Frequently Asked Questions

What is DeployContext?

DeployContext is a cloud hosting platform for Model Context Protocol (MCP) servers. It enables developers to deploy MCP servers directly from GitHub repositories, provides remote access via HTTP and SSE transports, and implements OAuth 2.1 authentication as required by Anthropic's specifications. Think of it as a one-click deployment platform purpose-built for MCPs.

What MCP runtimes are supported?

DeployContext supports three runtimes, automatically detected from your repository:

Node.js — Detected via package.json. TypeScript is fully supported.
Python — Detected via pyproject.toml or requirements.txt. Works with FastAPI and the MCP SDK.
Bun — Detected via bun.lockb. Bun-native projects are supported.

How does deployment work?

When you deploy an MCP, the platform performs the following steps:

Clones your GitHub repository
Detects the runtime (Node.js, Python, or Bun) and transport mode
Generates a Dockerfile and builds a container
Deploys the container to cloud infrastructure
Runs a health check to verify the MCP responds correctly
Registers the endpoint so it is accessible via your unique URL

Most deployments complete in under 5 minutes. See the Deploy Guide for full details.

What transport modes are supported?

DeployContext supports three transport modes, all detected automatically:

Native HTTP/SSE — For MCPs that implement HTTP endpoints directly. This is the recommended approach.
stdio Bridge — For MCPs that use standard input/output. The platform automatically wraps them with an HTTP/SSE bridge, requiring no code changes on your part.
SSE Flag — For MCPs that accept a --transport sse flag.

What is OAuth 2.1 and why is it required?

OAuth 2.1 is the authentication and authorization standard required by Anthropic for remote MCP servers. It ensures that only authorized users and applications can access your MCP's tools and data.

DeployContext implements the full OAuth 2.1 specification, including dynamic client registration (RFC 7591), PKCE (RFC 7636), resource indicators (RFC 8707), and token introspection (RFC 7662). This is handled by the platform so you do not need to implement OAuth yourself.

How do I connect my MCP to Claude Desktop?

After deploying your MCP, you can connect it to Claude Desktop:

Copy the endpoint URL from your DeployContext dashboard
Open your Claude Desktop configuration file
Add your MCP endpoint under the mcpServers section
Restart Claude Desktop to load the new configuration
Your MCP's tools will appear in Claude Desktop and be available for use

You can also test your MCP directly using the MCP Inspector before connecting it to Claude Desktop.

Is DeployContext free?

DeployContext offers a free Hobby tier to get started. For higher usage, additional features, and increased rate limits, Pro and Scale tiers are available. Visit the Monetization docs to learn about pricing options for both creators and subscribers.

What are the rate limits?

Rate limits are enforced at the edge for each API key:

MCP endpoints: 60 requests per minute per API key
Platform API (Hobby): 100 requests per hour
Platform API (Pro): 1,000 requests per hour
Platform API (Scale): 10,000 requests per hour

When limits are exceeded, the API returns HTTP 429 (Too Many Requests). Implement exponential backoff in your client for automatic retry. See the API Reference for more details.

Can I use private GitHub repositories?

Yes. When you sign in with GitHub and grant the repo scope, DeployContext can access your private repositories. Your GitHub access token is stored encrypted and is only used to clone your repository during deployment.

Can I add environment variables to my MCP?

Yes. You can configure environment variables for your MCP through the deployment settings in your dashboard. Environment variables are encrypted at rest and injected into your MCP at runtime. This is the recommended way to provide API keys, database credentials, and other configuration to your MCP server.