bmad-distillator

Compress multiple source documents into a single lossless distillate with token estimation.

Updated Mar 18, 2026
One-click install
npx skills add https://github.com/robotics-playground/skills --skill bmad-distillator-robotics-playground
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bmad-distillator
Source: https://github.com/robotics-playground/skills/tree/main/_bmad/core/skills/bmad-distillator
Command: npx skills add https://github.com/robotics-playground/skills --skill bmad-distillator-robotics-playground

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Distillate documents into a single lossless distillate that preserves every fact, decision, constraint, and relationship while removing overhead for downstream LLM workflows.

Core Features & Use Cases

  • Lossless compression of multiple source documents into a single, token-efficient distillate.
  • Stage-based processing (Analyze, Compress, Verify & Output, and optional Round-Trip Validate) to handle large inputs reliably.
  • Suitable for PRD creation, architecture synthesis, and any long-context LLM workflow requiring faithful fidelity.

Quick Start

Provide input source documents and optional downstream_consumer to generate a lossless distillate next to the sources.

Frequently Asked Questions about bmad-distillator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I compress multiple source documents into a single lossless distillate for LLM context?

Lossless document distillation preserves every fact, decision, constraint, and relationship from multiple source documents while removing overhead, ensuring high-fidelity inputs for downstream LLM workflows like PRD creation and architecture synthesis.

How does stage-based processing handle large sets of source documents for token compression?

The distillation process uses semantic splitting and stage-based processing across Analyze, Compress, Verify & Output, and optional Round-Trip Validate stages to handle large source sets and deliver predictable token footprints for downstream tasks.

Can I estimate the token footprint of a compressed document before running downstream LLM tasks?

Yes, the Skill provides token estimation and strict frontmatter metadata alongside the compressed distillate, allowing you to predict and control the token footprint before executing downstream LLM tasks.

Does lossless distillation work for PRD creation and architecture synthesis workflows?

Yes, lossless distillation is suitable for PRD creation and architecture synthesis, compressing multiple source documents into a token-efficient format that maintains faithful fidelity for long-context LLM workflows.

What is the best way to verify that document compression did not lose any facts or constraints?

The optional Round-Trip Validate stage verifies that document compression did not lose any facts or constraints by validating the generated distillate against the original source documents.

Do I need a downstream consumer specified to generate a lossless distillate from my sources?

No, specifying a downstream consumer is optional when generating a lossless distillate from source documents, but providing it helps tailor the compression output to meet specific downstream LLM workflow requirements.