klavis

Klavis’ MCP server provides the Strata-based integrations and REST endpoints to manage and run multiple MCP servers (such as Strata and other prebuilt connectors) via a single API surface. The quickstart shows you can deploy the server using Docker and then interact with it through the REST API or the Klavis SDKs to create and manage Strata instances or individual MCP servers. Once running, you can create a Strata server by sending a request to the Strata endpoint with user context and target servers (e.g., Gmail, Slack). You can also use the Python or TypeScript SDKs to programmatically create strata instances or individual MCP servers, which abstracts the REST calls into familiar client code. The REST API section of the Quick Start demonstrates how to initiate a Strata server or a single MCP server, including how to authenticate with your API key. This MCP server thus acts as a centralized control plane for configuring, starting, and orchestrating multiple AI agent integrations through Strata and other connectors.

Prerequisites:

• Docker installed on your machine (Docker Desktop or equivalent).
• Access to a Klavis API key if required by the deployment (as noted in env vars).

Install steps:

1. Ensure Docker is running on your machine.
2. Pull and run the MCP server image:

docker pull ghcr.io/klavis-ai/github-mcp-server:latest docker run -p 5000:5000 ghcr.io/klavis-ai/github-mcp-server:latest

3. (Optional) If you need to pass an API key or other environment variables, run the container with -e flags, e.g.:

docker run -p 5000:5000 -e KLAVIS_API_KEY=your-api-key ghcr.io/klavis-ai/github-mcp-server:latest

4. Verify the server is running by hitting the REST endpoint used by Klavis (until you have your own API key, you can refer to the REST usage examples in the docs):

curl -X GET http://localhost:5000/v1/status

Prerequisites recap:

• Docker installed
• Optional: Klavis API key for authenticated endpoints
• Basic familiarity with Docker commands and REST API basics

Tips and notes:

• The server is typically exposed on port 5000 (as shown in the Docker run example); adjust if you map to a different port.
• Use the REST API or the provided SDKs (Python or TypeScript) to create Strata instances or individual MCP servers programmatically.
• Keep your KLAVIS_API_KEY secure; rotate keys per your security policy.
• If you’re running locally behind a firewall or VPN, ensure the port you expose (e.g., 5000) is accessible to your client applications.
• When troubleshooting, check Docker container logs for errors, and consult Klavis docs for supported server names and endpoint formats.
• The MCP server supports batch operations via Strata to manage multiple connectors efficiently; refer to docs for available server names and configurations.