We now have an experimental version of a Frame.io MCP Server, selectively available in private beta, so you can drive your Frame.io activity with an AI agent.
If MCP’s new to you, it’s an open-source standard introduced by Anthropic for connecting AI applications to external systems. (You can find more on this in Anthropic’s original announcement from November 2024.)
The server exposes Frame.io as a set of tools that MCP-compatible clients (Claude, ChatGPT, and Cursor, to name a few) can use to carry out requests.
By connecting the MCP server to your client of choice, the agent can navigate workspaces and projects, read comments, create shares, search, handle uploads, and a fair bit more, all mapped to the API.
These are the same endpoints the API already exposed, so nothing here is net new functionality. What changes is how you combine them. Stringing calls together, passing output from one into the next, and shaping the result becomes a matter of prompting rather than scripting.
For example, you can prompt your AI agent to:
generate a report of all users that have uploaded content into your account in the last 30 days, and for each one, the name of the file, file type, and which project and workspace it lives in.
It will handle the traversal and pass IDs between calls on its own.
It’s still early days, and we’re looking for people who want to try this out with their creative workflows or other in-app automations.
If you’ve got work you’d hand to an agent, or if you want to see how you can take Frame.io actions right from the chat mode of your AI-tool of choice, reply here and we’ll chat about getting you set up. Your use cases and feedback will help shape how it develops from here.