lucid

Lucid MCP Server enables multimodal LLMs to access and analyze Lucid diagrams exported from Lucid Chart, Lucid Spark, and Lucid Scale. It exposes tools that let you discover documents, retrieve metadata, and optionally analyze visual content using AI providers. You can interact with the server using natural language prompts or call its built-in tools directly. The available tools include search-documents for listing Lucid documents, get-document for metadata with optional AI analysis of visuals, and get-document-tabs for lightweight page metadata. When AI analysis is requested, the server will use your configured AI provider keys (Azure OpenAI or OpenAI) to analyze diagram content and provide insights. This setup is particularly useful for integrating Lucid diagrams into conversational workflows or automated diagram analysis pipelines.

Prerequisites:\n- Node.js (v18 or higher) installed on your system.\n- npm or pnpm available.\n- A Lucid API key required for all features.\n- Optional: API keys for AI providers (Azure OpenAI or OpenAI) if you want AI-powered analysis.\n\nInstall steps:\n1) Install the lucid-mcp-server package globally:\nbash\nnpm install -g lucid-mcp-server\n\n2) Optionally verify installation and run via npx (as recommended by the README):\nbash\nnpx -y lucid-mcp-server\n\n3) Set up environment variables in your shell:\nbash\nexport LUCID_API_KEY="your_api_key_here"\n# Optional: Azure OpenAI settings\nexport AZURE_OPENAI_API_KEY="your_azure_openai_key"\nexport AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"\nexport AZURE_OPENAI_DEPLOYMENT_NAME="gpt-4o"\n# Optional: OpenAI settings (fallback if Azure not configured)\nexport OPENAI_API_KEY="your_openai_api_key"\nexport OPENAI_MODEL="gpt-4o"\n\n4) Verify the MCP Inspector (optional) to test the server:\nbash\nnpx @modelcontextprotocol/inspector lucid-mcp-server\n

Tips and notes:\n- The server requires a Lucid API key (LUCID_API_KEY) for all features; without it, document discovery and metadata retrieval will be limited.\n- For AI-powered diagram analysis, either configure Azure OpenAI keys or provide OpenAI keys. Azure is preferred if configured.\n- When using VS Code integration, the settings.json example in the README shows how to wire environment variables for the MCP server.\n- If you encounter issues with authentication or API access, double-check that your Lucid API key is valid and that the API endpoints for OpenAI or Azure OpenAI are reachable from your environment.\n- The MCP tools support exporting and analyzing individual documents by ID, and you can combine search and analysis flows in automation tasks.