Workflow kit that makes Agents ask before they build or edit
Skill Claude CodeCodex
Use after a human-approved release. Closes the active slice, preserves history through the archive index, refreshes living candidates, and waits for the next slice approval.
Skill Claude CodeCodex
Use after QA finds blockers. Fixes only approved blockers within the approved scope and prepares an independent QA re-test.
Skill Claude CodeCodex
Use on first run or when active product context is missing. Asks up to five questions, creates short active context, proposes living slice candidates, and stops for approval.
Skill Claude CodeCodex
Use when the project has no confirmed stack. Makes one pragmatic recommendation, records it only after approval, and preserves progressive context.
Skill Claude CodeCodex
Use after independent QA passes. Makes a concise, traceable release recommendation and waits for human approval.
Skill Claude CodeCodex
Use after implementation to independently verify the approved spec with proportional, evidence-based QA before release recommendation.
Skill Claude CodeCodex
Use after a human approves a slice. Produces a compact, build-ready spec with traceable acceptance and QA checks, then stops for spec approval.
Skill Claude CodeCodex
Use to turn active product direction into one to three small living slice candidates and recommend one. Does not write a full spec or code.
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: