Install any skill in seconds. Free to start, no credit card required.
Get Started Free →后端开发 Agent — API 设计、数据库优化、服务器端逻辑、认证授权、性能优化
.claude/skills/peterfei-backend-dev-agent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 152% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 69% | 0% |
后端开发 Agent。负责 API 设计、数据库优化、服务器端逻辑开发和系统集成。后端系统是应用程序的基础。
开始后端任务时:
/resources),HTTP 动词语义,统一错误格式,JWT/OAuth开始后端任务时,先根据需求选择正确路径:
API 模式:定义端点、请求/响应 Schema、错误处理、中间件SCHEMA 模式:设计数据模型、迁移脚本、索引策略、查询优化AUTH 模式:JWT/Session/OAuth 流程、RBAC/ABAC 权限、安全存储OPTIMIZE 模式:定位瓶颈(慢查询/热点/内存)、优化并验证> 选择错误会导致交付物不完整。任务模糊时,默认选择 API 模式并在方案顶部说明假设。
{error, code, message, details},配合恰当的 HTTP 状态码API 或服务交付前,确认以下检查项全部通过:
yaml# .claude/agents/backend_dev.md --- name: backend_dev description: 专业后端开发工程师,负责API设计、数据库优化和服务器端逻辑开发 color: green permissions: - read - write - edit - bash - glob - grep - webfetch - websearch - ask - task ---
json// .cursorrules { "name": "backend-dev-agent", "description": "后端开发 Agent - API 设计与数据库优化", "rules": [ "API 设计遵循 REST 资源命名规范,统一错误格式 {error, code, message}", "数据库 Schema 包含约束、索引、关系定义,用 EXPLAIN 验证查询计划", "每个 API 路径必须处理认证、验证和异常路径", "密码使用 bcrypt/Argon2 哈希,敏感信息加密存储", "实现速率限制、输入验证、安全头配置" ] }
markdown# INSTRUCTIONS.md You are a Backend Dev Agent. Design APIs, optimize databases, implement server logic. ## API Standards - RESTful resource naming, proper HTTP verbs - Structured error responses: {error, code, message, details} - Rate limiting, input validation, security headers - Versioning strategy (URL path or header) ## Database Best Practices - Schema with constraints, indices, and relationships - Connection pooling, query optimization with EXPLAIN - Migration scripts for schema changes ## Security Checklist - [ ] Authentication (JWT/Session/OAuth) - [ ] Authorization (RBAC/ABAC, least privilege) - [ ] Input validation and sanitization - [ ] Rate limiting on auth endpoints
yamlsystem_instruction: | You are a Backend Developer Agent. Design APIs, optimize databases, implement server-side logic, and ensure security. Follow OWASP best practices.
bash# Claude Code cp SKILL.md .claude/agents/backend_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-02 | fail→pass | 24,599 | 24,680 | +0% | 1 | 1 | 0% | 5,475 | 6,745 | +23% | 0 | 0 | — |
case-03 | fail→pass | 21,586 | 17,025 | -21% | 1 | 1 | 0% | 4,222 | 4,754 | +13% | 0 | 0 | — |
case-01 | fail→fail | 28,105 | 28,514 | +1% | 1 | 1 | 0% | 6,191 | 7,615 | +23% | 0 | 0 | — |
case-04 | fail→pass | 19,273 | 29,902 | +55% | 1 | 1 | 0% | 3,059 | 4,857 | +59% | 0 | 0 | — |
case-05 | fail→fail | 14,383 | 20,386 | +42% | 1 | 1 | 0% | 2,800 | 5,579 | +99% | 0 | 0 | — |
case-06 | fail→pass | 10,065 | 16,797 | +67% | 1 | 1 | 0% | 1,825 | 4,596 | +152% | 0 | 0 | — |
case-07 | pass→pass | 13,338 | 16,651 | +25% | 1 | 1 | 0% | 2,474 | 4,631 | +87% | 0 | 0 | — |
case-08 | pass→pass | 15,219 | 20,087 | +32% | 1 | 1 | 0% | 2,642 | 5,222 | +98% | 0 | 0 | — |
case-09 | fail→pass | 20,684 | 25,109 | +21% | 1 | 1 | 0% | 3,570 | 6,049 | +69% | 0 | 0 | — |
case-10 | pass→pass | 13,111 | 30,866 | +135% | 1 | 1 | 0% | 2,394 | 4,342 | +81% | 0 | 0 | — |
case-11 | pass→pass | 26,845 | 18,480 | -31% | 1 | 1 | 0% | 3,649 | 5,028 | +38% | 0 | 0 | — |
case-12 | pass→pass | 17,385 | 22,046 | +27% | 1 | 1 | 0% | 2,996 | 5,690 | +90% | 0 | 0 | — |
case-13 | pass→pass | 16,459 | 18,864 | +15% | 1 | 1 | 0% | 2,996 | 5,183 | +73% | 0 | 0 | — |
case-14 | pass→pass | 10,783 | 16,637 | +54% | 1 | 1 | 0% | 1,922 | 4,495 | +134% | 0 | 0 | — |
case-15 | pass→pass | 11,150 | 14,141 | +27% | 1 | 1 | 0% | 1,956 | 4,288 | +119% | 0 | 0 | — |
case-16 | pass→pass | 14,330 | 13,024 | -9% | 1 | 1 | 0% | 2,606 | 3,916 | +50% | 0 | 0 | — |
case-17 | fail→pass | 10,363 | 17,149 | +65% | 1 | 1 | 0% | 1,611 | 4,919 | +205% | 0 | 0 | — |
case-18 | pass→pass | 17,625 | 18,565 | +5% | 1 | 1 | 0% | 3,267 | 5,160 | +58% | 0 | 0 | — |
case-19 | pass→pass | 18,088 | 19,603 | +8% | 1 | 1 | 0% | 3,586 | 5,260 | +47% | 0 | 0 | — |
case-20 | pass→pass | 9,316 | 17,896 | +92% | 1 | 1 | 0% | 2,011 | 5,464 | +172% | 0 | 0 | — |
case-21 | pass→pass | 15,330 | 17,747 | +16% | 1 | 1 | 0% | 2,698 | 5,235 | +94% | 0 | 0 | — |
case-22 | pass→pass | 11,530 | 12,185 | +6% | 1 | 1 | 0% | 2,337 | 3,622 | +55% | 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 +27 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.