pluggedin-app

plugged.in is an all-in-one AI-CMS and MCP-enabled platform that emphasizes persistence, versioning, and interoperability. It embeds a robust MCP hub within a single web application, enabling you to connect 1,500+ MCP-compatible tools, filter and manage integrations, and use an in-process RAG engine for document indexing and semantic search. With this MCP server, you can discover, configure, and monitor external MCP servers, test model interactions, and run multiple tools in a unified interface. The system also supports live debugging, per-server configuration, and OAuth-based authentication to securely orchestrate tools across your workspace.

Prerequisites:

• Docker and Docker Compose installed on your system
• Sufficient disk space for a Postgres database, vector storage, and application assets

Step-by-step:

1. Install Docker
   • Follow instructions at https://docs.docker.com/get-docker/
2. Pull and run the plugged.in image via Docker (example)

# Pull and run the latest Plugged.in image (MCP-enabled)
docker run -d --name pluggedin-app -p 12005:12005 veriteknik/pluggedin:latest

3. Initialize environment and data (optional)
   • If your image requires a data directory or environment configuration, mount volumes and supply environment variables as needed.

# Example: with docker-compose (recommended for multi-service setup)
# Create a docker-compose.yml that defines the services (web, postgres, etc.)
# Then:
docker compose up -d --build

4. Access the MCP-enabled interface

   • Open http://localhost:12005 in your browser (or the port you configured).

5. Verify MCP connectivity

   • Use the MCP hub UI to discover and configure external MCP servers, test tool interactions, and validate OAuth flows.

Tips and notes:

• This setup uses a multi-arch Docker image; ensure your host supports amd64 or arm64. The image includes PostgreSQL with pgvector and an embedded in-process RAG engine.
• If upgrading from older versions, prefer a fresh start or follow the provided migration guidance in the repository notes.
• For security, enable OAuth 2.1 where supported and consider configuring TLS/HTTPS in a reverse-proxy setup for production.
• Monitor MCP activity via the built-in auditing and analytics features to detect misconfigurations or tool conflicts.
• The MCP cache and vector data are stored in persistent volumes; back them up regularly if you value long-term data retention.