# audit MCP server

> Scan, enumerate, and risk-score MCP servers configured on your machine to understand their declared capabilities and access levels.

- Repository: https://github.com/saagpatel/MCPAudit
- Page: https://hostedmcp.org/servers/saagpatel/mcpaudit
- Transport: stdio
- Stars: 4
- License: MIT
- Latest version: 3.0.0

## Install

### Claude Code

```sh
claude mcp add mcpaudit -- uvx mcp-audits
```

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

```json
{
  "mcpServers": {
    "mcpaudit": {
      "command": "uvx",
      "args": [
        "mcp-audits"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "mcpaudit": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "mcp-audits"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"mcpaudit","type":"stdio","command":"uvx","args":["mcp-audits"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "mcpaudit": {
      "command": "uvx",
      "args": [
        "mcp-audits"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add mcpaudit -- uvx mcp-audits
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.mcpaudit]
command = "uvx"
args = ["mcp-audits"]
```

### Gemini CLI

```sh
gemini mcp add -s user mcpaudit uvx -- mcp-audits
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "mcpaudit": {
      "command": "uvx",
      "args": [
        "mcp-audits"
      ]
    }
  }
}
```

## FAQ

### Does a high risk score mean a server is malicious?

No, a high score indicates broader capability exposure rather than confirming a server is malicious or unsafe.

### Does mcp-audit edit my configuration files?

No, the tool does not perform any edits to your configuration files.

### Can I use mcp-audit in my CI/CD pipeline?

Yes, it can write SARIF reports and evaluate local policy files for use in automated workflows.

### Does the tool report my environment variable values?

No, it only reports the names of the environment variable keys to provide context without exposing secrets.
