-check.ts`).
+
+## Architecture
+
+### Layered pipeline
+
+```
+files → ExtractionOrchestrator (tree-sitter) → DB (nodes/edges/files)
+              ↓
+       ReferenceResolver (imports, name-matching, framework patterns)
+              ↓
+       GraphQueryManager / GraphTraverser (callers, callees, impact)
+              ↓
+       ContextBuilder (markdown/JSON for AI consumption)
+```
+
+The public API surface is `src/index.ts` — the `CodeGraph` class wires all the layers and re-exports types. Library users only touch this file; the MCP server and CLI also drive it.
+
+### Module layout
+
+- `src/index.ts` — `CodeGraph` class: `init`/`open`/`close`, `indexAll`, `sync`, `searchNodes`, `getCallers`/`getCallees`, `getImpactRadius`, `buildContext`, `watch`/`unwatch`.
+- `src/db/` — `DatabaseConnection`, `QueryBuilder` (prepared statements), `schema.sql`, `sqlite-adapter.ts`. Backed by Node's built-in **`node:sqlite`** (`DatabaseSync`) — real SQLite with WAL + FTS5, exposed through a thin better-sqlite3-shaped adapter. The bundled runtime always ships Node ≥22.5, so `node:sqlite` is always available: **no native build step and no wasm fallback**. (Running from source needs Node ≥22.5.) `codegraph status` reports the live backend (`node-sqlite`, the sole backend).
+- `src/extraction/` — `ExtractionOrchestrator`, tree-sitter wrappers, per-language extractors under `languages/` (one file per language), plus standalone extractors for non-tree-sitter formats (`svelte-extractor.ts`, `vue-extractor.ts`, `liquid-extractor.ts`, `dfm-extractor.ts` for Delphi). `parse-worker.ts` runs heavy parsing off the main thread.
+- `src/resolution/` — `ReferenceResolver` orchestrates `import-resolver.ts` (with `path-aliases.ts` for tsconfig path aliases + cargo workspace member globs), `name-matcher.ts`, and `frameworks/` (Express, Laravel, Rails, FastAPI, Django, Flask, Spring, Gin, Axum, ASP.NET, Vapor, React Router, Next.js — `nextjs.ts`: pages and `route.ts` handlers from files, `router.push` / `redirect` / `NextResponse.redirect` as `navigates` edges, with `next-router-synthesizer.ts` for `<Link href>` — Expo Router, SvelteKit, Vue/Nuxt, Cargo workspaces). Frameworks emit `route` nodes and `references` edges. `callback-synthesizer.ts` holds the whole-graph synthesis passes (`SYNTH_PASSES`, merged in registry order — first-seen wins a duplicate pair) with the language gates; `tier-synthesizer.ts` is the cross-tier pass (a client's literal `fetch`/`axios` path onto its own route, a queue job onto its consumer, a bus / socket event onto its handler — `channel`, `tier`, `registeredAt` on every edge; registered before the in-process emitter pass so its more specific edge wins); `synth-utils.ts` has the helpers they share (`enclosingFn`, `enclosingValue`, `makeLineAt`). Express's `postExtract` composes `app.use('/prefix', router)` mounts onto a mounted file's route names, idempotently (the original path stays in `qualifiedName`).
+- `src/graph/` — `GraphTraverser` (BFS/DFS, impact radius, path finding) and `GraphQueryManager` (high-level queries), plus the shared query-time derivations more than one surface renders: `named-symbol-flow.ts` (the one path finder, behind `codegraph_explore`'s Flow section and the viewer's Flow strip), `dynamic-boundary-report.ts` (where the graph stops), `type-hierarchy.ts` (ancestors/subtypes and the implementation count explore prints and the viewer draws),
+  `dead-code.ts` (unreferenced symbols, and every reason a candidate is NOT claimed). A derivation that two callers render must live here, not in `ToolHandler` — two derivations eventually disagree.
+- `src/context/` — `ContextBuilder` + formatter for markdown/JSON output.
+- `src/search/` — full-text query parser and helpers for FTS5.
+- `src/sync/` — `FileWatcher` (native FSEvents/inotify/RDCW) with debounce + filter, and git-hook helpers.
+- `src/mcp/` — MCP server (`MCPServer`, `tools.ts`, `transport.ts`). `server-instructions.ts` is what the server returns in the MCP `initialize` response — keep it in sync with the user-facing tool guidance.
+- `src/installer/` — see below.
+- `src/bin/codegraph.ts` — CLI (commander). Subcommands: `install`, `init`, `uninit`, `index`, `sync`, `status`, `query`, `files`, `context`, `affected`, `serve --mcp`.
+- `src/ui/` — terminal UI (shimmer progress, worker).
+- `src/ui-server/` -- read-only JSON API for the `codegraph ui` browser viewer (`api/`: `node`, `flow`, `map`, `screens`, `steps`, `deadcode`, `trails`, `program`, ...) plus static server; Svelte viewer lives in `ui/` (see `docs/design/codegraph-ui-design-spec.md`). `screens`/`steps`/`program` share one fold (`via`/`when` via `graph/branch-guards.ts`); `api/effects.ts` curates calls that leave the index; `api/route-roots.ts` names where a route's code starts. Derivations rendered by more than one surface belong in `src/graph/`, not `ToolHandler`.
