seojoonkim

seojoonkim / agentlinter

Public

ESLint for AI Agents — AGENTS.md/CLAUDE.md 채점·진단·자동수정 | Position Risk Warning · Token Efficiency · Security Check

agentlinter.vercel.app agentlinter agents-md ai-agent claude claude-md
38
7
100% credibility
Found Feb 07, 2026 at 17 stars 2x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

AgentLinter scans AI agent instruction files to score their quality across structure, clarity, completeness, security, consistency, memory handling, runtime setup, and skill safety, offering diagnostics and fixes.

How It Works

1
📰 Discover AgentLinter

You stumble upon a friendly checker that helps make AI helpers smarter and safer while browsing tips for building assistants.

2
📁 Open your helper's folder

Head to the folder where your AI helper keeps its thinking instructions.

3
🔍 Run the quick check

Tap the simple starter to scan everything and reveal your helper's strengths and soft spots in seconds.

4
📊 See your score and tips

Get a clear report with a grade for smarts, safety, and flow, plus friendly suggestions for each tweak.

5
✏️ Make easy improvements

Follow the simple hints to sharpen instructions, fix fuzzy spots, and boost safety.

6
🔄 Check again

Run the checker once more to watch your score soar and confirm everything's better.

🚀 Your helper shines!

Celebrate as your AI assistant now thinks clearly, stays secure, and performs at its best.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 38 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is agentlinter?

AgentLinter is ESLint for AI agents: it scans your CLAUDE.md and workspace files, scores them 0-100 across eight dimensions like structure, clarity, security, and skill safety, then diagnoses issues and auto-fixes safe ones with `npx agentlinter --fix`. Built in TypeScript, it detects leaked secrets, vague instructions, cross-file inconsistencies, and runtime config risks in setups like Claude Code, OpenClaw, Cursor, or Windsurf. Run it locally with one command—no API keys or signup—for instant reports and shareable links.

Why is it gaining traction?

It auto-detects your agent framework and applies tailored rules, outputting GitHub Action-ready formats with annotations for CI workflows, much like eslint github action or github eslint config prettier. Developers love the `--fix` flag for quick wins on CLAUDE.md files, plus skill safety scans that flag curl-pipe-bash risks before install. Free forever, with templates via `npx agentlinter init --template coding`, it fills the gap left by Anthropic's basic tools.

Who should use this?

AI agent builders tweaking CLAUDE.md for Claude Code projects, or teams maintaining OpenClaw/Moltbot workspaces with multiple files like SOUL.md and TOOLS.md. Frontend devs using Cursor/Windsurf for agentic coding agents, or ops folks integrating eslint github ci to lint agent configs in PRs. Anyone diagnosing why their agent hallucinates or leaks secrets.

Verdict

Try it for early agent linting—solid TypeScript CLI with eslint-like workflow, but at 21 stars and 1.0% credibility, it's pre-1.0 with basic tests. Great docs and GitHub Action make it usable now; watch eslint github releases for maturity.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.