Feature request: expose DonorDock as an MCP (Model Context Protocol) server.
Otto is great inside the platform, but there's currently no way to let an outside AI client work with our DonorDock data. Today the only path is Zapier MCP, which is action-oriented — I can create a contact or a gift, but I can't ask questions across our records.
What I'd want from a native MCP server:
  • Read tools: query contacts, gifts, pledges, campaigns, appeals, funds, marketing lists, activities — with filters and date ranges
  • Write tools: create/update contact, create gift, create activity, log an interaction
  • Auth via the existing API key/secret + tenant ID, with scoped read-only vs read-write keys
  • A remote HTTPS endpoint so it works in Claude, ChatGPT, Cursor, etc. without local setup
  • What is the last name of the person who gave on Tuesday?
Concrete use cases: "Which recurring donors lapsed in the last 90 days?", "Summarize giving by campaign for the board packet", "Draft follow-ups for everyone who gave to the spring appeal but not the fall one." That's analysis work that's tedious to do by export-to-spreadsheet and is exactly what an agent is good at.
Also worth noting: publishing a public API reference would help even ahead of MCP — right now I can't find documented endpoints, so building our own wrapper isn't practical either.