+
+### NodeKind / EdgeKind
+
+Defined in `src/types.ts`. Both extractors and resolvers must use these exact strings.
+
+- **NodeKind**: `file`, `module`, `class`, `struct`, `interface`, `trait`, `protocol`, `function`, `method`, `property`, `field`, `variable`, `constant`, `enum`, `enum_member`, `type_alias`, `namespace`, `parameter`, `import`, `export`, `route`, `component`, `union`.
+- **EdgeKind**: `contains`, `calls`, `imports`, `exports`, `extends`, `implements`, `references`, `type_of`, `returns`, `instantiates`, `overrides`, `decorates`.
+
+### Multi-agent installer
+
+`src/installer/` is the entry point for `codegraph install` (and the bare `codegraph`/`npx @colbymchenry/codegraph` invocation). Architecture:
+
+- `targets/registry.ts` lists every supported agent.
+- `targets/types.ts` defines the `AgentTarget` interface — adding a 5th agent (Continue, Zed, Windsurf…) is **one new file in `targets/` + one entry in `registry.ts`**. Each target owns its config-file location and MCP-server JSON/TOML/JSONC writing. (Targets no longer write an instructions file — see below.)
+- Current targets: `claude.ts`, `cursor.ts`, `codex.ts`, `opencode.ts`.
+- `targets/toml.ts` is a hand-rolled TOML serializer scoped to `[mcp_servers.codegraph]` (used by Codex). Sibling tables and `[[array_of_tables]]` are preserved verbatim. No new dependency.
+- opencode reads `opencode.jsonc` by default; the installer prefers existing `.jsonc`, falls back to `.json`, and creates `.jsonc` for greenfield installs. Edits are surgical via `jsonc-parser` so user comments and formatting survive install/re-install/uninstall round-trips.
+- `instructions-template.ts` no longer holds an instructions body — it exports only the `<!-- CODEGRAPH_START -->`/`<!-- CODEGRAPH_END -->` markers. The installer **stopped writing** a `## CodeGraph` block into each agent's instructions file (`CLAUDE.md` / `~/.codex/AGENTS.md` / `~/.config/opencode/AGENTS.md` / `~/.gemini/GEMINI.md` / `.cursor/rules/codegraph.mdc` / Kiro steering doc) because it duplicated the MCP `initialize` instructions verbatim (issue #529). Each target's `install` (self-heal on upgrade) and `uninstall` use the markers to **strip** a block a previous install left behind. `server-instructions.ts` is the single source of truth for agent-facing guidance.
+- All installer changes need matching coverage in `__tests__/installer-targets.test.ts` — there are ~47 parameterized contract tests covering install idempotency, sibling preservation, uninstall reverses install, byte-equal re-runs returning `unchanged`, and partial-state recovery for Codex.
+
+### Cursor MCP working-directory quirk
+
+Cursor launches MCP subprocesses with the wrong cwd and doesn't pass `rootUri` in `initialize`. The installer injects `--path` into Cursor's MCP args — absolute path for local installs, `${workspaceFolder}` for global installs. If you touch Cursor wiring, preserve this.
+
+### MCP server instructions
+
+`src/mcp/server-instructions.ts` is sent back to the agent in the MCP `initialize` response. This is the *first* thing every agent sees about how to use the tools, and as of issue #529 it is the **single source of truth** for agent-facing tool guidance — the installer no longer writes a duplicate `## CodeGraph` instructions block into `CLAUDE.md` / `AGENTS.md` / `.cursor/rules/codegraph.mdc`. Edit tool guidance here and nowhere else.
+
+## Retrieval performance & dynamic-dispatch coverage (do not regress)
+
+CodeGraph's core value is letting an agent answer **structural/flow** questions ("how does X reach Y", trace, impact, callers) with a few **fast** codegraph calls and **zero Read/Grep**. The optimization target is **wall-clock latency + tool-call count** — *don't optimize for token cost*. (Cost is **lower**, not "flat" as earlier framing claimed: a current-build with-vs-without A/B across the 7 README repos, median of 4, saved on average **35% cost · 57% tokens · 46% time · 71% tool calls** — reproducing the published README. The mechanism is **far fewer turns over a much smaller accumulated context** — NOT cache-ability: the without-arm's huge token volume is *mostly* cheap cache-reads, which is why token-count savings (57%) look bigger than cost savings (35%). Measure tokens by **summing per-turn assistant usage**, not `result.usage` (last-turn only in current Claude Code). See `docs/benchmarks/call-sequence-analysis.md`.) The mechanism that drives everything here: **an agent falls back to Read/Grep the instant a codegraph answer is insufficient.** So every change is judged by one question — is codegraph's answer sufficient enough to *stop* the agent from reading?
+
+**Target behavior:** a flow question resolves in **1 codegraph call on small repos, scaling to 3–5 on large**, with **Read/Grep = 0**. When reviewing a PR or trying something new, do not regress this.
