Audit and trim AI agent instruction files to remove redundant documentation.
npx skills add https://github.com/borkweb/skills --skill agents-md-lint
Please help me install this Agent Skill. Skill: agents-md-lint Source: https://github.com/borkweb/skills/tree/main/skills/core/agents-md-lint Command: npx skills add https://github.com/borkweb/skills --skill agents-md-lint
This Skill automates the auditing and pruning of AI agent instruction files to remove redundant information, thereby conserving valuable context tokens for more efficient AI operations.
Run the 'agents-md-lint' skill to remove redundant documentation from your instruction files.
High-intent search queries and answers about installing and using this skill.
You can eliminate redundant context tokens by auditing and trimming AI agent instruction files to remove documentation that can be derived from the codebase. This optimization frees up context tokens for more efficient AI performance.
AI agent instruction files like AGENTS.md, CLAUDE.md, CONVENTIONS.md, and .cursorrules provide guidelines for AI coding assistants. They require pruning when redundant documentation consumes context tokens and limits the AI's available working context.
Yes, the auditing and trimming process supports repositories containing CLAUDE.md and .cursorrules files, as well as AGENTS.md and CONVENTIONS.md. It analyzes these files to identify and remove documentation that can be derived from code.
You audit AGENTS.md files by running an automated process that utilizes a sub-agent to simulate blind auditing from code-only context. This identifies unnecessary facts that can be derived from the codebase for removal.
Yes, codebase access is required to analyze and trim CONVENTIONS.md files. The auditing process must compare the instruction file documentation against the code to determine which facts are redundant and can be removed.
flightctlScan codebase structure and convention files to generate or refresh AGENTS.md.
deltatreeCompresses source documents into a lossless, token-efficient format for LLM consumption.
EndUser123Audit Claude Code configuration artifacts for over-engineering, redundancy, and token waste.
stephanjRefactor monolithic CLAUDE.md or AGENTS.md files into modular linked documentation.
