Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/garagon/aguara/llms.txt

Use this file to discover all available pages before exploring further.

Aguara can automatically discover MCP server configurations across 17 popular AI coding assistants and agent frameworks, making it easy to scan all your MCP servers without manually specifying config paths.

​
Supported Clients

Aguara auto-detects MCP configurations from the following clients:
ClientConfig Locations
Claude Desktop~/Library/Application Support/Claude/ (macOS)
~/.config/claude/ (Linux)
%APPDATA%\Claude\ (Windows)
Cursor~/.cursor/mcp.json
VS Code~/.vscode/mcp.json
Cline~/.cline/mcp_settings.json
Windsurf~/.windsurf/mcp.json
~/.codeium/windsurf/mcp_config.json
OpenClaw~/.openclaw/openclaw.json
OpenCode~/.config/opencode/opencode.json
Zed~/.config/zed/settings.json
Amp~/.config/amp/settings.json
Gemini CLI~/.gemini/settings.json
Copilot CLI~/.copilot/mcp-config.json
Amazon Q~/.aws/amazonq/mcp.json
Claude Code~/.claude.json
Roo CodeVS Code global storage
Kilo CodeVS Code global storage
BoltAI~/.boltai/mcp.json (macOS)
JetBrains~/.junie/mcp/mcp.json

​
Discovery Commands

​
List All MCP Configs

Discover which MCP clients are configured on your system without scanning: 
aguara discover
Example output: 
Found 2 MCP configuration(s):

  Claude Desktop  /Users/you/Library/Application Support/Claude/claude_desktop_config.json
    ├── filesystem                node /path/to/filesystem-server/index.js
    └── git                       npx -y @modelcontextprotocol/server-git

  Cursor  /Users/you/.cursor/mcp.json
    └── web-search                python -m mcp_server_duckduckgo

Total: 3 MCP servers across 2 clients

​
JSON Output

Get discovery results as JSON for programmatic use: 
aguara discover --format json

{
  "clients": [
    {
      "client": "claude-desktop",
      "path": "/Users/you/Library/Application Support/Claude/claude_desktop_config.json",
      "servers": [
        {
          "name": "filesystem",
          "command": "node",
          "args": ["/path/to/filesystem-server/index.js"],
          "env": {}
        }
      ]
    }
  ]
}

​
Auto-Discovery + Scan

Discover all MCP configurations and scan them in one command: 
aguara scan --auto
This is equivalent to:
  1. Running aguara discover to find all MCP configs
  2. Extracting the command, args, and env from each server
  3. Scanning all referenced files and scripts

​
How It Works

​
Config Parsing

Aguara reads MCP client configuration files and extracts server definitions. Most clients use a standard format: 
{
  "mcpServers": {
    "server-name": {
      "command": "node",
      "args": ["server.js"],
      "env": {
        "API_KEY": "sk-..."
      }
    }
  }
}
Client-Specific Variations:
  • VS Code: Uses servers instead of mcpServers
  • Zed: Uses context_servers
  • OpenCode: Nested under mcp.servers
  • Claude Code: Nested under mcp.mcpServers
  • OpenClaw: Custom TOML-like format in openclaw.json
Aguara handles all these variations automatically.

​
Path Resolution

For each discovered server, Aguara:
  1. Resolves the command (e.g., node, python, npx)
  2. Locates the script file from args (e.g., server.js, -m module_name)
  3. Scans the script and any related files for security issues
  4. Checks environment variables for hardcoded credentials

​
Security Analysis

When you run aguara scan --auto, all discovered servers are analyzed for:
  • Hardcoded API keys and secrets in config env blocks
  • Unpinned npx packages (e.g., npx -y package without version)
  • Dangerous command flags (--allow-all, shell=True)
  • Docker security issues (--cap-add, --privileged, --network=host)
  • Server script vulnerabilities (prompt injection, command execution, exfiltration)

​
Use Cases

​
Pre-Deployment Audit

Before deploying an MCP server to production, scan all configured servers: 
aguara scan --auto --severity high --verbose

​
CI Integration

Add MCP config scanning to your GitHub Actions workflow: 
- name: Scan MCP configurations
  run: |
    aguara scan --auto --ci

​
Development Workflow

After installing a new MCP server, check for security issues: 
# Install server
claude mcp add my-server -- node server.js

# Scan it
aguara scan --auto

​
Multi-Client Monitoring

If you use multiple AI assistants (Claude Desktop + Cursor + VS Code), discover and scan all of them at once: 
aguara scan --auto --format json -o scan-results.json

​
Limitations

  • File System Access: Discovery requires read access to client config directories
  • Cross-Platform: Some clients (like BoltAI) are macOS-only
  • Custom Locations: Non-standard config paths won’t be auto-detected (use aguara scan <path> instead)
  • Server Code: Aguara scans local server scripts, but can’t analyze remote npx packages until they’re installed

Build docs developers (and LLMs) love

Get started for freeTalk to us