AgentLint CLI
A linter for the agent harness checks a repository's AGENTS.md, CLAUDE.md, CI and hooks and scores six dimensions.
- 65 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- `agentlint check` reads repository configuration, and extended checks launch sub-agents and read session logs.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/0xmariowu-agentlintDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- `agentlint check` reads repository configuration, and extended checks launch sub-agents and read session logs.
This is third-party code. Review the repository files before installing.
What it does
AgentLint looks at what the agent reads rather than at code: `AGENTS.md`, `CLAUDE.md`, CI configuration, hooks and `.gitignore`. It runs 51 deterministic checks across six dimensions: findability, instructions, workability, safety, continuity and harness. The result is a score out of 100 with a per-dimension breakdown. For a deeper review there are seven opt-in extended checks that launch AI sub-agents and read local Claude Code session logs when present. Run it with `agentlint check` in any git repository. After `npx agentlint-ai install`, Claude Code gets the `/al` command with a scan, fix and report loop.
Who it is for. Developers and teams who want to check instruction files and the repository harness the way they check code.
Good fit when
- Before rolling agents into a repository, how clear `AGENTS.md` and `CLAUDE.md` are gets assessed.
- A check is added to CI so the harness does not degrade.
- In Claude Code `/al` returns fixes for weak spots after a scan.
- Scores of several organisation repositories are compared.
Not a fit when
- You need style and bug checks for the code itself, as ESLint does.
- The repository has no instruction files for an agent.
- Sub-agents and reading session logs, needed for extended checks, are not allowed.
Example request
Check this repository's harness with AgentLint and show the three weakest dimensions with fixes.Limitations
The base 51 checks are deterministic, but the score depends on the chosen rule set, so it reflects the authors' conventions rather than universal truth. The first command, `npm install -g agentlint-ai`, installs only the CLI and does not touch `~/.claude/`. The second, `npx agentlint-ai install`, adds the `/al` command to `~/.claude/commands/` and registers the marketplace plugin, with details in `INSTALL.md`. Extended checks use AI and local session logs.
How to disable. Remove the package with `npm uninstall -g agentlint-ai`. If you installed the `/al` command, delete it from `~/.claude/commands/` and remove the plugin with a Claude Code command; the steps are in `INSTALL.md`.
FAQ
How do I install AgentLint?
Run `npm install -g agentlint-ai`. A separate command, `npx agentlint-ai install`, adds the Claude Code plugin.
How do I run a check?
In a git repository run `agentlint check`. In Claude Code after installing the plugin call `/al`.
Which dimensions are scored?
Findability, instructions, workability, safety, continuity and harness.
Does installation change Claude Code settings?
`npm install -g agentlint-ai` does not touch `~/.claude/`. Only the opt-in `npx agentlint-ai install` does.
What are extended checks?
Seven opt-in Deep and Session checks. They launch sub-agents and read local Claude Code session logs.
Discussion
No questions or reviews yet.
Related
CLI toolsDeterministic AGENTS.md and CLAUDE.md checks: compares a change with its baseline and blocks regressions in CI with no model.
A CLI that analyses a repository in six model phases and writes AGENTS.md or CLAUDE.md, with ExecPlan planning on top.
agnix lints CLAUDE.md, AGENTS.md, SKILL.md, hooks and MCP configs with 702 rules, autofix and editor extensions for VS Code and JetBrains
A skill that analyzes a repository and generates files for AI agents: AGENTS.md, pointers, reviewers, CI and templates
Agentic OS adds a five-step workflow and checks in git hooks and CI so that a step without evidence does not count as done
A Claude Code and Codex plugin with skills for code research, tests, refactoring and independent review of finished changes
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| AgentLint (this card) | no | 65 | |
| AgentLint | no | 156 | a hook-based guard that blocks risky commands and secret writes by an agent; AgentLint by 0xmariowu scores instruction files and the repository harness |
| Dev Machine Guard | no | 197 | scans a developer machine for agents and risky settings; AgentLint scores instruction files and the harness inside a repository |
README in short
The core idea is that an agent is a model plus a harness, and the harness can be linted like code. Docs and guides live on the project site.