# Calllint

> Calllint is an open-source MCP server for security. Static safety scan of MCP configs that returns SAFE, REVIEW, BLOCK or UNKNOWN verdicts. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI
- **Website:** https://calllint.com
- **Repository:** https://github.com/calllint/calllint
- **MCP Registry name:** io.github.calllint/calllint
- **Categories:** security, developer tools
- **Listing updated:** 2026-11-15

## About

Calllint reads an MCP or agent-tool configuration and works out what each server entry would let an agent read, write, execute, connect to, send or mutate, without ever running, installing or connecting to the server. Each finding cites the config field it came from, and results roll up into a risk class from S0 (metadata only) to S5 (financial or irreversible).

Thirteen static detectors cover secret-looking environment keys, broad filesystem paths, unknown remote hosts, prompt poisoning and hidden instructions in tool metadata, shell and package-runner commands, unverified local sources, message sending, OAuth scopes, gateway runtimes, payment actions and unpinned packages. It is offline and deterministic by default; network lookups only happen with `--online` and are advisory.

The `calllint-mcp` package lets an agent run the preflight itself before approving another server. A baseline and verify pair records an approved surface and flags later drift, and a CI mode gates changes to config files. The maintainers say verdicts are heuristic decision support, and SAFE means no blockers were observed, not a proof of runtime safety.

## Features

- **Evidence-backed verdicts:** Every finding points to the exact config field behind a SAFE, REVIEW, BLOCK or UNKNOWN result.
- **Thirteen static detectors:** Covers secrets, filesystem scope, network, prompt poisoning, shell execution, messaging, OAuth scopes, payments and unpinned packages.
- **Auto-discovery:** `npx calllint scan --auto` finds configs for Cursor, Claude Code, Claude Desktop, VS Code and Windsurf.
- **Drift detection:** `baseline` and `verify` flag approved servers whose risk surface later changes.
- **CI gate:** A `--ci` mode can block merges that change MCP config files.
- **Agent-run preflight:** The calllint-mcp server lets an agent scan configs before approving new servers.

## Good for

- Review an unfamiliar MCP server's config before installing it
- Gate pull requests that change .cursor/mcp.json in CI
- Detect rug-pulls after a package version changes
- Find unpinned packages and broad filesystem grants in a team's config
- Have an agent self-check before approving another tool

## Install

### Claude Code (Terminal)

```sh
claude mcp add calllint -- npx -y calllint-mcp
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "calllint": {
      "command": "npx",
      "args": [
        "-y",
        "calllint-mcp"
      ]
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "calllint": {
      "command": "npx",
      "args": [
        "-y",
        "calllint-mcp"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "calllint": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "calllint-mcp"
      ]
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "calllint": {
      "command": "npx",
      "args": [
        "-y",
        "calllint-mcp"
      ]
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "calllint": {
      "command": "npx",
      "args": [
        "-y",
        "calllint-mcp"
      ]
    }
  }
}
```

## FAQ

### Does it run the servers it scans?

No. It never executes, installs or connects to the target server.

### Does it need network access?

Not by default. It works offline and only goes online when you pass --online, for advisory data.

### Does SAFE mean the server is safe to run?

No. The site says SAFE means no blockers were observed, not a proof of runtime safety.

### How do I run it?

Use npx calllint scan --auto, or give a path such as npx calllint scan .cursor/mcp.json.

### How do I add it to an agent?

Add the calllint-mcp package, for example with npx -y calllint-mcp.

### Is Calllint free?

Yes. Calllint is open source, so you can use and self-host it for free.

### How do I add Calllint to Claude Code?

Run this command in your terminal: claude mcp add calllint -- npx -y calllint-mcp

### What are the best alternatives to Calllint?

The closest alternatives to Calllint 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/calllint
