Richer PR summaries + new issue summarizer - #18
Conversation
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Neither path captured it before — nothing needed it. Both GitHub's webhook payload and the REST issues-list response already include it, so this widens existing capture rather than adding a new GitHub call. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
SUMMARIZER_SYSTEM_PROMPT drops the terse structured What changed/Why convention for 2-3 sentences of plain prose that leads with the concrete change and includes detail the PR body supports. ISSUE_SUMMARIZER_SYSTEM_PROMPT is new: a plain restatement plus what needs doing, only when the issue actually supports it. Both share one Workers AI call core (makeWorkersAiSummarizer); workersAiSummarizer is renamed workersAiPrSummarizer to sit alongside the new workersAiIssueSummarizer. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Mirrors summarizePrSeam for issues, gated on action:"assigned" only — unassigned/opened/closed issue events never need a summary since they never surface in anyone's to-do. Runs alongside progressSeam, not instead of it. The "both seams fire" test also seeds a milestones row matching the fixture's github_ref, since progressSeam (applyEventProgress) only writes a milestone_progress row when a matching milestone already exists — mirrors the seedMilestone pattern already used in test/progress.test.ts. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The old parseStructuredSummary-based skip check would permanently fail for every future PR summary once the prompt moves to plain prose (no **What changed:** marker) — Sync would re-summarize every closed PR on every run forever. structuredCount is renamed prSummarizedCount to match the new, text-shape-independent semantics. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Mirrors the existing pr_summaries join used for previousActivity. todoCard rendering is unchanged/out of scope for this change — this makes the summary available on the DTO end-to-end. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Mirrors the PR loop's resummarize-unless-done / budget-check / call / pace pattern, limited to issues that currently have an assignee. Shares ONE AI-call budget with the PR loop rather than a separate allowance — PRs are processed first, so a large backlog spends the whole budget on PRs before any issue gets summarized; that's expected and self-corrects across Sync clicks. Also fixes two pre-existing backfill tests that used the assigned openIssue fixture without an explicit issueSummarizer stub: the new issue loop now summarizes that assigned issue too, so those tests needed an injected stub (never hit env.AI) and, in one case, an updated shared-budget assertion. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The Sync modal tracked only PR summarization (structuredCount/prsTotal). Now that Sync also summarizes issues (src/tools/backfill.ts), it shows "N of M PRs summarized" and "N of M issues summarized" as two independent bars. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
"Issues are never summarized" is no longer true — update My Work's description to match the new issue_summaries capture path. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The issue-summarization loop counted a fallback-to-excerpt AI failure as "done", unlike the PR loop it mirrors — the Sync UI's issue progress bar could show 100% while masking an AI failure the PR bar would correctly surface. Also cleans up a stale "skip-if-structured" comment, disambiguates the shared summarizer's failure log, and updates CLAUDE.md's migrations list (0016, 0017 were both missing). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 48 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (20)
✨ Finishing Touches🧪 Generate unit tests (beta)
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 |
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
canopy | 21c656d | Commit Preview URL Branch Preview URL |
Jul 04 2026, 04:57 PM |
1 participant
Summary
**What changed:**/**Why:**markdown format to 2-3 sentences of richer plain prose that leads with the concrete change and grounds detail in the PR body.issue_summariestable) that generates a plain restatement of an issue plus what needs doing — only where the issue actually supports it — at capture time when the issue is assigned.MyWorkTodo.summary).model !== 'excerpt'.Test plan
npm run typecheck— cleannpm test— 347/347 passing (50 files)npm run build:web— succeedsDesign spec:
docs/superpowers/specs/2026-07-04-worker-summarizer-prompts-design.mdImplementation plan:
docs/superpowers/plans/2026-07-04-worker-summarizer-prompts-plan.md🤖 Generated with Claude Code