raqueljezweb/anythingllm-mcp-server

AnythingLLM MCP Server

An MCP (Model Context Protocol) server that enables seamless integration between AnythingLLM and MCP-compatible clients like TypingMind, Claude Desktop, and more.

Features

• 🚀 Complete AnythingLLM API Integration: Access all AnythingLLM features through MCP
• 💬 Workspace Management: Create, update, delete, list workspaces and manage their settings
• 🤖 Chat Integration: Send messages, manage chat history, and stream responses
• 📄 Document Management: Upload, list, delete documents, process URLs, and manage vectors
• 🔍 Embedding & Search: Embed text/webpages and search within workspaces
• 👥 User Management: Create, update, delete users and manage permissions
• 🔑 API Key Management: Programmatically manage API keys
• ⚙️ System Administration: Configure LLM providers, vector databases, and system settings
• 🤖 Agent Support: Create and manage AI agents with custom prompts and tools
• 📊 System Monitoring: Access system information and statistics
• 🔐 Secure Authentication: API key-based authentication with environment variable support

Prerequisites

• Node.js 18+ installed
• AnythingLLM instance running with API access enabled
• AnythingLLM API key (obtain from your AnythingLLM instance at /settings/api-keys)

Installation

Option 1: Using npx (Recommended)

No installation needed! You can run the server directly with npx when configuring your MCP client.

Option 2: Global Installation

npm install -g anythingllm-mcp-server

Option 3: Local Installation

git clone https://github.com/yourusername/anythingllm-mcp-server.git
cd anythingllm-mcp-server
npm install

Configuration

For TypingMind.com

Setup Instructions

1. Go to TypingMind.com

2. Click on Settings → MCP Servers

3. In the MCP configuration editor, add the AnythingLLM server:

   {
    "mcpServers": {
      "anythingllm": {
        "command": "npx",
        "args": ["-y", "anythingllm-mcp-server@2.0.0"]
       }
     }
   }
   

4. Click Save - TypingMind will restart the MCP servers

5. Look for the "Connected" status

First Time Usage

After the server connects, initialize it with your AnythingLLM credentials:

1. In the chat, type:

   Use the initialize_anythingllm tool with apiKey: "your-api-key" and baseUrl: "http://localhost:3001"
   

2. Replace:

   • "your-api-key" with your AnythingLLM API key (get it from AnythingLLM → Settings → API Keys)
   • "http://localhost:3001" with your AnythingLLM instance URL

3. You'll see: "AnythingLLM client initialized successfully"

Example Commands

After initialization, use natural language to interact with AnythingLLM:

• "List all my workspaces"
• "Create a new workspace called 'Project Documentation'"
• "Chat with the 'research' workspace about the latest findings"
• "Show me all documents in the 'knowledge-base' workspace"

TypingMind will automatically use the appropriate AnythingLLM tools.

For Claude Desktop

Add to your Claude Desktop configuration file:

On macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
On Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "anythingllm": {
      "command": "npx",
      "args": ["-y", "anythingllm-mcp-server@1.0.3"]
    }
  }
}

Initialization

When first using the server in your MCP client, you'll need to initialize it with your AnythingLLM credentials by calling the initialize_anythingllm tool:

Use the initialize_anythingllm tool with apiKey: "your-api-key" and baseUrl: "http://localhost:3001"

• apiKey: Get this from your AnythingLLM instance at /settings/api-keys
• baseUrl: Your AnythingLLM instance URL (default: http://localhost:3001)

You can also set these as environment variables:

• ANYTHINGLLM_API_KEY: Your API key
• ANYTHINGLLM_BASE_URL: Your instance URL

Available Tools

Initialization

Workspace Management

Chat Operations

Document Management

Embedding & Search

User Management

API Key Management

System Administration

LLM Provider Management

Vector Database Management

Agent Management

Usage Examples

Once configured, you can interact with AnythingLLM through your MCP client:

List all workspaces

"Use the list_workspaces tool to show me all available workspaces"

Create a new workspace

"Create a new workspace called 'Project Documentation' using the create_workspace tool"

Chat with a workspace

"Use chat_with_workspace to ask the 'research' workspace: What are the key findings from the latest report?"

Manage documents

"List all documents in the 'knowledge-base' workspace"
"Delete document with ID 'doc123' from the 'archive' workspace"

API Reference

This MCP server connects to the following AnythingLLM API endpoints:

• /api/v1/workspaces - Workspace operations
• /api/v1/workspace/{slug}/chat - Chat functionality
• /api/v1/workspace/{slug}/documents - Document management
• /api/v1/system/settings - System configuration

For complete API documentation, visit /api/docs on your AnythingLLM instance.

Troubleshooting

Connection Issues

• ✅ Ensure your AnythingLLM instance is running and accessible
• ✅ Verify the API key has proper permissions
• ✅ Check that the base URL is correct (include http/https)
• ✅ Make sure the AnythingLLM API is enabled in your instance

API Errors

• Most errors will include descriptive messages
• Check AnythingLLM logs for detailed error information
• Ensure your API key hasn't expired
• Verify workspace slugs are correct when accessing specific workspaces

Common Issues

"Client not initialized" error:

• Make sure you've initialized the server with your API key and base URL

"Workspace not found" error:

• Check that the workspace slug is correct
• Use list_workspaces to see available workspaces

Connection refused:

• Verify AnythingLLM is running on the specified port
• Check firewall settings

Security Notes

• 🔐 Keep your API key secure and never commit it to version control
• 🔒 Use environment variables for sensitive configuration in production
• 🛡️ Restrict API key permissions to only what's needed
• ⚠️ Never share your API key publicly

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the file for details.

Support

• 🐛 Issues: GitHub Issues
• 📖 AnythingLLM Docs: docs.useanything.com
• 💬 AnythingLLM Community: Discord

Acknowledgments

• AnythingLLM by Mintplex Labs
• Model Context Protocol by Anthropic
• TypingMind for MCP support

Made with ❤️ for the AnythingLLM community