Trang chủ Khám phá Trợ giúp
Đăng ký Đăng nhập
haiany
/
deepseek-harness
mirror of https://github.com/deepseek-ai/deepseek-harness.git
1
0
Fork 0
Các tập tin Các vấn đề 0 Wiki
Branch: master
Branches Tags
deepseek-harnes...
/
packages
/
session
/
session-telemetry
/
README.md

README.md 8.5 KB
Permalink Lịch sử Raw

description: "Session-telemetry capture seam for deployments and backend authors choosing a reporting backend, mounting redaction rules, or implementing the backend contract."

kind: "package-library"

@deepseek-ai/dsh-session-telemetry

English | 中文

Summary

Session telemetry lets deployments send ordered copies of session activity for reporting while preserving the canonical session log. Deployments choose one reporting backend and can redact each outbound copy before delivery; without redaction rules, captured data leaves the process unchanged. The handoff is non-blocking, so reporting does not delay session processing. Delivery is best effort, and queued records may be lost if the process crashes.

Table of Contents

Use this package

As a deployment, choose a backend, mount it, and add redaction rules when records must not leave the process as captured. As a backend author, implement the three-member contract and compose the coordinator with a capture mode.

Choosing and mounting a backend

Load exactly one backend plugin; it registers ctx.sessionTelemetry with the capture coordinator and its delivery pipeline. A duplicate load throws. The required sharing member reports the deployment mode, not per-session admission or delivery. A consumer may report "not configured" only when no telemetry service is mounted. The /feedback command confirms recording without reading this policy.

The backend contract

A backend implements three members: emit(record) must be a non-blocking enqueue because it runs synchronously on the session-event path; optional flush() is a fire-and-forget hint after a turn ends, which most backends omit in favor of their SDK's own batching schedule; shutdown() drains queued records and resolves when the SDK stops, and disposal awaits it. A backend that implements flush() must order concurrent flushes with the final shutdown() drain.

What gets captured

Capture runs in one of two modes. live capture follows session events as they are appended, replays already-live sessions at mount time, and records lifecycle markers; on-demand capture reads the canonical session log only when the backend requests a prefix through captureSession(session, throughSeq?). Coordinator options select whether stored history is included. Every canonical session event maps to one ledger record in order. An assistant/message or assistant/attempt record carries its complete embedded compact stream, including failed and retried output. Each ledger record also carries session.id, session.format_version, the numeric event identity, optional header facts, and a pre-mapped severity (error for tool/result.isError, turn/end error reasons, and agent-error; info otherwise).

The sharing disclosure

Every backend discloses its deployment mode through sharing: full, feedback-only, or disabled. A backend may additionally restrict eligible Sessions. This property is not a delivery receipt; handoff is a non-blocking enqueue, and batching, retry, and loss policy belong to the backend SDK.

Redacting records

Every outbound record passes the sessionTelemetry/record waterfall after the coordinator copies its canonical event. This package ships no rules: with no listener mounted, records reach the backend exactly as captured, so exported data is as clean as the rules a deployment mounts. Listeners stack by transforming next()'s return value; a throwing listener withholds that one record fail-closed. Redaction applies to the outbound copy only — the canonical session log is never rewritten.

Understand the implementation

Implementation internals — click to expand This section explains the capture design; the observable behavior is fully covered in [Use this package](#use-this-package). ### Design concept The seam is built on one boundary: the harness's aspect ends at `emit()`. Complete event capture, redaction, and the handoff cursor live here; batching, retry, queueing, and loss policy are the reporting SDK's, deliberately not modelled or wrapped. The design and rejected alternatives are pinned in the [revival Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md). ### Source map | File | Role | |---|---| | [`src/index.ts`](src/index.ts) | Service Definition: `SessionTelemetryBackend`/`SessionTelemetrySink` contract, record vocabulary, `session-telemetry/record` waterfall declaration | | [`src/coordinator.ts`](src/coordinator.ts) | Capture: live listeners, lifecycle-local on-demand replay, redaction, handoff cursor, containment | ### Capture flow Live capture registers Session events, flush hints, shutdown markers, and agent/error observers through the composing fiber’s effects. On-demand capture registers only the disposal effect and reads the requested canonical-log prefix under its history policy. Synchronous handlers contain failures so they cannot affect the agent loop or other listeners. ### The handoff cursor A module-scope `WeakMap` records the highest sequence handed off, not delivered. Re-adopting the same object resumes after that cursor. Capture normally starts at `firstLiveSeq`; explicit `includeHistory: true` starts an unhanded object at seq 0, including restored or fork history. The backend owns capture authorization. Stored history does not itself authorize capture; the OTel backend waits for new explicit feedback. Receivers deduplicate repeated records by `(session.id, session.format_version, event.seq)`.

Further Exploration

Read these pages when the seam contract is not enough. They move from the shipped backend to the subsystem reference and the decision evidence.

Model Experience

None, as the seam observes the session stream and hands redacted copies outward; it registers nothing model-facing.

KV Cache effect

None; the package neither assembles nor sends a provider request.

Known Limitations and Deferred Work

These limits define the delivery and data-protection guarantees a deployment gets. They are current package constraints.

  • Best-effort delivery — the cursor marks handed-off, not delivered; a session torn down inside a reload window cannot be re-adopted, and whatever sits in a backend queue at crash time is lost. A durable outbox (spool, per-sink cursors, at-least-once) is deferred until a deployment states a crash-loss requirement.
  • No built-in redaction rules — with no sessionTelemetry/record listener mounted, records leave the process exactly as captured, including any credentials embedded in file contents or command output; a deployment exporting to a shared collector owns its rule set.
  • On-demand redaction uses current state — uncaptured events exist only in the canonical session log; a later captureSession() deep-copies and redacts their current values with the policy mounted at that time, and there is no capture-time telemetry snapshot or durable pre-capture spool.

Dev Note

Working context for maintainers — click to expand None.

Runtime invariant: No companion is published. The package's whole output is the backend handoff — a synchronous emit() call outside every authoritative event stream — and its capture side never appends session events, so no event/data relation exists for an independent companion to observe.