PostStream Docs

Connect an AI agent — Claude, Cursor, or any MCP client — to your PostStream account, and it can schedule posts, check what published, and explain failures, all in plain language. Everything it does still runs through the same checks as the dashboard, and nothing publishes without your approval.

Start here

How AI agents work

What an MCP agent is, the 9 things it can do for you, and how it stays safe.

Connect an agent

Step-by-step: Claude Desktop, Cursor, or any MCP client. 5 minutes.

Connect an agent

Claude Desktop

Create a token, paste one config block, restart Claude. Done.

Cursor

The same idea, in Cursor's mcp.json.

Any MCP client / one-click Connect

Point a remote client at our URL with no token and approve it in the browser.

Troubleshooting

401s, 402s, wrong URL, expired token — and the fix for each.

The plugin

The @poststreamai/mcp package

Install, configure, upgrade. The small program that bridges your agent to PostStream.

npm package page↗

Versions, install size, changelog.

Reference

Tools & permissions

Every tool the agent can call and the permission it needs.

Auth, errors & limits

Tokens, scopes, endpoints, error codes, rate limits.

OpenAPI / Swagger spec↗

The machine-readable REST spec for the backend, for developers.

Help

FAQ

Do I need a paid plan? Is my data safe? Can it post without me? Answered.