fix(ask): find related source evidence automatically - #688
Conversation
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
…to HEAD # Conflicts: # frontend/src/components/OperationsDashboard.stories.tsx # frontend/src/components/OperationsDashboard.test.tsx # frontend/src/components/OperationsDashboard.tsx
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
…textualWisdomLab/LineageWeave into fix/pr688-restack-child # Conflicts: # backend/app/post_content_worker.py
…to fix/ask-auto-source-composer
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
…textualWisdomLab/LineageWeave into fix/pr688-restack-child
…to fix/ask-auto-source-composer
…textualWisdomLab/LineageWeave into fix/pr688-restack-child
This reverts commit a7cda9b.
This reverts commit 1d16c40.
This reverts commit 891e0f9.
…to fix/ask-auto-source-composer
| where missing.post_id = any($1::uuid[]) | ||
| and job.status_code = $2 | ||
| and nullif(btrim(post.post_body), '') is not null | ||
| order by post.post_id | ||
| """, | ||
| [UUID(sibling_id) for sibling_id in sibling_ids], | ||
| SUCCEEDED, | ||
| ) |
There was a problem hiding this comment.
📝 Info: Re-analysis only reaches siblings with missing facts
The SUCCEEDED reclaim check in _claim_job still tests only source_body_sha256, not the new analysis_input_sha256, so a completed post never self-reclaims on a stale evidence window. Re-analysis depends entirely on _requeue_project_missing_case_jobs, which requeues only siblings that still carry missing facts. A completed sibling whose facts are all present is never re-analyzed when new project evidence arrives.
Was this helpful? React with 👍 or 👎 to provide feedback.
| "--agents", | ||
| str(agents_path), | ||
| "--auto-discover-model-agents", |
There was a problem hiding this comment.
🔍 Orchestrator start no longer tolerates model-discovery failures
start.py drops --allow-discovery-failures while chat agents still carry blank models expanded via --auto-discover-model-agents. If gateway /v1/models discovery fails, the orchestrator now fails to start instead of continuing, taking the whole LLM/Vision/embedding boundary down. This is coupled to the pinned-commit bump and is not documented in ADR 0083.
(Refers to this code)
Was this helpful? React with 👍 or 👎 to provide feedback.
| def operations_analysis_input_sha256( | ||
| sources: tuple[OperationsEvidenceSource, ...], context: str | ||
| ) -> str: | ||
| """Digest the exact ordered source window and context sent for analysis.""" | ||
| payload = { | ||
| "context": context, | ||
| "sources": [ | ||
| { | ||
| "post_id": source.post_id, | ||
| "title": source.title, | ||
| "input_sha256": source.input_sha256, | ||
| } | ||
| for source in sources | ||
| ], | ||
| } | ||
| encoded = json.dumps( | ||
| payload, ensure_ascii=False, separators=(",", ":"), sort_keys=True | ||
| ).encode("utf-8") | ||
| return hashlib.sha256(encoded).hexdigest() | ||
|
|
There was a problem hiding this comment.
📝 Info: Input fingerprint omits source clock and axis
operations_analysis_input_sha256 digests context and each source's post_id, title, and text digest, but not observed_at or time_axis_code. Milestones bind their instant to observed_at, so a source whose clock changes while text/title/id stay identical produces the same fingerprint and does not invalidate reuse.
Was this helpful? React with 👍 or 👎 to provide feedback.
1 participant
Summary
Evidence
uv run pytest -q tests/test_post_chat_ingestion.py tests/test_post_content_worker.py(27 passed)corepack pnpm exec vitest run src/components/OperationsDashboard.test.tsx --maxWorkers=1(12 passed)corepack pnpm exec vitest run src/App.test.tsx --maxWorkers=1 -t "renders safe Ask Agent evidence under each cited post"(1 passed, 95 skipped)corepack pnpm lintStacked on #640. No source names or production identifiers are included.