metorial-platform

Metorial Platform provides a ready-to-run MCP server that powers agentic AI integrations. It acts as a bridge between AI models and external APIs, data sources, and tools using the Model Context Protocol. With Metorial, you can connect your AI applications to thousands of APIs via a single function call, leverage the embedded MCP Explorer in the dashboard for testing, and manage multiple MCP connections across environments. The platform emphasizes developer-focused tooling, including first-class SDKs for JavaScript/TypeScript and Python, a powerful API, and a self-hosted option if you prefer to run everything locally or in your own infrastructure. Once running, you can register MCP servers, configure contexts and tools, and monitor usage and debugging data through the Metorial Dashboard.

The MCP integration capabilities enabled by Metorial include creating reusable contexts for external data sources, orchestrating calls to multiple APIs within a single function, and wiring up tools (like web services, databases, or data processing utilities) to a unified interface. The included SDKs simplify building clients that call into the MCP server, while the dashboard and monitoring features help you observe session activity, errors, and performance in real time. This makes it suitable for building agentic AI applications that interact reliably with external systems at scale.

Prerequisites:

• Docker installed on your machine or server
• Internet access to pull the Metorial Platform image
• Optional: a hosting environment if deploying to production (cloud provider, Kubernetes, etc.)

Install steps (Docker-based):

1. Verify Docker is installed

   • macOS: open Docker Desktop
   • Windows: open Docker Desktop
   • Linux: ensure the docker daemon is running

2. Pull and run the Metorial Platform image

   • Pull the latest image: docker pull metorial/metorial-platform:latest
   • Run the container (adjust ports as needed, e.g., 8080 for dashboard, 4000+ for MCP endpoints): docker run -i -p 8080:8080 -p 4001:4001 metorial/metorial-platform:latest

3. Configure environment (optional but recommended)

   • If the image requires environment variables (database URLs, secrets, etc.), pass them with -e flags. Example: docker run -i
     -p 8080:8080
     -p 4001:4001
     -e METORIAL_DB_URL="postgres://user:pass@dbhost:5432/metorial"
     -e METORIAL_API_KEY="your-api-key"
     metorial/metorial-platform:latest

4. Access and initial setup

   • Dashboard: http://localhost:8080
   • Follow the on-screen prompts or consult the API docs to create MCP server configurations and connect tools/APIs.

5. Optional: Run in the background with Docker compose (example)

   • Create a docker-compose.yml with the appropriate image and ports, then run: docker compose up -d

6. Verify the service

   • Check logs: docker logs <container-id>
   • Ensure MCP endpoints are reachable and the dashboard loads.

• Environment variables: If the image requires database connections, cache stores, or API keys, provide them via -e flags (e.g., METORIAL_DB_URL, METORIAL_API_KEY). Check the docs for the exact variable names.
• Ports: Ensure the exposed ports (dashboard and MCP endpoints) are accessible from your network and not blocked by firewalls.
• Data persistence: If running with databases or storage, mount volumes to persist data across restarts.
• Health checks: Use the dashboard or API endpoints to verify MCP server health and session activity.
• Security: When deploying publicly, restrict access to the dashboard and MCP endpoints, and rotate credentials regularly.
• Troubleshooting: If the container fails to start, inspect logs for missing env vars, misconfigurations, or resource constraints (CPU/memory).