+
+### Adapt the tool to the agent — don't try to change the agent
+
+The lever that decides whether a retrieval change lands. **Test before building anything here: does this make a tool the agent _already calls_ do more with the input it _already gives_? If it instead needs the agent to behave differently — pick a different tool, query differently, learn from examples — it hits the low-salience wall and won't land.**
+
+CodeGraph's only channels to influence the agent are low-salience: the MCP `initialize` instructions (`server-instructions.ts`) and the tool descriptions. Changing them does **not** reliably move the agent's tool _choice_ or query style — validated: trace-first steering ported into the server-instructions + tool descriptions (3 wording variants) never reproduced what a CLI `--append-system-prompt` achieved, and **regressed** wall-clock vs baseline. New tools fare worse (rarely chosen — the agent under-picks even `trace`); "better examples" is the same steering. The agent's tool-choice does improve on its own as host models get better at tool use — but that is not ours to force.
+
+What works is meeting the agent where it already is:
+- **explore-flow** — `codegraph_explore` is the PRIMARY tool the agent reliably calls; its query is a precise bag of symbol names (incl. qualified `Class.method`) spanning the flow the agent is after; explore finds the call path _among those named symbols_ (riding synthesized edges) and leads its output with it. (`buildFlowFromNamedSymbols`: segment/co-naming disambiguation; ≤1 unnamed bridge so it never wanders a god-function's fan-out. Overload-aware: a PascalCase type token in the query biases an overloaded name to that type's own def — `DataRequest task` → DataRequest's `task`, not the abstract base; named-symbol files sort first.)
+- **Sufficiency** — make the tool's output complete enough that the agent stops. `codegraph_node` returns the full body + the caller/callee trail, and for an AMBIGUOUS name returns **every overload's body in one call** (so the agent never Reads a file to find the right overload — validated on Alamofire/gin). This is the after-explore depth tool (labeled SECONDARY).
+- **Errors teach abandonment** — one or two `isError: true` responses early in a session and the agent stops calling codegraph entirely (maintainer-observed, repeatedly). `isError` is reserved for genuine "stop trying" cases: security refusals (`PathRefusalError`) and real malfunctions (which carry a retry-once note). Every expected/recoverable condition — project not indexed, symbol not found, file not in the index — returns a **SUCCESS-shaped response carrying the guidance** (`NotIndexedError` → `textResult`, see `ToolHandler.execute`'s catch). The same principle is why the tool surface is **always exposed, even at an un-indexed root** (the old empty-`tools/list` gate was removed in #964 — it broke monorepos where only sub-projects carry a `.codegraph/`, and hid the tools from a session that started before `codegraph init`): safety comes from the response SHAPE (success-shaped guidance, never `isError`), not from hiding tools. An un-indexed root's `initialize` sends a per-project variant (`SERVER_INSTRUCTIONS_NO_ROOT_INDEX` — "pass `projectPath` to a project that has a `.codegraph/`"), not an "inactive" note; indexing is still deliberately the user's call, never the agent's.
+
+What fails is the inverse — folding a precise answer into a **fuzzy-input** tool: the now-removed `codegraph_context` took a description, not symbols, so it couldn't disambiguate a flow's endpoints and surfaced the _wrong feature_ (which is why it was cut). Precise output needs precise input — explore takes a symbol bag for exactly this reason. (`codegraph_trace` was likewise removed: explore-flow does its job and the agent under-picked it.)
+
+The remaining lever under this axis is **coverage**: every flow made to connect statically (a new dynamic-dispatch synthesizer, or extracting symbols static parsing skipped — e.g. object-literal store actions in `create((set,get)=>({...}))`) is then surfaced automatically by explore-flow, no agent change needed. Reactive/reconciler runtimes (Halo's `ReactiveExtensionClient`, MediatR, Vue Proxy) are the frontier — flows there have no static edges, so nothing surfaces (correctly — silent beats wrong). Full investigation + A/B record: `docs/benchmarks/call-sequence-analysis.md` + auto-memory `project_codegraph_read_displacement`.
+
+### Explore budget — keep BOTH budgets monotonic with repo size
+
+Two functions in `src/mcp/tools.ts` scale explore with indexed file count. This is the expected resolution (a regression here silently forces agents back to Read):
+
+| Repo | files | explore calls | chars/call | per-file |
+|---|---|---|---|---|
+| express (small) | 147 | 1 | 18K | 3800 |
+| excalidraw/django (medium) | 643–3043 | 2 | 28K | 6500 |
+| vscode (large) | 10446 | 3 | 35K | 7000 |
+| ~20k / ~40k | — | 4 / 5 | 38K | 7000 |
+
+- `getExploreBudget(fileCount)` → **call** budget: `<500→1, <5000→2, <15000→3, <25000→4, ≥25000→5` (max 5).
+- `getExploreOutputBudget(fileCount)` → **per-call** output (chars / files / per-file). **Invariant: a larger tier must never get a smaller `maxCharsPerFile` than a smaller tier.** (Regression that motivated this doc: the `<5000` tier's 2500 was *below* the `<500` tier's 3800, so on a god-file repo — excalidraw's 415 KB `App.tsx` — one explore returned <1% of the file and forced a Read.)
+- Explore output must **never tell the agent to "use Read"** — steer to another `codegraph_explore` and "treat returned source as already Read."
