Besides ChatGPT and Claude, Magisterium AI works with any client that speaks the Model Context Protocol (MCP) or our agent-to-agent (A2A) endpoint.

Supported clients (examples)

• Cursor, VS Code Copilot, and other IDEs with MCP support.
• n8n, Zapier, and similar automation platforms via their MCP integrations.
• Custom agents built on frameworks like LangChain, LlamaIndex, or your own.
• Cloud agent platforms with A2A support.

The specific setup varies by client, but the pattern is the same: point the client at our MCP or A2A endpoint and authenticate with your Magisterium AI API key.

Configuration

• MCP endpoint URL — see magisterium.com/developers/docs/mcp.
• API key — create one at magisterium.com/developers/api.

Paste both into your client's MCP config.

Common tools exposed

Once connected, your client can use tools such as:

• Search the Magisterium AI library (retrieve relevant magisterial or theological passages).
• Query documents for passage-level citations.
• Scripture lookups across Catholic translations.

Tool names and schemas are documented on the Developers page.

A2A

For agent-to-agent workflows that don't support MCP natively, we offer an A2A endpoint. The documentation on magisterium.com/developers covers the setup.

Plan-based limits

MCP and A2A both require a paid plan. Free accounts get a PLAN_REQUIRED error.

• Pro / Organization: Unlimited requests per day, 15 / minute combined MCP + A2A.
• Enterprise / Large Organization: Custom — contact us.
• The Q&A pipeline (chat / catholic_qa) is capped at 2 / minute on paid plans.

Rate-limit response headers tell your client what's available.

Help with a specific client

If you're setting up a new or unusual client and something isn't working, contact us and choose API. Include your client name, the error you're seeing, and any logs.