rules

This Skill enforces a strict file whitelist that blocks agent-created .md, .json, and script files outside approved locations, allowing only essential code files (.tsx, .py, .css, .prisma) and orchestrator-controlled documentation in .build/. This maintains project hygiene by stopping pollution before writes occur.

Agents can create code files (.tsx, .py, .css, .prisma) and standard configuration files. All project documentation (.md files) must be created by the orchestrator in .build/ as centralized sources of truth, preventing redundant or scattered documentation across the project root.

This Skill routes all documentation creation to the orchestrator within .build/, enforcing files like context.md, timeline.md, tasks.md, issues.md, specs.md, and decisions/*.md. When agents attempt unauthorized .md creation, the Skill blocks it and instructs structured information return to the orchestrator.

The Skill intercepts the write before execution, stops the unauthorized file creation, and routes structured violation information to the orchestrator for manual review. This prevents non-standard configurations and unnecessary scripts from entering your codebase.

Yes. The Skill applies different rules: it permits orchestrator-written items like context.md and decisions/*.md in .build/, while blocking equivalent agent-created files. This dual enforcement ensures clean separation between approved centralized documentation and agent outputs.

Yes. By strictly controlling which files enter .build/ and preventing arbitrary .md, .json, and script proliferation, this Skill eliminates redundant documentation and non-standard configurations. The result is a cleaner codebase with reduced technical debt and enforced project structure.