+
+### Dynamic-dispatch coverage — the flow must EXIST in the graph end-to-end
+
+Static tree-sitter extraction misses computed/indirect calls, so flows break at dynamic dispatch and the agent reads to reconstruct them. Synthesizers/resolvers bridge these so `codegraph_explore` connects them end-to-end (`src/resolution/callback-synthesizer.ts`, `src/resolution/frameworks/`). Channels today: callback/observer, EventEmitter, **React re-render** (`setState`→`render`), **JSX child** (`render`→child component), **React Native native→JS events** (`sendEvent(withName:)` / JVM `emit` → the `addListener` handler, named or inline, `rn-event-channel`), django ORM descriptor. The JS→native direction is a *resolver* (`frameworks/react-native.ts`: `RCT_EXPORT_METHOD`, `RCT_EXTERN_MODULE` Swift shims, TurboModules), which trusts receiver evidence — an alias bound to `NativeModules.X` — over the import resolver. All synthesized edges are `provenance:'heuristic'` with `metadata.synthesizedBy` + `registeredAt` (the wiring site), surfaced inline in `codegraph_explore`'s Flow section and the `codegraph_node` trail.
+
+**Principle: partial coverage is WORSE than none.** Bridging one boundary but not the next reveals a hop the agent then drills + reads to finish. Measured on excalidraw: react-render alone *raised* reads to 5–7; only completing the flow (adding the jsx-child hop) dropped it to 0–1. **Always close the flow end-to-end and re-measure** — never ship a half-bridged flow.
+
+
+### Validation methodology & worked examples
+
+**Required** for every new language/framework: validate on small/medium/large real repos with >=3 flow prompts; deterministic probes (`scripts/agent-eval/probe-*.mjs`) then agent A/B (`scripts/agent-eval/run-all.sh` / `ab-new-vs-baseline.sh`). Pass bar: ~0 Read/Grep within the explore-call budget, faster than without-codegraph, no control-repo regression.
+
+Full methodology (feedback metrics, CLI contamination guard, Sonnet/`--effort high` model policy, daemon pre-warm), the Excalidraw worked example, and coverage matrix live in:
+- `docs/AGENTS.md` (nested; also loaded when cwd is under `docs/`)
+- `docs/design/dynamic-dispatch-coverage-playbook.md`
+- `docs/design/callback-edge-synthesis.md`
+- `docs/benchmarks/call-sequence-analysis.md` / `docs/benchmarks/agent-eval-feedback-metrics.md`
+
+
+Tests live in `__tests__/` and mirror the module they cover. Notable ones beyond the obvious:
+
+- `installer-targets.test.ts` — parameterized contract suite across all 4 agent targets (see installer notes above).
+- `evaluation/` — `runner.ts` + `test-cases.ts` exercise codegraph against synthetic projects and score the results; run via `npm run eval` (builds first). Not part of `npm test`.
+- `sqlite-backend.test.ts` / `node-sqlite-backend.test.ts` — pin that `node:sqlite` is the sole backend: `getBackend()` reports `node-sqlite` and the DB comes up in WAL.
+- `pr19-improvements.test.ts`, `frameworks-integration.test.ts` — regression coverage for specific past PRs/incidents; don't rename these, the names anchor to git history.
+
+Tests create temp dirs with `fs.mkdtempSync` and clean up in `afterEach`. They write real files and exercise real SQLite — there is no DB mocking.
+
+### Windows-gated tests
+
+Behavior that differs by platform (path resolution, drive letters, `SENSITIVE_PATHS`, `%APPDATA%` config dirs, CRLF) must be gated, not assumed. Use `it.runIf(process.platform === 'win32')(...)` for Windows-only assertions and `it.runIf(process.platform !== 'win32')(...)` for POSIX-only ones — e.g. `/etc` is sensitive on POSIX but resolves to `C:\etc` (non-existent) on Windows, so an ungated `/etc` assertion fails on Windows. Validate the Windows side for real (see below); don't merge a Windows-gated test you haven't seen run.
+
+## Cross-platform validation
+
+The dev machine — and the default `npm test` target — is **macOS**, so local runs cover the macOS path. The other two platforms aren't here; when a change is platform-sensitive (file watching, sockets / named pipes, path & symlink handling, process lifecycle, inotify budget) validate them for real rather than guessing.
+
+### Linux (Docker)
+
+When asked to test or validate on Linux, use **Docker** — there's no Linux box, but Docker runs on the macOS host. Build a throwaway image from the repo and run the suite inside it:
+
+- `FROM node:22-bookworm`; `COPY` the repo with a `.dockerignore` excluding `node_modules`/`dist`/`.git`/`.codegraph`; `RUN npm ci && npm run build`. Don't reuse the Mac `node_modules` — `esbuild`/`rollup` ship platform-specific binaries.
+- Run with **`docker run --rm --init`**. The `--init` is load-bearing for any process-lifecycle test (daemon reaping, the #277 PPID watchdog, idle-timeout): without a zombie-reaping PID 1, a SIGKILL'd/exited process lingers as a zombie and `process.kill(pid, 0)` still reports it *alive*, so exit-detection assertions false-fail even though the process did exit.
