Prerequisites
- LibreChat self-hosted — librechat.ai
- Node.js 18+ on the LibreChat server
- Agent MCP Studio at agentmcp.studio
Step-by-Step Setup
Start bridge.js on the LibreChat server
curl -O https://agentmcp.studio/bridge.js && npm install ws # Keep running (use pm2 or systemd for persistence): node bridge.js wss://agentmcp.studio/api/relay/YOUR-UUID
Open the studio → Settings → MCP Relay Bridge → click Connect.
-
Edit librechat.yaml
In your LibreChat installation, edit
librechat.yamland add the MCP server configuration:mcpServers: agent-mcp-studio: command: node args: - /path/to/bridge.js - wss://agentmcp.studio/api/relay/YOUR-UUID type: stdio Restart LibreChat and enable tools
Restart the LibreChat server. In the chat interface, select a model that supports tool use and enable Plugins or Tools mode. Your browser tools will appear in the available tools list.
Frequently Asked Questions
Yes — LibreChat supports MCP via its librechat.yaml configuration. Add your MCP server under the mcp section and LibreChat's AI agents can call your Agent MCP Studio tools during conversations.
LibreChat is an open-source, self-hosted ChatGPT alternative that supports multiple AI providers (OpenAI, Anthropic, Google, local Ollama models). It offers a feature-rich UI with plugins, RAG, agents, and MCP tool support.
MCP servers are configured in your librechat.yaml file in the LibreChat root directory. Add a mcp key with server entries specifying the command and arguments for bridge.js.
MCP tool calls depend on the AI model's function calling support. In LibreChat, MCP tools work with Claude (full support), GPT-4o, and Gemini. Models without function calling won't invoke MCP tools.
No — LibreChat reads librechat.yaml at startup. Just restart the LibreChat server process after editing the YAML and your MCP tools will be available without rebuilding.