Conversation
process.exit(existsSync(p) ? 0 : 1) caused the Bash tool to report "Error: Exit code 1" on every normal capture (no collision). Swap to ? 1 : 0 so the happy path exits 0 and collision exits 1. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Introduces two new domain terms for the doc context enrichment feature: - Doc Context: aggregated summaries of work items and Confluence pages injected as a preamble into every Review Aspect agent's prompt - Doc Context Sub-agent: short-lived agent that fetches and summarises a single source in a diff-aware way Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
ADR-0010: Confluence fetch logic lives as an inline .mjs script in pr-review/ rather than depending on unic-confluence or extracting a shared package — no runtime package resolution for installed plugins without a bundler; credential lookup mirrors unic-confluence convention (env vars → ~/.unic-confluence.json) for the shared Unic audience. ADR-0011: New work item and doc sources are added as additive parallel paths, not a plugin registry — one source each does not justify the abstraction overhead. Condition to revisit: when a third distinct source type is introduced. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
… Confluence Adds a Doc Context gathering phase (step 4a) that runs before review agents and injects structured, diff-aware summaries of linked ADO work items and any Confluence pages referenced in their descriptions. Context is internal to agents only — no new PR comments are posted. Key design decisions captured in the spec: - Always-on; degrades gracefully at every tier (no work items, no creds, fetch errors) with console warnings where actionable - Two levels of parallel Doc Context Sub-agents: per work item and per Confluence page - Summaries are diff-aware and structured with source provenance labels - Step 4a runs in parallel with steps 5–7; step 8 waits for it Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
… scenarios - extend-pr-review-to-check-docs: original idea (PR description, ADO work items, Confluence pages as reviewer context) - alternative-work-item-sources-for-doc-context: Jira, GitHub Issues, etc. as future linked work item sources - alternative-doc-sources-for-doc-context: GitHub Wiki, Notion, etc. as future doc link destinations; flags auto-detect vs. config-file design question for future grilling Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
… range Remove hardcoded 0.2.0 target from spec 10 — the correct version depends on what the re-review feature ships at and must be determined at bump time. Extend the PROMPT.md spec scan range from the hardcoded "00 → 09" to "in order" so Ralph picks up spec 10 without manual edits. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…d apply triage Publishes the full issue set for the doc context enrichment feature: - PRD.md: parent document (status: closed) - 01: Confluence page client + node:test suite (ready-for-agent) - 02: Work item Doc Context enrichment — step 4a (ready-for-agent) - 03: Confluence page Doc Context enrichment (ready-for-agent, blocked by 01+02) - 04: Version bump + CHANGELOG + docs (ready-for-human) Issues 01–03 include agent briefs. Issue 04 includes a note explaining why it cannot be delegated to an agent (version depends on re-review shipping first; prose authorship benefits from human judgment). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…suffix Interim commit before version bump — part of spec-01.
All emitted signatures now use asterisk italics and include the
— Iteration {LATEST_ITERATION_ID} suffix. Detection prefix documented
in Comment signature and Notes sections.
…ead fetch Inserts Step 3.5 between the PR metadata fetch and file listing steps. Paginates ADO pullRequestThreads, scans all comments for the bot signature prefix, and exports IS_REREVIEW, PRIOR_THREADS_FILE, SUMMARY_THREAD_ID, and PRIOR_ITERATION_ID for use by downstream re-review steps.
Step 3.5 unconditionally fetches all ADO PR threads via paginated API, scans every comment for the bot signature prefix, and exports four variables (IS_REREVIEW, PRIOR_THREADS_FILE, SUMMARY_THREAD_ID, PRIOR_ITERATION_ID) for downstream re-review specs to consume.
…(v0.2.1) Two prior summaries caused isSummaryThread=true on both entries. Post-process the array to mark only the max-threadId candidate, and use last() instead of first() when reading SUMMARY_THREAD_ID.
…th LATEST_ITERATION_ID - Add Step 3.6 to fetch pullRequestIterations and resolve LATEST_ITERATION_ID and LATEST_COMMIT_ID via jq max_by(.id) - Add PRIOR_COMMIT_ID resolution for IS_REREVIEW=true (direct lookup + timestamp fallback for legacy comments without the — Iteration N suffix) - Replace hardcoded iterationId=1 in Step 4 with $LATEST_ITERATION_ID - Update Notes section and CLAUDE.md rule
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Branch Step 5 on IS_REREVIEW for incremental vs full diff - Early exit when PRIOR_COMMIT_ID == LATEST_COMMIT_ID (no new commits) - Fallback warnings with both commit IDs on git fetch failure - Export DIFF_HUNKS_FILE (JSON) for spec 05 thread classification - Mark spec 04 done
…s checks README row for spec 04 was left as pending after the implementation commit, which would cause the loop to re-target and re-implement it. Also removed the dead numeric comparisons (.status == 1 / .status == 6) from the Path C jq filter — ADO returns thread status as strings, making these unreachable. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…dules - Extract parseSignature, classifyThread, matchFinding, detectPriorReview into scripts/re-review/ - Add 11 fixture JSON files under tests/fixtures/ covering all classification scenarios - Write 25 tests across 4 test files (node:test); all pass - Wire modules into commands/review-pr.md; fix printf vs echo for zsh JSON safety Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…evelopment" This reverts commit a1b90a3
… guard Step 11.5, dedup parseSignature - Move partial-prior-run check from Step 10 Path B to end of Step 3.5 so the entire pipeline (diff range, agent analysis, posting) runs in a consistent mode from the start - Guard Step 11.5 completion-marker post with `if [ -n "$SUMMARY_THREAD_ID" ]` to avoid a failed API call when no summary thread was detected - Reuse `parseSignature` in `detect-prior-review.mjs` instead of duplicating the `/Iteration ([0-9]+)/` regex; eliminates dead-code risk for parse-signature.mjs Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
… missing tests, MARKER_FOUND guard - match-finding.mjs: skip threads where start/end is null instead of defaulting to line 0, which caused false positive matches near top of file - detect-prior-review.mjs: use max() over all parsed iteration IDs instead of last-seen; thread insertion order no longer affects priorIterationId - review-pr.md (Step 3.5): add exit-code guard and value validation for MARKER_FOUND to prevent silent wrong-direction fallback on node failure - tests: add 6 new cases (31 total) covering null-end classifyThread, deleted-file [0,0] hunks, numeric ADO status 2/5, null-start/end matchFinding, and priorIterationId max-ordering in detectPriorReview Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…hment feat(pr-review): re-review feature v0.8.0 + doc context enrichment planning
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
1 participant
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.
Summary
Release PR:
develop→mainfor the pr-review re-review feature.iterationIdselection, missing tests,MARKER_FOUNDguard, partial-run timing, Step 11.5 guard, dedupparseSignatureOnce merged, the release workflow will detect that
pr-review@0.8.0has no tag and create it automatically.Test plan
pnpm --filter pr-review verify:changelogpassespnpm --filter pr-review test— 25 tests, 0 failures🤖 Generated with Claude Code