+- Linux is where the inotify watch budget actually bites: count a process's watches via `/proc/<pid>/fdinfo/*` (sum `^inotify ` lines on the fd whose `readlink` is `anon_inode:inotify`).
+
+### Windows (Parallels VM + SSH)
+
+For any Windows-specific PR, bug, or implementation, validate it on the real Windows VM rather than guessing. Connection details live in the gitignored **`.parallels`** file at the repo root (VM name, guest IP, SSH user/key). `prlctl exec` needs Parallels Pro and is unavailable, so SSH is the bridge.
+
+- Connect / run from the Mac host: `ssh <user>@<guest_ip> "..."`. For multi-line work, pipe PowerShell over stdin and **refresh PATH from the registry** first (sshd's session has a stale PATH after winget installs):
+  ```
+  ssh colby@10.211.55.3 "powershell -NoProfile -ExecutionPolicy Bypass -Command -" <<'PS'
+  $env:Path = [Environment]::GetEnvironmentVariable("Path","Machine") + ";" + [Environment]::GetEnvironmentVariable("Path","User")
+  Set-Location C:\dev\codegraph
+  PS
+  ```
+- Clone fresh into a **Windows-local** path (`C:\dev\codegraph`) and `npm ci` there — never run npm against the shared Mac repo, since `esbuild`/`rollup` ship platform-specific binaries.
+- Guest toolchain (winget): Node LTS, Git, and the **VC++ ARM64 redistributable** (required by `@rollup/rollup-win32-arm64-msvc`, which vitest pulls in).
+- Fetch a contributor PR head straight from their fork to dodge `pull/<n>/head` lag: `git fetch <fork-url> <branch>` then `git checkout -f FETCH_HEAD`.
+- Known pre-existing Windows failures (they reproduce on `main`, unrelated to your change — confirm against `origin/main` before blaming your PR, and don't let them mask new regressions): `security.test.ts > Session marker symlink resistance > does not follow a pre-planted symlink` (symlink creation needs privileges on Windows); and the `mcp-initialize.test.ts` / `mcp-roots.test.ts` suites, which fail in `afterEach` with `EPERM` removing the temp dir because a spawned `serve --mcp` (its `--liftoff-only` re-exec grandchild) still holds the cwd / SQLite file open — a Windows file-locking quirk, not a logic bug.
+
+## Releases
+
+Released to npm and mirrored as [GitHub Releases](https://github.com/colbymchenry/codegraph/releases). `CHANGELOG.md` is the source of truth; GitHub Release notes are extracted from it.
+
+### Writing changelog entries
+
+**Default: write entries under `## [Unreleased]`** — that's the section reserved for work landing between releases. **Don't pre-create a `## [X.Y.Z]` block** for the next release: the Release workflow's first step is `scripts/prepare-release.mjs`, which automatically promotes everything under `[Unreleased]` into a new `## [X.Y.Z] - <YYYY-MM-DD>` block at release time (or merges into a pre-existing `[X.Y.Z]` block if one exists — but you don't need one). Pre-staging is what caused the v0.9.5 sparse-release-notes incident: a sparse `[0.9.5]` block hand-added before the rest of the work landed got picked by the extractor over the much-larger `[Unreleased]` section above it. Don't do that.
+
+Formatting rules for any entry (anywhere — `[Unreleased]` or otherwise):
+
+1. **Write friendly, user-facing notes — not engineer-facing ones.** Group under `### New Features` and `### Fixes` (sentence-case). Surface `### Breaking Changes` and `### Security` as their own sections **only when the release has them**; fold improvement-flavored changes into New Features. Omit empty sections. (This replaces the old Keep-a-Changelog `Added/Changed/Fixed/Removed/Deprecated` grouping: the GitHub Release page extracts each version block **verbatim** via `scripts/extract-release-notes.mjs`, and the old dense, implementation-focused entries rendered as an unreadable wall of text — so the whole CHANGELOG was rewritten to this format and every published release re-noted to match.)
+2. **One plain-language sentence per bullet:** what changed and why it matters to a user. Lead with the capability, or with the symptom that's now fixed.
+3. **Strip the internals.** No internal file paths (`src/...`), no internal symbol / function / class names, no benchmark numbers / percentages / node-or-edge counts. **Keep:** language & framework names (Go, Spring, NestJS, …), things a user types or sets (`codegraph install`, `codegraph_explore`, the `CODEGRAPH_*` env vars), agent / IDE names (Claude Code, Cursor, opencode, Kiro, …), and a brief `Thanks @user` when a contributor is credited.
+4. Issue / PR references in entries are by number (`(#403)` etc.); the GitHub renderer auto-links them in the published release notes.
+5. **Don't add a `[X.Y.Z]: https://...` link reference yourself** — `prepare-release.mjs` appends it automatically when it promotes the version (idempotent: a re-run is a no-op if it already exists).
