Skip to main content
Glama

Peekaboo MCP

by steipete
install-mcp-claude-desktop.mdβ€’5.82 kB
# Installing Peekaboo MCP in Claude Desktop This guide explains how to install the Peekaboo MCP server in Claude Desktop application. ## Prerequisites 1. **macOS 14.0 (Sonoma) or later** 2. **Claude Desktop** installed from [claude.ai/download](https://claude.ai/download) 3. **Peekaboo CLI** built and working ## Quick Installation ### Method 1: Direct Binary Installation (Recommended) 1. **Build Peekaboo** (if not already built): ```bash cd /Users/steipete/Projects/Peekaboo npm run build:swift ``` 2. **Edit Claude Desktop configuration**: ```bash # Open Claude Desktop config code ~/Library/Application\ Support/Claude/claude_desktop_config.json ``` 3. **Add Peekaboo MCP server** to the configuration: ```json { "mcpServers": { "peekaboo": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve"], "env": { "PEEKABOO_AI_PROVIDERS": "anthropic/claude-opus-4-20250514,ollama/llava:latest" } } } } ``` 4. **Restart Claude Desktop** to apply changes. ### Method 2: Using NPM Package (When Published) Once the npm package is published: ```bash # Install globally npm install -g @steipete/peekaboo-mcp # Configure Claude Desktop ``` Then add to `claude_desktop_config.json`: ```json { "mcpServers": { "peekaboo": { "command": "peekaboo-mcp" } } } ``` ## Configuration Options ### Environment Variables You can add environment variables to configure Peekaboo: ```json { "mcpServers": { "peekaboo": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve"], "env": { "PEEKABOO_AI_PROVIDERS": "anthropic/claude-opus-4-20250514", "PEEKABOO_LOG_LEVEL": "info", "OPENAI_API_KEY": "sk-...", "ANTHROPIC_API_KEY": "sk-ant-...", "PEEKABOO_USE_MODERN_CAPTURE": "true" } } } } ``` ### Available Environment Variables - `PEEKABOO_AI_PROVIDERS`: Comma-separated list of AI providers (e.g., `anthropic/claude-opus-4,openai/gpt-4.1`) - `PEEKABOO_LOG_LEVEL`: Logging verbosity (debug, info, warn, error) - `OPENAI_API_KEY`: OpenAI API key for GPT models - `ANTHROPIC_API_KEY`: Anthropic API key for Claude models - `X_AI_API_KEY` or `XAI_API_KEY`: xAI API key for Grok models - `PEEKABOO_OLLAMA_BASE_URL`: Ollama server URL (default: http://localhost:11434) - `PEEKABOO_USE_MODERN_CAPTURE`: Use modern capture API (true/false) ## Verifying Installation 1. **Check Claude Desktop logs**: ```bash # Monitor Claude Desktop logs tail -f ~/Library/Logs/Claude/mcp*.log ``` 2. **Check Peekaboo MCP logs**: ```bash # Monitor Peekaboo logs ./scripts/pblog.sh -f ``` 3. **Test in Claude Desktop**: - Open Claude Desktop - Start a new conversation - Type: "Can you take a screenshot of my desktop?" - Claude should use the Peekaboo MCP tools ## Troubleshooting ### Common Issues 1. **"Tool not found" errors** - Ensure the binary path is absolute, not relative - Check that the binary is executable: `chmod +x /path/to/peekaboo` 2. **Permission errors** - Grant Screen Recording permission: System Settings β†’ Privacy & Security β†’ Screen Recording - Grant Accessibility permission: System Settings β†’ Privacy & Security β†’ Accessibility 3. **MCP server not starting** - Check Claude Desktop logs for errors - Try running the command manually: `/path/to/peekaboo mcp serve` - Verify the binary exists and is the correct architecture 4. **API key issues** - Ensure API keys are properly set in the env section - Or configure them globally in `~/.peekaboo/credentials` ### Debug Mode Enable debug logging to troubleshoot issues: ```json { "mcpServers": { "peekaboo": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve", "--log-level", "debug"], "env": { "PEEKABOO_LOG_LEVEL": "debug" } } } } ``` ## Advanced Configuration ### Multiple Instances You can run multiple Peekaboo instances with different configurations: ```json { "mcpServers": { "peekaboo-local": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve"], "env": { "PEEKABOO_AI_PROVIDERS": "ollama/llama3.3" } }, "peekaboo-cloud": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve"], "env": { "PEEKABOO_AI_PROVIDERS": "anthropic/claude-opus-4,openai/gpt-4.1" } } } } ``` ### Custom Working Directory Set a custom working directory for file operations: ```json { "mcpServers": { "peekaboo": { "command": "/Users/steipete/Projects/Peekaboo/peekaboo", "args": ["mcp", "serve"], "cwd": "/Users/steipete/Desktop" } } } ``` ## Available Tools Once installed, Claude will have access to all Peekaboo tools: - **Screen Capture**: `image`, `see` - **UI Automation**: `click`, `type`, `scroll`, `hotkey`, `swipe`, `drag` - **Window Management**: `window`, `list`, `space` - **App Control**: `app`, `menu`, `dock` - **System**: `permissions`, `sleep`, `dialog` - **AI Analysis**: `analyze`, `agent` - **Utilities**: `move`, `clean` ## Security Considerations 1. **API Keys**: Store sensitive API keys in `~/.peekaboo/credentials` (chmod 600) 2. **Permissions**: Only grant necessary system permissions 3. **File Access**: Be aware that MCP tools can read/write files 4. **Screen Content**: Screenshots may contain sensitive information ## Support - **Issues**: [GitHub Issues](https://github.com/steipete/peekaboo/issues) - **Documentation**: [Peekaboo Docs](https://github.com/steipete/peekaboo/tree/main/docs) - **MCP Spec**: [Model Context Protocol](https://modelcontextprotocol.io)

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/steipete/Peekaboo'

If you have feedback or need assistance with the MCP directory API, please join our Discord server