编程开发

Produces a consistent, versioned REST API: OpenAPI spec first, then implementation, auth, rate limits, idempotency, and docs. Great for new services.

Generates a robust Python CLI with argparse, logging, retries, dry-run, and typed config — ideal for automation, data pipelines, and internal ops scripts.

(golden-master / snapshot tests), then run them against the new version. Non-goals: do…

or adapter) so old and new can coexist.

elsewhere. List all call sites that must change.

quirks, bugs, implicit type coercions, and ordering side effects. Flag which behaviors…

Use when migrating old code (e.g. jQuery → React, Python 2 → 3, callbacks → async/await, class components → hooks) while preserving behavior exactly.

matrix covering normal, boundary, and invalid cases. 4. Write tests using {{pytest /…

Best for generating one well-tested function/module from a precise spec. The deliberate-questions-first pattern reduces hallucinated APIs.

request/response JSON, status codes), state management, and error handling. 3. Wait for…

and performance expectations. Do not guess — list every assumption explicitly.

Use this when you need a complete, production-ready feature end to end (database schema, API, validation, frontend, tests). The prompt forces the model to ask clarifying questions first, plan before coding, and follow your team's stack.

生成符合 Conventional Commits 的提交信息和 PR 描述，便于团队协作与回溯。

模拟算法/系统设计面试，中文追问、压力测试与复盘。

先测量再优化：给出各语言/系统的分析工具和证据驱动的优化顺序。

理清 Promise 并发、事件循环，以及让 TS 类型精确又不晦涩。

Python 项目从脚本到工程：依赖、类型、异步、打包。

E2E 与接口测试选型、数据隔离与防脆弱测试。

教你写出有意义的测试：用例设计、命名、mock 边界。

用 MQ 解耦时：顺序、重复、积压、丢失四大问题的工程对策。

Redis 使用模式：键设计、过期策略、分布式锁、常见事故预防。

面试或实战的高并发系统设计：容量估算、缓存、队列、降级。

单体要拆分或规划微服务时：按业务能力划分、定边界与通信方式。

从提交到部署的流水线：测试门、构建缓存、灰度与回滚。