+6. **Every release opens with a `### Highlights` block — the only part most people read.** At most ~8 one-line bullets, in plain language for someone who doesn't read code, ordered by what a typical user notices first (new agent/IDE support and setup changes, then answer quality, then reliability), plus a one-sentence upgrade note when a re-index is needed. Write or refresh it in `[Unreleased]` when a release is being prepared — not per PR — and keep the detailed `### New Features` / `### Fixes` entries below it. When `### Fixes` grows past ~15 entries, group them under `####` sub-headings (`Better answers from codegraph_explore`, `Finding your project, live updates, and the CLI`, `Indexing reliability and disk usage`, `Language and framework accuracy`) so a skimmer can find their area.
+
+Multi-word headings like `### New Features` are safe on the normal release path: `prepare-release.mjs` **Case A** moves the whole `[Unreleased]` body verbatim into `[X.Y.Z]`. (Only its rarely-used **Case B** *merge* splits sub-sections with a single-word `^### (\w+)$` regex that wouldn't match them — and Case B fires only if a `[X.Y.Z]` block was pre-created, which rule above already forbids.)
+
+### Release flow (the user runs these)
+
+Releases are built and published by the **GitHub Actions "Release" workflow**
+(`.github/workflows/release.yml`). It runs `scripts/prepare-release.mjs` to
+promote `[Unreleased]` into `[<version>]` (and auto-commit + push that
+CHANGELOG change back to `main` so on-disk truth matches the published
+notes), then bundles a Node runtime per platform (`scripts/build-bundle.sh`)
+and publishes both the GitHub Release and the npm thin-installer
+(`scripts/pack-npm.sh`: a shim package + per-platform packages).
+Publishing manually is **wrong** now — a plain `npm publish` ships the root
+package (non-bundled), which breaks anyone on Node < 22.5.
+
+**Claude does NOT bump the version unless explicitly asked.** The maintainer
+typically does it themselves — often by editing `package.json` directly via
+the GitHub web UI. Don't proactively commit a version bump as part of
+unrelated work, and don't propose one when summarizing a PR.
+
+When the maintainer DOES bump the version, the only edit strictly required is
+to `package.json` — the workflow's "Sync package-lock.json" step detects a
+mismatch between `package.json` and `package-lock.json`, runs
+`npm install --package-lock-only --ignore-scripts` to rewrite the lock file's
+version fields (top-level + `packages.""`), and auto-commits + pushes the
+result back to `main` with `[skip ci]`. So a GitHub-web-UI single-file edit to
+`package.json` is enough to kick off a clean release. (If they edit both files
+locally, that's fine too — the sync step no-ops.)
+
+Once `package.json` is at the target version on `main`, trigger
+**Actions → Release → Run workflow** (on `main`). The workflow:
+
+1. Syncs `package-lock.json` to `package.json`'s version if they've drifted; commits + pushes that change.
+2. Runs `prepare-release.mjs <X.Y.Z>` → promotes `[Unreleased]` → `[X.Y.Z] - <today>` in `CHANGELOG.md`, appends the link reference, commits + pushes the move with `[skip ci]`.
+3. Builds every platform bundle on one runner, generates `SHA256SUMS`.
+4. Creates the GitHub Release with notes from the freshly-promoted `[X.Y.Z]` block.
+5. Publishes the npm shim + per-platform packages. Requires the `NPM_TOKEN` repo secret.
+
+**Do not run `npm publish`, `git push`, or `git tag` yourself** — these are
+publish actions on shared state. Write the files, hand the user the commands.
+
+## House rules
+
+- The `0.7.x` line is in active multi-agent rollout. Any change to `src/installer/` (especially `targets/`) needs corresponding test coverage and a CHANGELOG entry — installer regressions break every new install silently.
+- When changing what the MCP tools do or how agents should use them, edit `src/mcp/server-instructions.ts` — it is the **single source of truth** for agent-facing tool guidance (issue #529). The installer no longer writes a duplicate instructions block into `CLAUDE.md` / `AGENTS.md` / `GEMINI.md` / `.cursor/rules/codegraph.mdc` / Kiro steering, so there's nothing to keep in sync anymore. (The repo's own checked-in `.cursor/rules/codegraph.mdc` is dogfooding config — update it too if you use Cursor on this repo, but it ships nowhere.)
+- **Before adding or extending a router, a web framework, or a language's `WHEN` rules, read `docs/design/framework-coverage.md`.** It is the standing answer to "what is supported and what is left" across the three axes (route nodes → Entry points, `navigates` edges → Screens, branch-guard rules → the `WHEN` labels), with what each remaining item needs, the traps that have already cost debugging time, and the queries to re-verify it. Update it in the same change that moves a row.
+- CodeGraph provides **code context**, not product requirements. For new features, ask the user about UX, edge cases, and acceptance criteria — the graph won't tell you.
+- **When the user references issues, PR comments, or external reports, anchor them to a date and version before drawing conclusions.** Check the comment's `createdAt` against:
+  - The **last released version** — `grep -m1 '^## \[' CHANGELOG.md` shows the top-of-file version (older releases follow). A comment dated before the latest `## [X.Y.Z] - YYYY-MM-DD` is reacting to *released* state — work that's only on `main` or on an unmerged branch doesn't apply.
+  - The **last main commit** — `git log --first-parent main -1 --format='%ai %h %s'`. A comment after the last release but before a fix on main may already be addressed there but unreleased.
