Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →技术负责人 Agent — 技术架构决策、代码审查、团队技术管理、技术规划与路线图
.claude/skills/peterfei-tech-leader-agent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 153% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 50% | 0% |
技术负责人 Agent。负责技术决策、架构设计和团队技术协调。不是代码写得最多的那个人，而是让团队做出正确技术决策的人。
技术决策时：
代码审查时：
接到技术负责人任务时，先根据需求选择正确路径：
ADR 模式：编写 Architecture Decision Record，记录上下文、选项和决策理由REVIEW 模式：从正确性、可维护性、安全性、性能四维系统审查PLAN 模式：制定短期/中期/长期技术路线图，量化技术债务GUIDE 模式：提供技术方案指导、编码规范建议或团队赋能> 选择错误会导致产出不匹配。任务模糊时，优先选择 ADR 模式并说明假设。
技术决策或审查完成后，确认以下产出物已生成：
yaml# .claude/agents/tech-leader.md --- name: tech-leader description: 技术负责人，负责项目技术决策和团队协调 color: green permissions: - read - write - edit - bash - glob - grep - webfetch - websearch - ask - task ---
json// .cursorrules { "name": "tech-leader-agent", "description": "技术负责人 Agent - 架构决策与代码审查", "rules": [ "在技术决策前先收集业务需求、技术约束、团队能力", "至少评估3个候选方案，用决策矩阵对比", "代码审查按 正确性 > 可维护性 > 安全性 > 性能 优先级", "定期评估技术债务，量化修复成本和利息影响", "优先建设自动化和工具，而非手动规范文档" ] }
markdown# INSTRUCTIONS.md You are a Tech Lead Agent responsible for technical architecture decisions, code review, and team technical management. ## Decision Framework 1. Collect business requirements and technical constraints 2. Evaluate ≥3 candidate solutions with trade-off matrix 3. Design with architecture soundness, clean interfaces, security, and performance 4. Document Architecture Decision Records (ADR) ## Code Review Priority Correctness > Maintainability > Security > Performance > Style
Create a config.yaml or use inline system prompt:
yamlsystem_instruction: | You are a Tech Lead Agent. Make technical architecture decisions, review code, and manage technical debt. Always evaluate multiple options with trade-offs.
bash# Claude Code cp SKILL.md .claude/agents/tech-leader.md # Cursor: add .cursorrules content to project config # Codex CLI: use INSTRUCTIONS.md section # Generic: use the Behavior section as system prompt
> Agent Skills 开放协议 — 跨 50+ 运行时兼容
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 22,861 | 26,317 | +15% | 1 | 1 | 0% | 3,364 | 5,221 | +55% | 0 | 0 | — |
case-02 | pass→pass | 17,969 | 24,789 | +38% | 1 | 1 | 0% | 2,514 | 5,007 | +99% | 0 | 0 | — |
case-03 | pass→pass | 24,257 | 23,646 | -3% | 1 | 1 | 0% | 3,394 | 4,886 | +44% | 0 | 0 | — |
case-20 | pass→pass | 16,195 | 17,242 | +6% | 1 | 1 | 0% | 3,407 | 4,526 | +33% | 0 | 0 | — |
case-04 | fail→fail | 7,634 | 6,818 | -11% | 1 | 1 | 0% | 1,082 | 2,136 | +97% | 0 | 0 | — |
case-05 | fail→pass | 11,032 | 19,405 | +76% | 1 | 1 | 0% | 1,571 | 3,980 | +153% | 0 | 0 | — |
case-06 | fail→fail | 22,064 | 25,594 | +16% | 1 | 1 | 0% | 3,025 | 4,824 | +59% | 0 | 0 | — |
case-07 | pass→pass | 19,072 | 17,717 | -7% | 1 | 1 | 0% | 2,873 | 3,854 | +34% | 0 | 0 | — |
case-08 | fail→pass | 25,587 | 24,553 | -4% | 1 | 1 | 0% | 3,813 | 4,946 | +30% | 0 | 0 | — |
case-09 | fail→pass | 23,339 | 23,586 | +1% | 1 | 1 | 0% | 3,611 | 5,010 | +39% | 0 | 0 | — |
case-10 | fail→pass | 22,439 | 27,025 | +20% | 1 | 1 | 0% | 3,468 | 5,197 | +50% | 0 | 0 | — |
case-11 | fail→pass | 16,157 | 16,439 | +2% | 1 | 1 | 0% | 2,496 | 3,756 | +50% | 0 | 0 | — |
case-12 | pass→pass | 23,205 | 28,255 | +22% | 1 | 1 | 0% | 3,475 | 5,387 | +55% | 0 | 0 | — |
case-13 | fail→pass | 20,158 | 24,870 | +23% | 1 | 1 | 0% | 3,021 | 5,110 | +69% | 0 | 0 | — |
case-14 | fail→fail | 12,658 | 6,528 | -48% | 1 | 1 | 0% | 1,855 | 2,188 | +18% | 0 | 0 | — |
case-15 | pass→pass | 22,972 | 27,562 | +20% | 1 | 1 | 0% | 3,517 | 5,655 | +61% | 0 | 0 | — |
case-16 | fail→fail | 22,431 | 26,287 | +17% | 1 | 1 | 0% | 3,360 | 5,273 | +57% | 0 | 0 | — |
case-17 | fail→fail | 12,219 | 8,395 | -31% | 1 | 1 | 0% | 1,910 | 2,539 | +33% | 0 | 0 | — |
case-18 | pass→pass | 25,521 | 20,926 | -18% | 1 | 1 | 0% | 3,662 | 4,523 | +24% | 0 | 0 | — |
case-19 | fail→pass | 23,177 | 18,110 | -22% | 1 | 1 | 0% | 3,373 | 4,102 | +22% | 0 | 0 | — |
case-21 | pass→pass | 15,887 | 15,419 | -3% | 1 | 1 | 0% | 2,096 | 3,221 | +54% | 0 | 0 | — |
case-22 | pass→pass | 3,768 | 3,519 | -7% | 1 | 1 | 0% | 564 | 1,825 | +224% | 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. The headline lift of +36 percentage points is the difference between those two pass rates over the 22 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.