claude-md-progressive-disclosurer

Structure CLAUDE.md using a two-level progressive disclosure architecture with Level 1 essentials and Level 2 references. This separates critical commands from detailed SOPs, optimizing information density so the LLM accesses depth only when explicitly triggered by errors or questions.

Progressive disclosure in prompt engineering is an architecture that organizes documentation into essential rules and detailed references. It enforces a reference-driven workflow with mandatory trigger conditions, allowing LLMs to reason about where to read details only when needed during long conversations.

Trigger detailed SOP reads by codifying explicit trigger conditions within the CLAUDE.md architecture. The system uses a trailing trigger index to guide long conversations, ensuring Level 2 references are fetched only when specific errors occur or questions arise.

Yes, progressive disclosure works for team prompt guidelines and debugging workflows by providing a two-tier architecture. It maintains clarity and traceability across multiple entry points, balancing quick access to essential commands with maintainable detailed edge cases.

Avoid a two-tier progressive disclosure architecture when documentation prompts are lightweight and lack complex SOPs or reference materials. It is designed for documentation-heavy prompts where rules must be quickly accessible yet maintainable, adding unnecessary overhead to simple prompt structures.