Repository navigation
docs(audit): runtime identity/HITL consistency audit + correct langgraph version claim - #18
Merged
Xander-Xai merged 2 commits intoOct 18, 2026
Conversation
…aph version claim Audit scope: AgentRuntime / PostgreSQL checkpoint / Celery / Redis lease / idempotent tool execution / HITL / crash recovery. Verified against live main (8e45f99) with real PostgreSQL + Redis. No new reproducible code defect found. The Issue #12 (Part A/B + partial-resume), #13 and #14 fixes re-verified as PASS on real infrastructure (199 runtime integration + 3481 unit + chaos/delivery/verify artifacts). One stale fact corrected: live docs/comments claimed HITL LangGraph semantics were "实测 langgraph 1.2.12", but requirements-lock.txt pins 1.2.2, the venv installs 1.2.2, and a fresh chaos artifact records 1.2.2 — so a lockfile install cannot reproduce a 1.2.12 claim. Align the live references (code comments, CLAUDE.md, HITL design doc, test docstring) to 1.2.2 and close the docs/limitations.md section 9 debt entry, with the HITL interrupt/resume test suite as regression evidence. Historical audit snapshots are left untouched (snapshot-not-backfilled convention). Adds docs/audit/2026-10-19-runtime-identity-hitl-consistency-audit.md with the capability -> file/symbol/test map, the reproduced fault matrix, boundaries (at-least-once / not exactly-once / real ERP NOT_VERIFIED / production cluster NOT_VERIFIED / RAG metrics NOT_VERIFIED / cache hit-rate NOT_MEASURED).
Xander-Xai
marked this pull request as ready for review
October 18, 2026 18:06
Xander-Xai
deleted the
audit/runtime-identity-hitl-consistency-20261019
branch
October 18, 2026 18:06
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What & why
Consistency audit of the claimed AgentRuntime / PostgreSQL checkpoint / Celery /
Redis lease / idempotent tool execution / HITL / crash recovery capabilities
against live
main(8e45f99), executed on real PostgreSQL + Redis.No new reproducible code defect. The Issue #12 (Part A/B + partial-resume),
#13 and #14 fixes are re-verified as PASS on real infrastructure.
One stale fact was corrected: live docs/comments claimed HITL LangGraph
semantics were "实测 langgraph 1.2.12", but
requirements-lock.txtpins1.2.2, the venv installs 1.2.2, and a fresh chaos artifact records
1.2.2 — so a lockfile install cannot reproduce a 1.2.12 claim. Live
references are aligned to 1.2.2 and the
docs/limitations.md§9 debt entry isclosed, with the HITL interrupt/resume suite as regression evidence. Historical
audit snapshots are untouched (snapshot-not-backfilled).
Deliverable
docs/audit/2026-10-19-runtime-identity-hitl-consistency-audit.md— HEAD/syncrecord, capability → file/symbol/test map, reproduced fault matrix, fast-path
vs durable-run authorization boundary, unresolved risks, interview-usable
phrasing.
Verification (real PG + Redis)
pytest tests/unit -qpytest tests/integration/runtime -qpytest tests/integration/runtime/test_celery_delivery_semantics.pypython scripts/test_worker_crash_recovery.pypython scripts/verify_distributed_runtime.pyoverall_status: PASSpytest test_hitl_langgraph_interrupt.py test_runtime_doc_drift.pypython scripts/audit_doc_consistency.pyruff check/ruff format --check(changed files)Boundary (unchanged, not overclaimed)
At-least-once, not exactly-once; external end-to-end exactly-once
NOT PROVIDED; real ERP write NOT_VERIFIED; production cluster /
multi-replica NOT_VERIFIED; no monotonic fencing token; RAG formal metrics
NOT_VERIFIED; cache hit-rate / latency deltas NOT_MEASURED.
Draft pending review + CI.