Coordinate cross-AI code review across Claude Code, Codex, and Gemini.
npx skills add https://github.com/ValorVie/custom-skills --skill discuss-multi-ai
Please help me install this Agent Skill. Skill: discuss-multi-ai Source: https://github.com/ValorVie/custom-skills/tree/main/skills/discuss-multi-ai Command: npx skills add https://github.com/ValorVie/custom-skills --skill discuss-multi-ai
在多個 AI 工具（如 Claude Code、Codex、Gemini 等）之間進行程式碼變更的交叉審議，透過標準化流程與可追蹤的紀錄，提升審議的一致性與透明度。
提交變更內容或 commit hash 以啟動跨 AI 審議並產生交接文字。
High-intent search queries and answers about installing and using this skill.
Cross-AI code review standardizes handoffs between tools like Claude Code and Gemini by generating structured review summaries and traceable documentation for each commit. It ensures consistent messaging and documented decisions across multiple tools.
To start a multi-AI code review, paste your code changes or a specific commit hash to trigger the deterministic workflow. The system automatically detects the current AI identity and generates copyable handoff text for the next AI tool.
Yes, the workflow evaluates each change for logical correctness, side effects, security, and project convention consistency. It automatically records these review criteria and documented decisions in standardized files under the docs/discuss-ai/ directory.
Yes, the workflow features automatic identity detection to identify whether the executing environment is Claude, Codex, Gemini, or another tool. This ensures clear routing and traceable handoffs when passing the discussion to a different AI.
The system uses a standardized documentation format placed under the docs/discuss-ai/ directory to trace review history. These files include specific fields like source commit, reviewer identity, and date to ensure traceable decisions and change tracking.
wikimediaFetch a Gerrit change by ID and generate a structured code review.
dusktreaderGenerate structured handoff documents as .pending.md files for AI agent transfers.
RYN6666999Route major code changes into pending decision queues for human approval.
HsnSaboorPackage code repositories into single XML, Markdown, JSON, or plain text files.
