Metorial
Author: metorial
Description: Open-source integration platform powered by MCP (Model Context Protocol) to connect AI models to many external APIs/tools with a unified interface. Provides hosted and self-hosted options, SDKs (TypeScript/JavaScript + Python), OAuth session management for tool auth, and an ecosystem of MCP server deployments/containers.
Stars: 3.2k
Forks: 358
License: Apache License 2.0
Category: Enterprise
Overview
Installation
.run()
Run an agent conversation loop with automatic session management, using specified MCP server deployments and an LLM client/model.
metorial.oauth.sessions.create()
Create an OAuth session for a given server deployment and return a user-facing authorization URL and session ID.
metorial.oauth.waitForCompletion()
Wait for one or more OAuth sessions to complete so the authenticated sessions can be used in subsequent runs (JavaScript/TypeScript).
metorial.oauth.wait_for_completion()
Wait for one or more OAuth sessions to complete so the authenticated sessions can be used in subsequent runs (Python).
FAQs
How do I pin a specific Docker image SHA for Metorial mcp-containers to ensure reproducible builds?
Replace the image tag with its SHA digest in your Docker config using the format `ghcr.io/metorial/mcp-container--[author]--[project]--[server]@sha256:[full-hash]`. Obtain the SHA by running `docker inspect --format='{{.Id}}' [image-name]` after pulling, or by checking the GitHub Container Registry package page for the specific server you need.
Can I use Metorial mcp-containers with Cursor or VS Code instead of Claude Desktop, and what configuration changes are needed?
Yes, both Cursor and VS Code likely support Metorial containers through their documented MCP features. Cursor requires adapting the Claude Desktop Docker command into your MCP settings, though note resource feature limitations documented in their forum. VS Code needs the config placed in `.vscode/mcp.json` following their MCP API extension guide format. The core Docker flags `-i` and `--rm` remain essential regardless of client.
How do Metorial mcp-containers handle automatic updates, and what are the risks of running unversioned images in production?
Metorial images pull from upstream sources without version tags or change logs, meaning docker pull fetches the latest build automatically. Production risks include breaking changes appearing without notice, inability to rollback to known-good states, and compliance failures when audit trails require exact dependency versions. Always pin images by SHA digest and test updates in staging environments first.
What is the difference between using Metorial mcp-containers standalone versus the full Metorial hosted platform with OAuth and monitoring?
The standalone containers require you to manage secrets through environment variables and monitor server health yourself, while the hosted platform automates credential refresh, provides observability dashboards, and handles infrastructure scaling. Think of standalone as self-service Docker images versus a managed service layer that removes operational overhead for production deployments requiring uptime guarantees and compliance controls.
How do I securely pass API keys and secrets to Metorial MCP containers without exposing them in process arguments?
Use Docker's `--env-file` flag to load secrets from a file instead of inline environment variables. Create a file like `.env` containing `KEY=value` pairs, set restrictive file permissions with `chmod 600 .env`, and reference it in your config's args array as `["run", "-i", "--rm", "--env-file", "/absolute/path/.env", "ghcr.io/..."]`. This keeps credentials out of process listings and command history while maintaining Docker's stdio transport.
How does Metorial mcp-containers compare to Composio for teams that need managed OAuth authentication across multiple integrations?
Metorial requires manual environment variable handling for credentials in each container, while Composio centralizes OAuth token management server-side. Teams needing pre-built authentication flows across hundreds of SaaS providers should choose Composio's managed approach. Metorial suits teams comfortable managing secrets via env files who prioritize runtime isolation and local Docker control over authentication convenience.