agent-docs-upgrader

Audits and rewrites AGENTS.md and CLAUDE.md files as concise agent operating manuals.

Updated Apr 6, 2026
One-click install
npx skills add https://github.com/QT-7274/dotfiles --skill agent-docs-upgrader-qt-7274
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-docs-upgrader
Source: https://github.com/QT-7274/dotfiles/tree/main/agent-docs-upgrader
Command: npx skills add https://github.com/QT-7274/dotfiles --skill agent-docs-upgrader-qt-7274

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agent documentation files like AGENTS.md and CLAUDE.md often become stale, verbose, contradictory, or force AI agents to read huge docs for small edits, degrading model behavior instead of improving it. ## Core Features & Use Cases - Audit and Scoring: Reviews existing agent docs against a rubric covering routing, length, specificity, progressive disclosure, freshness, and drift between mirrored files. - Rewrite and Creation: Produces concise repo operating manuals with routing tables, decision tables, short real examples, and paired do/don't rules. - Multi-Repo Governance: Enforces a decision gate for multi-repo workspaces and keeps AGENTS.md and CLAUDE.md synchronized to prevent drift. - Use Case: A team with a sprawling monorepo asks to modernize their CLAUDE.md; the skill audits it, cuts stale paths and generic advice, and rewrites it as a 100-line routing document with trigger-based spec references. ## Quick Start Ask the AI to audit and rewrite the AGENTS.md and CLAUDE.md files in this repository as concise agent operating manuals.

Frequently Asked Questions about agent-docs-upgrader

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

FAQPage Schema
How do I improve my CLAUDE.md or AGENTS.md file?▼

Run an audit against criteria like routing speed, length, specificity, and freshness, then rewrite the file as a concise operating manual. Keep entry docs around 60-150 lines with decision tables, real commands, and short examples instead of long prose.

What should a good AGENTS.md file contain?▼

A good AGENTS.md contains a project map, common workflow tables, real build/test/lint commands, local rules with reasons, and trigger-based references to large specs. It should route the agent to the right module quickly rather than forcing it to read every document.

Should AGENTS.md and CLAUDE.md be identical?▼

Yes, by default keep AGENTS.md and CLAUDE.md identical to reduce drift and prevent different agents from reading different project rules. Only isolate small platform-specific sections when explicitly requested, keeping shared rules synchronized.

How do I handle agent docs in a multi-repo workspace?▼

Use a root AGENTS.md as a routing document with a workspace map pointing to each sub-repo's entry doc. Before creating or omitting sub-repo docs, choose a strategy: root-only routing, docs for specified sub-repos, or docs for all detected sub-repos.

Why is a long agent instruction file worse than none?▼

Long, stale, or contradictory docs force agents to over-read, cause wrong edits, and slow routing. A bad doc is worse than no doc when it contains old paths, obsolete commands, or unconditional instructions to read all documentation first.