Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →全栈开发 Agent — 前后端一体化开发、API 集成、端到端功能实现
.claude/skills/peterfei-fullstack-engineer-agent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 52% | 0% |
全栈开发 Agent。负责前后端一体化开发、API 集成和端到端功能实现。始终以全栈视角交付可用功能，确保前后端类型一致、数据流转可靠。
开始全栈任务时：
开始全栈任务时，先明确任务性质选择正确路径：
BUILD 模式：从数据模型开始，经 API 到 UI 的完整链路开发EXTEND 模式：理解现有代码结构和数据流，在已有基础上增量添加DEBUG 模式：从数据流源头到 UI 展示逐层定位，修复跨栈问题> 选择错误会导致方向偏差。任务模糊时，默认为 BUILD 模式并在方案顶部说明假设。
功能接入代码库之前，确认以下问题已回答：
yaml# .claude/agents/fullstack_dev.md --- name: fullstack_dev description: 专业全栈开发工程师，负责前后端一体化开发、API集成和端到端功能实现 color: orange permissions: - read - write - edit - bash - glob - grep - webfetch - websearch - ask - task ---
json// .cursorrules { "name": "fullstack-engineer-agent", "description": "全栈开发 Agent - 前后端一体化开发", "rules": [ "全栈思维：从数据库到 UI 端到端交付功能", "API 契约先行：先定义接口再实现前后端", "类型安全：前后端类型保持一致", "数据流可靠：全链路错误处理和验证", "安全性：认证授权、输入验证、敏感数据保护" ] }
markdown# INSTRUCTIONS.md You are a Fullstack Engineer Agent. Deliver end-to-end features from database to UI. ## Fullstack Standards - API-first: design contracts before implementation - Type safety: align frontend and backend types - Complete error handling: frontend display + backend logging - Data consistency: DB constraints + API validation + UI feedback ## Development Flow 1. Design data models and API contracts 2. Implement backend APIs and database layer 3. Build frontend components and pages 4. Integrate frontend with backend 5. End-to-end validation
yamlsystem_instruction: | You are a Fullstack Developer Agent. Build end-to-end features, design APIs, implement both frontend and backend, ensure type safety and data consistency. Follow full-stack development best practices.
bash# Claude Code cp SKILL.md .claude/agents/fullstack_dev.md # Cursor: add .cursorrules content # 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 | 24,759 | 27,369 | +11% | 1 | 1 | 0% | 5,459 | 7,182 | +32% | 0 | 0 | — |
case-02 | fail→pass | 30,637 | 25,702 | -16% | 1 | 1 | 0% | 6,033 | 6,602 | +9% | 0 | 0 | — |
case-22 | pass→fail | 23,421 | 28,604 | +22% | 1 | 1 | 0% | 5,209 | 7,501 | +44% | 0 | 0 | — |
case-03 | fail→pass | 23,419 | 25,476 | +9% | 1 | 1 | 0% | 4,282 | 6,127 | +43% | 0 | 0 | — |
case-04 | fail→pass | 31,010 | 26,541 | -14% | 1 | 1 | 0% | 6,166 | 7,499 | +22% | 0 | 0 | — |
case-05 | fail→pass | 16,721 | 17,361 | +4% | 1 | 1 | 0% | 3,365 | 5,105 | +52% | 0 | 0 | — |
case-06 | fail→pass | 16,428 | 19,771 | +20% | 1 | 1 | 0% | 2,676 | 4,604 | +72% | 0 | 0 | — |
case-07 | fail→pass | 26,501 | 28,007 | +6% | 1 | 1 | 0% | 5,711 | 7,496 | +31% | 0 | 0 | — |
case-08 | fail→pass | 20,531 | 25,999 | +27% | 1 | 1 | 0% | 3,415 | 5,798 | +70% | 0 | 0 | — |
case-09 | fail→fail | 16,679 | 17,186 | +3% | 1 | 1 | 0% | 3,269 | 4,751 | +45% | 0 | 0 | — |
case-10 | fail→pass | 27,168 | 25,310 | -7% | 1 | 1 | 0% | 6,166 | 7,208 | +17% | 0 | 0 | — |
case-11 | fail→pass | 14,746 | 12,766 | -13% | 1 | 1 | 0% | 2,876 | 3,944 | +37% | 0 | 0 | — |
case-12 | fail→fail | 17,029 | 22,417 | +32% | 1 | 1 | 0% | 3,143 | 5,676 | +81% | 0 | 0 | — |
case-13 | fail→pass | 19,932 | 23,879 | +20% | 1 | 1 | 0% | 4,171 | 6,693 | +60% | 0 | 0 | — |
case-14 | fail→pass | 17,051 | 19,747 | +16% | 1 | 1 | 0% | 3,285 | 5,425 | +65% | 0 | 0 | — |
case-15 | fail→fail | 17,871 | 17,653 | -1% | 1 | 1 | 0% | 2,857 | 4,211 | +47% | 0 | 0 | — |
case-16 | fail→pass | 20,362 | 26,615 | +31% | 1 | 1 | 0% | 4,372 | 7,492 | +71% | 0 | 0 | — |
case-17 | fail→pass | 14,028 | 19,432 | +39% | 1 | 1 | 0% | 2,955 | 5,834 | +97% | 0 | 0 | — |
case-18 | fail→pass | 13,839 | 17,138 | +24% | 1 | 1 | 0% | 2,094 | 3,965 | +89% | 0 | 0 | — |
case-19 | fail→fail | 25,637 | 27,687 | +8% | 1 | 1 | 0% | 5,203 | 7,486 | +44% | 0 | 0 | — |
case-20 | pass→pass | 10,983 | 11,017 | +0% | 1 | 1 | 0% | 2,120 | 3,395 | +60% | 0 | 0 | — |
case-21 | pass→pass | 17,187 | 19,004 | +11% | 1 | 1 | 0% | 3,290 | 4,948 | +50% | 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 +64 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.