reference-enrichment

What problem does it solve?

Reference files for agents and skills often stay shallow or incomplete, which makes outputs generic, slow to improve, and hard to validate against progressive-disclosure standards.

Core Features & Use Cases

• Analyze reference coverage: Detects which sub-domains are missing reference files based on what the skill claims it covers.
• Research and compile Level 3 references: Creates concrete, domain-specific reference files with detection commands, version notes, and error-fix mappings.
• Decompose bloated bodies into references: Extracts catalog-like and spec-heavy content out of overly long SKILL.md/agent bodies into focused references/ files, then wires a loading table.
• Validate before integrating: Runs deterministic and rubric-driven checks to ensure references reach Level 3+ depth and are signal-loadable.

Quick Start

Run reference-enrichment for a target skill by invoking it with the skill or agent name (optionally adding --decompose to extract oversized sections into references).