report-review-run
Use when the PR review harness starts or ends a phase or loop, records a pushed fix, or posts the end-of-run GitHub report comment, or when the user says /report_review_run.
- License 许可
- MIT
- 所属仓库
- sazlin/loadout
- 最近更新
- 2026-11-14
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- report-review-run
- metadata
- { "loadout.sha": "49eb9bd", "loadout.source": "skills/report-review-run/SKILL.md", "loadout.managed": "true" }
- description
- Use when the PR review harness starts or ends a phase or loop, records a pushed fix, or posts the end-of-run GitHub report comment, or when the user says /report_review_run.
02 资源文件 1
- 脚本scripts/review_run_report.py26.3 KB
03 SKILL.md 正文
Report review run
Record wall-clock timings in REVIEW_RUN.json and render the final PR
comment from that log. Do not invent durations. Do not hand-write the
report body.
When to use
review_orchestratorstarts or ends a panel loop, resolve wave, verify loop, risk classification, or merge wrap-up- After
issue_resolvercommits land on PR head - At the end of a harness run (ok, blocked, or aborted)
- The user asks
/report_review_run
Do not use to review, fix, or merge. Do not commit REVIEW_RUN.json.
Steps
- Script:
python3 .claude/skills/report-review-run/scripts/review_run_report.py. If.claude/is missing, useskills/report-review-run/scripts/review_run_report.py. Run--helpfor flags. Default file isREVIEW_RUN.jsonin the project root. - Start a step before the work:
begin --section "Panel Review" --label "loop 1". Sections:Panel Review,Resolve Issues,Verifiers,Risk Classification,Merge. Labels:loop N,wave N, or aTASK-00Nid.begincloses any still-open step. - End a step when that work returns:
end. - Record a pushed fix after a clean cherry-pick/push:
change --sha <short> --task TASK-00N --summary "..." --path <file>:+N,-M(repeat--path). Take N/M fromgit show --numstat --format= <sha>. Source files only. Never logTASKS_TO_RESOLVE-*.mdorREVIEW_HISTORY.md. - Stage cells before render, matching the Decision/Aborted table:
stage --panel "✅|2 loops" --resolve "✅|3 tasks" --verifiers "✅|4/4" --risk "🟢|\low`" --merge "✅|done"`. - Dashboard:
dashboard https://cursor.com/agents/<id>oncerun-inforeturns. Never invent an id. - Render and post at the end of the run, including abort:
render --out /tmp/review-run-report.mdthengh pr comment <n> --body-file /tmp/review-run-report.md. Never--edit-last. Never paste a hand-written stand-in. - Delete
REVIEW_RUN.jsonafter the comment is posted (or after max comment retries). Missing file is a no-op.
Guardrails
- Never invent timestamps or durations
- Never skip
begin/endaround a phase or loop - Never post a Run report that did not come from
render - Never commit
REVIEW_RUN.json - The mermaid fence in
renderstdout is already closed. Do not wrap it in anothermermaidfence
04 同仓库更多 Skills
demo
Demo skill
writing-skills
Use when creating new skills, editing existing skills, or verifying skills work before deployment
writing-plans
Use when you have a spec or requirements for a multi-step task, before touching code
verification-before-completion
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
using-superpowers
Use when starting any conversation - establishes how to find and use skills, requiring skill invocation before ANY response including clarifying questions
using-git-worktrees
Use when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback