context-compress

To reduce LLM context token usage, compress dense documentation into a token-efficient, layered Context Bible that preserves semantic fidelity while achieving over 60% token reduction.

The best way to load large documentation progressively is using a three-layer architecture (L0 Executive Kernel, L1 Core Foundation, L2 Detailed Specs) with a machine-readable index for quick retrieval.

A canonical dictionary unifies terminology across multiple documents by assigning canonical IDs and performing cross-document synthesis to resolve conflicts and maintain consistent references.

Yes, you can use glob patterns to compress multiple documentation files at once, as the workflow supports inputs from individual files, directories, or glob patterns to generate the Context Bible.

Yes, documentation compression preserves regulatory constraints and version limits by using a deterministic, template-driven workflow that strictly maintains all original boundaries and semantic details.

You retrieve detailed specs without bloating the active LLM context by using on-demand references and assets support, which allows deep dives into specific layers only when required.