What problem does it solve?
This Skill eliminates the token bloat and information loss of manual document summarization for LLM workflows, preserving every fact, decision, constraint, and relationship from source documents while stripping unnecessary human-oriented prose to maximize context efficiency.
Core Features & Use Cases
- Lossless Compression: Preserves all critical information from source documents unlike lossy summaries, ensuring no data is missing for downstream workflows.
- Multi-Document Support: Handles single or multiple source files with automatic routing for small sets and fan-out compression for large document collections, plus optional semantic splitting for very large content.
- Validation & Quality Assurance: Includes optional round-trip reconstruction testing to confirm zero information loss, ideal for high-stakes use cases like PRD creation or architecture design.
- Use Case: If you have a 25-page product brief paired with 15 pages of discovery notes, use this skill to create a single dense distillate that a downstream LLM can consume as full context to generate a complete PRD without missing any key requirements or decisions.
Quick Start
Use the bmad-distillator skill to compress your source documents into a lossless, token-efficient distillate for your downstream LLM workflow.