Connect Octopost to Claude, ChatGPT, Cursor or any MCP client. Ask about your analytics, review what is scheduled, and draft posts in plain language — using the same permissions and approval rules as the app.
The server URL is the same everywhere. Your client walks you through signing in to Octopost and picking the workspace it may access.
https://octopost.ai/mcp
Run this in your terminal:
claude mcp add --transport http octopost https://octopost.ai/mcp
Settings → Connectors → Add custom connector, then paste the server URL above. A browser window opens for you to sign in and choose a workspace.
Settings → Connectors → Create, paste the server URL, and authorize when prompted. Requires a plan where custom connectors are available.
Any client that supports remote MCP over HTTP with OAuth works. Point it at the server URL — discovery and sign-in are handled for you.
A connected assistant can draft, schedule, edit and cancel posts out of the box. It cannot publish straight to your social accounts — a published post reaches real followers and Octopost cannot take it back.
If you want that, enable Allow publishing immediately for that specific client in Settings → API → App Clients. Turn it back off at any time; existing connections respect the change on their next request.
Fourteen tools, grouped by the permission they need. Your assistant only sees the ones you granted.
The workspace you pick when authorizing is the only one that connection can touch. To work in another workspace, add a second connection and authorize it separately.
If your workspace requires editor approval, posts an editor creates through MCP land in the approval queue just as they would in the app. Approving is done by a person, in Octopost.
MCP uses the same API access and monthly scheduling limits as the REST API. If a limit is reached, your assistant is told why and what to do about it.
Delete the client in Settings → API → App Clients and every open session stops working on its next request.
Tool calls carry JSON only, so a file cannot travel over the protocol. Upload media in Octopost, or have your assistant import it from a public URL.
Building your own integration instead? The REST API reference covers the same capabilities over plain HTTP.