Agent Lint
- 47 repo stars
- Updated July 24, 2026
- 0xmariowu/AgentLint
AgentLint is a harness linter delivered as a Claude Code plugin that audits agent instruction files so developers can improve the context AI tools read before coding.
About
agent-lint is a Claude Code marketplace plugin from AgentLint that scores how well a repository supports AI coding agents. It runs deterministic checks on CLAUDE.md, AGENTS.md, Cursor rules, Copilot instructions, and related harness files, then offers fixes, CI bootstrap, and an interactive /al workflow. Install it when agent output feels inconsistent and you suspect missing or low-quality repo guidance rather than model limits.
- 51 deterministic core checks across six harness dimensions
- Scans CLAUDE.md, AGENTS.md, Cursor rules, and Copilot instructions
- CLI agentlint check plus interactive /al scan-fix-report in Claude Code
- Optional bootstrap for CI, hooks, and agent templates
Agent Lint by the numbers
- Data as of Jul 28, 2026 (Skillselion catalog sync)
/plugin marketplace add 0xmariowu/AgentLintAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| repo stars | ★ 47 |
|---|---|
| Last updated | July 24, 2026 |
| Repository | 0xmariowu/AgentLint ↗ |
Why do AI agents ignore my repo conventions—and how do I systematically audit CLAUDE.md, AGENTS.md, and related harness files?
Score and fix repo harness files like CLAUDE.md and AGENTS.md so AI coding agents get reliable context.
Who is it for?
Teams using Claude Code, Codex, Cursor, or Copilot who want evidence-backed scores and fixes for agent-facing repository docs.
Skip if: Repositories with no AI agent workflow or teams seeking application runtime bug finding instead of harness quality.
What you get
Returns a harness score report, suggested fixes, and optional CI hooks or templates to raise agent readiness.
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
Recommended Marketplaces
FAQ
Does AgentLint send my code to a cloud model by default?
The core 51 checks are deterministic and local; extended Deep and Session modes are opt-in and may use sub-agents or local logs.
Which files does it inspect?
CLAUDE.md, AGENTS.md, .cursor/rules, .github/copilot-instructions.md, and related harness assets if present.
How do I run it inside Claude Code?
Install the marketplace plugin, run npx agentlint-ai install once, then use the /al slash command for scan-fix-report.