code reviewby frank890417
🧬 Taiwan.md — PR Review
frank890417/twmd-pr-review
Review a Taiwan.md PR via canonical MAINTAINER-PIPELINE. TRIGGER when: user says "審 PR", "Review PR", "PR #N", "免疫巡邏", "跑 PR review".
What it changes
Passed 13 of 19 test cases with the skill, and 6 of 19 without it.
Without skill32%
With skill68%
Checks it passed that the model alone missed
- Executes gh pr merge
- Includes the DNA emoji 🧬 signature
- Uses gh pr comment to deliver the response
and 7 more in the full report
8 fixed7 passed either way7 still failing0 broke
- Pass-rate lift
- +36 pp
- Tokens
- -22%
- Agent turns
- 0%
- Response time
- -41%
Model gemini-3.6-flashMeasured 8/15/2026v115 of 22 cases passed with skill · 19 comparable
See it in action
One of 8 test cases the skill fixed. Same model, same prompt, with and without the skill.
PromptWhich repository document defines the core PR review strategy, 30-second triage rules, and decision flags for this project?
What was checked, without and with the skill| Checked | Without skill | With skill |
|---|
| References docs/pipelines/MAINTAINER-PIPELINE.md | Missed | Passed |
|---|
What's inside
The instructions your model reads when the skill loads.
SKILL.md13 linessingle file
🧬 Taiwan.md — PR Review
-
你是 Taiwan.md（簽名 🧬）。如未甦醒先跑 /twmd-become。
-
嚴格完整讀取並執行 docs/pipelines/MAINTAINER-PIPELINE.md §PR 審核策略（含 §1b Git merge 優先）。
-
依該 pipeline 的三級判斷（merge / merge+heal / leave open / close）+ gh pr comment 用貢獻者語言感謝。
-
HARD — merge-first-then-heal：ship 時先 gh pr merge（PR 必須變 MERGED），再 main heal。禁止 content 進 main 後 gh pr close 當收割（2026-07-23 哲宇校正）。
故意最小化。30 秒判斷 / 紅黃旗 / Checklist / 翻譯 PR 上游檢查 / 合併策略 / §1b 優先序全部在 pipeline canonical。
Track record
Benchmark
13 / 19
test cases passed, vs 6 / 19 without
Adoption
- Users
- 0
- Times used
- 0
Safety and source details
SkillSafety
✓ 3 of 3 passed✓Code scan15 of 15 checks clean
0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026
View 15 scanner checks
live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing
✓IntentSafe
LLM judge · does the skill do what it claims
✓ContentClean
LLM judge · policy & harmful-content review
Source
ImportedImported 8/3/2026Version v1