+  - The **current branch's tip** — your own unmerged work obviously can't be what the comment is reacting to.
+  Always disambiguate "released," "merged-but-unreleased," and "in-progress" before agreeing that a user-reported problem is unfixed (or that a fix is incomplete). A user saying "your fix only covers X" about a recent PR is usually pointing at the *released* shortcomings — your in-flight branch may already address them but they have no way to know that.
+- **Version-tag every image referenced in `README.md`.** GitHub caches README images (`raw.githubusercontent.com` with a 5-minute TTL; third-party hosts sit behind the long-lived camo proxy), so updating an asset in place can keep showing the stale version. Give each README image URL a `?v=N` query tag and **bump `N` in the same commit whenever the asset bytes change** — e.g. `assets/waitlist.svg?v=2`. The changed URL sidesteps every cache so the new image shows immediately instead of waiting on a TTL to expire.

CHANGELOG.md

@@ -25,6 +25,8 @@ and adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ### New Features
 
+- **Codex and Astra read project guidance from `AGENTS.md`.** The canonical agent guide now lives in `AGENTS.md` (with a nested `docs/AGENTS.md` for long validation notes); `CLAUDE.md` is a thin `@AGENTS.md` wrapper for Claude Code. Codex/Astra no longer miss the old CLAUDE-only instructions.
+
 - **A busy screen's picture is laid out by the parts of the screen.** A screen is a set of handlers with no order between them, so on a hub screen the old rows-by-distance collapsed into one enormous row — the main screen of one app put 89 boxes side by side on a canvas over 28,000px wide, every line a near-horizontal sweep across all of it. The Steps tab now groups a screen's picture by region — the component that owns each handler, named in a small caption over its boxes — with each region a column where a step sits above what it sets in motion, tiled in the screen's own source order. At rest the picture hides only two things: the screen's own fan-out — one line into each region stands in for it — and lines that point back up; every other line draws where it leads, between two regions included, and selecting a step brings out its whole story in the side panel, link by link. A box nothing points at is the screen's own doing — run on render or mount, or from a binding written inline — the key says so, and selecting it lights its line from the screen with what fires it. The same app's widest screen now lays out under 3,500px with every line local, and the whole picture fits on screen when it opens. Endpoints, handlers and the in-order reading are untouched, and nothing needs a re-index: the regions come from the same walk that draws the steps.
 
 - **Where the code chooses, the picture says so once.** A helper that ends `return (await hasSeenWelcome(id)) ? '/home/' : '/welcome/'` sends the app to one of two screens, but the Steps picture drew that as two separate arrows, each carrying the whole condition with one of them negated and both cut off at the same forty characters — and before you clicked anything, neither arrow was labelled at all, so nothing said it was a choice. Now sibling arrows out of one box that are the arms of one `if`, `switch` or ternary are drawn as the choice they are: the condition is written once under the box that decides it, and each arrow out says only which way it is — `yes`, `no`, or a case's own value. They are the only arrows labelled before you select anything, so the picture reads at a glance without becoming a wall of text. A one-sided guard — an early exit, an `if` with only one side drawn — still carries its condition on the arrow, and an arrow that is reached whether or not the condition holds never claims a side. Nothing needs a re-index: the decision is read from the source at request time.

docs/AGENTS.md

@@ -0,0 +1,36 @@
+# AGENTS.md (docs/)
+
+Nested Codex guidance under `docs/`. Loaded with the repo-root `AGENTS.md` when cwd is under `docs/` (Codex walks root to cwd; shared `project_doc_max_bytes` budget).
+
+Root `AGENTS.md` already carries the non-negotiable retrieval principles (adapt-the-tool, explore budgets, end-to-end synthesis). This file holds the longer validation methodology and Excalidraw worked example that were moved out of root to fit the budget.
+
+### Validation methodology (REQUIRED for every new language/framework)
+
+For each **language × framework**, validate on **small, medium, and large** real repos with **≥3 different flow prompts** each:
+
+1. **Pick the canonical flow** for the framework ("how does X reach Y": state→render, request→handler→view, query→SQL, action→reducer→store…).
+2. **Deterministic probes** (`scripts/agent-eval/probe-{node,explore}.mjs` against the built `dist/`): `codegraph_explore` with the flow's symbol names connects from→to end-to-end with no break (its Flow section shows the path); **no node explosion** (`select count(*) from nodes` stable before/after re-index); synthesized-edge **precision** spot-check (`select … where provenance='heuristic'`).
+3. **Agent A/B** (`scripts/agent-eval/run-all.sh <repo> "<Q>"`): with vs without codegraph, **≥2 runs/arm** (run-to-run variance is large — never conclude from n=1). Record **duration, total tool calls, Read, Grep**. Optional forced-Read-0 sufficiency proof via the block-read hook (`scripts/agent-eval/hook-settings.json`).
