compress-skill

Compress oversized SKILL.md files to under 200 lines while preserving functional rules and examples.

3|Updated Apr 5, 2026
One-click install
npx skills add https://github.com/dvy1987/agent-loom --skill compress-skill
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: compress-skill
Source: https://github.com/dvy1987/agent-loom/tree/main/.agents/skills/compress-skill
Command: npx skills add https://github.com/dvy1987/agent-loom --skill compress-skill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill reduces verbose SKILL.md files to a concise, high-impact form so agents can operate more reliably and avoid context-window noise while preserving all functional behavior.

Core Features & Use Cases

  • Rule-preserving compression: Identifies and keeps hard gates, gotchas, output formats, routing triggers, and at least one example.
  • Structured classification and extraction: Moves BACKGROUND and EDGE_CASE content into references/, collapses redundant prose, and converts workflows to one-liners.
  • Safe split decision: Invokes split-skill when genuinely core content prevents safe compression, and validates with regression checks before committing.
  • Use Case: Automatically shrink an oversized skill after an edit or when AGENTS.md triggers compression to keep a living skill library fast and reliable.

Quick Start

Compress the target SKILL.md to under 200 lines while preserving hard gates, gotchas, output format, and at least one example and run the regression checks before committing.

Frequently Asked Questions about compress-skill

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

FAQPage Schema
How do I compress a SKILL.md file to under 200 lines without losing functional behavior?

You can compress a SKILL.md file by collapsing redundant prose into one-liners while preserving hard gates, gotchas, output formats, routing triggers, and at least one example. Background and edge case content is moved into references.

What should I keep when reducing markdown documentation to avoid breaking agent routing?

When reducing markdown documentation, you must keep hard gates, gotchas, output formats, routing triggers, and at least one example. This ensures agents operate reliably without context-window noise.

When should I split a skill instead of compressing its markdown file?

You should split a skill instead of compressing its markdown file when genuinely core content prevents safe compression. The compression process invokes split-skill automatically after determining the file cannot be safely reduced below 200 lines.

How do I automate skill documentation compression after editing a file?

Automated skill documentation compression triggers automatically after an edit or when AGENTS.md requests compression. It validates the compressed file with regression checks before committing to ensure functional behavior remains intact.

Does compressing an oversized SKILL.md move background content to another location?

Yes, compressing an oversized SKILL.md moves BACKGROUND and EDGE_CASE content into the references/ directory. This structured classification and extraction preserves necessary context while keeping the active skill concise.