# skilllint — vet an agent skill before you install it

> skilllint — vet an agent skill before you install it is a freemium MCP server for security. Lint a SKILL.md for missing triggers, leaked secrets and injection markers before install. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Freemium
- **Starting price:** $19 a month
- **Platforms:** API, CLI, self-hosted
- **Website:** https://rta.lat/skilllint/pricing
- **Repository:** https://github.com/mrpkk/skilllint
- **MCP Registry name:** lat.rta/skilllint
- **Categories:** security, developer tools
- **Listing updated:** 2026-11-14

## About

Skilllint checks an agent skill file against eleven numbered rules and returns a PASS, WARN or FAIL verdict with a score. Findings range from critical, such as missing frontmatter, secret-shaped literals and prompt-injection or exfiltration markers, down to low-severity items like files over 60 KB. Other rules cover an invalid name, an overlong description, a description with no when-to-use trigger, TODO or placeholder text, references to missing files, world-writable scripts and instructions with unbounded autonomy.

The README contrasts it with scanners that send the inspected content to a provider: the check runs locally and only a verdict, score and hash leave the machine. The verdict is tied to the sha256 of the file so it can be signed and verified later, and a compare tool ranks several candidate skills from different registries.

The hosted endpoint gives 50 free calls a day, then $0.02 a call in USDC over x402 with no account, or $19 for 30 days with no daily cap. The code is MIT licensed, and interfaces are available in English and Hindi.

## Features

- **Eleven lint rules:** Checks structure, triggers, secrets, injection markers, links and autonomy.
- **Three-level verdict:** Returns PASS, WARN or FAIL with a numeric score.
- **Hash-bound result:** Ties each verdict to the file's sha256 so it can be signed later.
- **Local checking:** Runs on your machine; only verdict, score and hash go out.
- **Skill comparison:** Ranks several candidate skills from different registries.
- **Pay per call:** 50 free calls a day, then USDC over x402 or a monthly pass.

## Good for

- Vet a downloaded skill before adding it to an agent
- Rank competing skills for the same job
- Catch secrets accidentally pasted into a SKILL.md
- Gate a skill allowlist on a reproducible hash

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http skilllint https://rta.lat/skilllint/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://rta.lat/skilllint/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "skilllint": {
      "url": "https://rta.lat/skilllint/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "skilllint": {
      "type": "http",
      "url": "https://rta.lat/skilllint/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "skilllint": {
      "serverUrl": "https://rta.lat/skilllint/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "skilllint": {
      "type": "http",
      "url": "https://rta.lat/skilllint/mcp"
    }
  }
}
```

## Tools

- `lint_skill`: Vet a SKILL.md before you install it: structure, trigger quality, leaked secrets, prompt-injection markers, broken links and autonomy claims. Returns a verdict, a score and a sha256 of the checked file.
- `compare_skills`: Compare candidate skills from different registries and rank them by verdict and evidence quality. Takes a comma-separated list of paths.
- `lint_rules`: List the linter's rules with their identifiers and severity levels.

## FAQ

### Does my skill file leave my machine?

The README says the check runs locally and only a verdict, a score and a hash go out.

### What does the verdict mean?

PASS needs a score of 80 or more and no critical finding; any critical finding is FAIL; WARN sits between.

### What does it cost?

Free for 50 calls a day, $0.02 a call over x402, or $19 for 30 days with no cap.

### Which tools does it expose?

lint_skill, compare_skills and lint_rules.

### Is skilllint — vet an agent skill before you install it free?

skilllint — vet an agent skill before you install it has a free plan, with paid plans for more usage or features. Paid plans start at $19 a month.

### How do I add skilllint — vet an agent skill before you install it to Claude Code?

Run this command in your terminal: claude mcp add --transport http skilllint https://rta.lat/skilllint/mcp

### What are the best alternatives to skilllint — vet an agent skill before you install it?

The closest alternatives to skilllint — vet an agent skill before you install it are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Concord MCP](https://ai-powered.com/mcp/concord-mcp): Concord MCP is a freemium MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/skilllint-vet-an-agent-skill-before-you-install-it
