@wed840313

wenwei-code-positioning targets code search and traversal from problem statement plus single entry point, converging to minimal critical implementation via component refs, method calls, and cross-layer associations. outputs file paths, line numbers, and core snippets without generating complete module documentation.

wenwei-cosmic automates COSMIC-standard workload decomposition from requirements documents into functional points and subprocess details (E/R/W/X data movements), generating markdown and CSV artifacts through a five-step workflow with sub-agent delegation.

wenwei-context-build recursively traces code call chains from entry points to extract technical implementation details, type signatures, and cross-layer dependency mappings. outputs structured markdown for ai consumption across frontend and backend architectures.

在开发完成后，对代码实现进行评审，结合功能点文档、需求/澄清文档与概要设计文档，评估需求完成度，检查代码改动是否严格按照概要设计落地，识别未完成项、BUG与偏离设计的问题。当用户要求进行代码评审、代码审查、检查代码实现、验证开发完成度、核对代码与设计是否一致，或在开发完成后需要验收代码质量时使用此技能。注意：此技...

在代码提交 git 前，通过 git diff 对比修改内容，逐文件审查变更点，检测潜在 BUG、隐含风险与可优化项，输出结构化审查报告。当用户表达提交前检查、提交检查、commit 前检查、pre-commit review、提前审查代码、检查一下改动、看看能不能提交等意图时使用此技能。注意：此技能与「代码评审...