+   - **Every run also reports three feedback metrics** — residual context occupancy, explore sufficiency (what the agent did NEXT after each explore), and allocation efficiency (share of returned bytes the answer cited) — under each run, plus a side-by-side arm table (`compare-arms.mjs`). Entry point: `docs/benchmarks/agent-eval-feedback-metrics.md`. Reading them: `Read a file we returned` is an allocation miss, `Read a file we did NOT return`/`Grep` is recall; allocation efficiency is **relative** (attribution is by citation) so it is only valid between builds on the same question; occupancy *shares* are Claude Code / 200k and don't transfer to another host — the arm ratio does.
+   - **The `codegraph` CLI is blocked in every arm** (`no-cli-shim.sh`: sanitized PATH + a PreToolUse hook, shared by both harnesses). Without it 14 of 15 without-arm runs in one 7-repo pass reached codegraph through Bash. Check the contamination row before believing any number: `CLI calls that RETURNED output` > 0 invalidates the run (in a new-vs-baseline A/B it silently drops calls from all three metrics, since a CLI explore is not a tool call).
+   - **Model policy — every A/B arm runs Claude with `--model sonnet --effort high`. Always. Never Opus/Fable.** All `scripts/agent-eval/*.sh` default to this (`MODEL`/`EFFORT` env override exists — don't raise it without an explicit reason from the maintainer). Two reasons, and the second matters more than cost: (a) Sonnet doesn't burn tokens; (b) **Sonnet is the deliberate floor model** — codegraph's real users attach it to whatever agent they already run (Cursor Composer, Gemini, etc.), so we validate on a "dumber" model on purpose: a stronger model's tool-use covers up the salience/sufficiency problems a weaker one exposes. An affordance that lands on Sonnet generalizes up to every host; one that only works on Opus/Fable doesn't generalize down to the agents most users actually have. Both arms always use the same model.
+   - **MCP attach is a startup-latency issue, not a hard block.** On a multi-step task the agent dives into Read/grep before codegraph finishes its ~2-3s startup (worse when the eval is itself run nested inside a Claude session, under CPU contention), so it runs with no codegraph. Fix: **pre-warm a persistent daemon** for the target (`CODEGRAPH_DAEMON_IDLE_TIMEOUT_MS` high; spawn `serve --mcp --path <target> </dev/null &`; wait for `.codegraph/daemon.sock`) **and skip the startup re-exec** (`CODEGRAPH_WASM_RELAUNCHED=1`) so claude connects before the agent's first turn. Don't trust claude's `init` snapshot — it can read `status:"pending"` / 0 tools even when it then connects; judge by actual codegraph usage in `parse-run.mjs`'s `by type`. To isolate a change — **new-build vs baseline-build, both codegraph-on** (vs run-all.sh's with-vs-without) — use `scripts/agent-eval/ab-new-vs-baseline.sh <indexed-repo> "<task>" [baseline-ref]` (it bakes in the pre-warm).
+4. **Pass bar:** a normal flow question reaches **~0 Read/Grep within the repo's explore-call budget**, runs **faster** than without-codegraph, and shows **no regression on a control repo**. Record the numbers in `docs/design/dynamic-dispatch-coverage-playbook.md` (the coverage matrix).
+
+Full playbook + per-mechanism design: `docs/design/dynamic-dispatch-coverage-playbook.md` and `docs/design/callback-edge-synthesis.md`.
+
+### Worked example — Excalidraw (TS/React, medium, 643 files)
+
+The template to replicate per language/framework. Question: *"how does updating an element re-render the canvas on screen?"* (the full flow crosses three React boundaries: observer callback, `setState`→`render`, and JSX child).
+
+| Stage | duration | Read | Grep | codegraph |
+|---|---|---|---|---|
+| Without codegraph | 115–139s | 9–10 | 10–11 | 0 |
+| Broken (explore-budget regression) | 131–139s | 5–10 | 3–5 | 6–14 |
+| Fixed (budget + msgs + synthesis) | 64–112s | 0–2 | 2–4 | 3–**10** |
+| + trace-first steering | **51–74s** | **0–2** | 0–4 | **3–4** |
+
+n=4 unhooked runs/stage, same prompt. After steering flow questions to `codegraph_trace` first: **best run 0 Read / 0 Grep / 3 codegraph / 51s**; **2 of 4 fully clean** (0 Read, 0 Grep). Steering eliminated the over-drill variance — call count tightened from 3–10 to 3–4, trace adoption went 3/4 → 4/4, and the `search`+`callers` path-reconstruction floundering dropped to 0. Run-to-run variance is still real; report the range, never a single run. **Residual reads/greps are all the nonce data-flow** (`canvasNonce` — a local prop with no graph edges); that's the def-use/data-flow frontier, left deliberately uncovered (tracking every local would explode the graph). Validated: `trace(mutateElement, renderStaticScene)` connects in **6 hops** across all three boundaries (`mutateElement → triggerUpdate → [callback] triggerRender → [react-render] render → [jsx] StaticCanvas → renderStaticScene`), each hop showing inline source + the wiring site; node count stable at 9,289; 1 callback + 46 react-render + 280 jsx-render synthesized edges (no explosion, precision-checked).
+
+
+Also see: `docs/design/dynamic-dispatch-coverage-playbook.md`, `docs/design/callback-edge-synthesis.md`, `docs/benchmarks/call-sequence-analysis.md`, `docs/benchmarks/agent-eval-feedback-metrics.md`.