Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →自动同步项目中的 Agents.md、claude.md 和 gemini.md 文件，保持内容一致性。支持自动监听和手动触发。
.claude/skills/bilal140202-doc-sync-tool/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 43% | 0% |
这个工具用于自动同步项目中的 AI Agent 配置文档，确保 Agents.md、claude.md 和 gemini.md 三个文件内容保持一致。
bashcd /Users/ben/Downloads/go\ to\ wild/auto-website-system/_skills/doc-sync-tool pnpm install
bash# 在项目根目录执行 node /Users/ben/Downloads/go\ to\ wild/auto-website-system/_skills/doc-sync-tool/sync.js # 或者使用 npm script pnpm run sync
bash# 启动文件监听服务 node /Users/ben/Downloads/go\ to\ wild/auto-website-system/_skills/doc-sync-tool/watch.js # 或者使用 npm script pnpm run watch
bash# 使用 PM2 在后台运行 pm2 start /Users/ben/Downloads/go\ to\ wild/auto-website-system/_skills/doc-sync-tool/watch.js --name doc-sync # 查看状态 pm2 status # 停止服务 pm2 stop doc-sync
Agents.md、claude.md、gemini.md 文件可以在 sync.js 中修改以下配置：
javascriptconst CONFIG = { targetFiles: ['Agents.md', 'claude.md', 'gemini.md'], // 目标文件列表 scanPath: process.cwd(), // 扫描路径（默认当前目录） excludeDirs: ['node_modules', '.git', '.next', 'dist'] // 排除目录 };
node_modules、.git、.next、dist 等目录bashchmod +x sync.js watch.js
需要 Node.js 14+ 版本
bashrm -rf node_modules package-lock.json pnpm install
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | fail→pass | 5,084 | 3,344 | -34% | 1 | 1 | 0% | 933 | 1,420 | +52% | 0 | 0 | — |
case-19 | pass→pass | 9,269 | 9,871 | +6% | 1 | 1 | 0% | 1,582 | 2,391 | +51% | 0 | 0 | — |
case-01 | fail→pass | 20,191 | 2,561 | -87% | 1 | 1 | 0% | 1,756 | 1,285 | -27% | 0 | 0 | — |
case-02 | fail→pass | 12,644 | 6,071 | -52% | 1 | 1 | 0% | 2,490 | 1,974 | -21% | 0 | 0 | — |
case-03 | pass→pass | 15,648 | 11,214 | -28% | 1 | 1 | 0% | 2,611 | 2,626 | +1% | 0 | 0 | — |
case-04 | fail→pass | 9,860 | 5,285 | -46% | 1 | 1 | 0% | 1,576 | 1,732 | +10% | 0 | 0 | — |
case-09 | fail→pass | 7,453 | 6,401 | -14% | 1 | 1 | 0% | 1,381 | 1,970 | +43% | 0 | 0 | — |
case-05 | fail→pass | 12,139 | 7,108 | -41% | 1 | 1 | 0% | 1,788 | 2,144 | +20% | 0 | 0 | — |
case-06 | fail→fail | 12,220 | 5,302 | -57% | 1 | 1 | 0% | 1,672 | 1,829 | +9% | 0 | 0 | — |
case-07 | pass→pass | 7,938 | 2,298 | -71% | 1 | 1 | 0% | 1,359 | 1,127 | -17% | 0 | 0 | — |
case-08 | pass→pass | 2,972 | 3,378 | +14% | 1 | 1 | 0% | 558 | 1,172 | +110% | 0 | 0 | — |
case-11 | fail→pass | 11,836 | 2,274 | -81% | 1 | 1 | 0% | 1,725 | 1,180 | -32% | 0 | 0 | — |
case-12 | fail→pass | 6,654 | 2,388 | -64% | 1 | 1 | 0% | 915 | 1,205 | +32% | 0 | 0 | — |
case-13 | pass→pass | 6,545 | 2,527 | -61% | 1 | 1 | 0% | 1,216 | 1,183 | -3% | 0 | 0 | — |
case-14 | fail→pass | 9,691 | 3,609 | -63% | 1 | 1 | 0% | 1,621 | 1,466 | -10% | 0 | 0 | — |
case-15 | fail→pass | 5,049 | 4,236 | -16% | 1 | 1 | 0% | 818 | 1,485 | +82% | 0 | 0 | — |
case-16 | fail→pass | 4,255 | 3,318 | -22% | 1 | 1 | 0% | 760 | 1,245 | +64% | 0 | 0 | — |
case-17 | fail→pass | 7,780 | 7,596 | -2% | 1 | 1 | 0% | 1,483 | 2,200 | +48% | 0 | 0 | — |
case-18 | pass→pass | 8,305 | 3,964 | -52% | 1 | 1 | 0% | 1,358 | 1,464 | +8% | 0 | 0 | — |
case-20 | pass→pass | 11,016 | 5,831 | -47% | 1 | 1 | 0% | 1,766 | 1,794 | +2% | 0 | 0 | — |
case-21 | pass→pass | 5,191 | 1,971 | -62% | 1 | 1 | 0% | 795 | 1,094 | +38% | 0 | 0 | — |
case-22 | pass→pass | 8,286 | 2,199 | -73% | 1 | 1 | 0% | 1,295 | 1,136 | -12% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +55 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.