Skill Claude CodeCodex
A procedure for examining a software repository's layout, conventions, build and test commands, linting, and continuous-integration setup.
A workflow-oriented agent system that runs the same set of engineering roles across modern AI coding assistants (opencode, Cursor, Claude Code, VS Code + Copilot).
This repository also configures its own agents. See what aia_kaiwu tells them →
Skill Claude CodeCodex
A procedure for examining a software repository's layout, conventions, build and test commands, linting, and continuous-integration setup.
Skill Claude CodeCodex
Planning. Use when the request is multi-step (>=3 steps), has an architectural choice on the table, touches multiple modules, or the scope is not obvious.
Skill Claude CodeCodex
Multi-session wayfinding for work bigger than one context. Use for foggy, multi-session efforts that need a decision-ticket map before execution.
Skill Claude CodeCodex
Reason one starting point (a raw problem, or a SCAMPER angle) through the Golden Circle (Why to How to What) grounded in first principles, producing one result from that single vantage. Dispatched by job-scamper, or standalone. Triggers - "why how what", "golden circle analysis", "first-principles pass on this".
Skill Claude CodeCodex
Concrete single-scope implementation task. Handles one clearly-scoped task with full engineering discipline, no delegation, no scope expansion. Prefer over task-multi-build when single-scope.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: