Schedule & Chill MCP server

Schedule & Chill MCP Server

A hosted Model Context Protocol server that lets an AI agent schedule and publish social media posts — LinkedIn, Bluesky, Mastodon and YouTube.

No install, no API key, no glue code. Paste one URL into your MCP client, approve a consent screen, and the agent gets 22 typed tools.

https://schedulenchill.com/mcp-oauth

Free. There is no paid plan and no credit card.

Connect it

Claude Code

claude mcp add --transport http schedule-and-chill https://schedulenchill.com/mcp-oauth

Then run /mcp and approve the browser consent screen.

Claude Desktop / ChatGPT / Cursor / VS Code / Zed

Add a remote (HTTP) MCP server with the URL https://schedulenchill.com/mcp-oauth. Each app words it differently — "Add custom connector", "Add MCP server → HTTP", "Add remote server" — but they all want the same URL. A browser window opens once for consent.

Anything that only takes a config file

{
  "mcpServers": {
    "schedule-and-chill": {
      "type": "http",
      "url": "https://schedulenchill.com/mcp-oauth"
    }
  }
}

API key instead of OAuth

There is a second endpoint for scripts and clients that would rather send a header:

https://schedulenchill.com/mcp     Authorization: Bearer YOUR_API_KEY

Same tools, same behaviour. Create a key in Settings → API keys.

You will need a free account at schedulenchill.com and at least one connected social channel.

Two transports, one server

Both are streamable HTTP and expose an identical surface: 22 tools, 4 resources, 2 prompts.

What the agent can do

Full reference with parameters: docs/tools.md.

Three things that will save you a bug report

"next_slot", or an ISO-8601 timestamp. Publishing is irreversible, so it is never assumed — omitting when is an error, not a silent publish.

1. schedulepost requires when.** There is no default. Pass "draft", "now",

silently dropped and the call still reports success. A guessed key does not error; it publishes with the platform default, which can mean a post you meant to restrict goes out public.

1. Call getcapabilities before sending platformoptions. Unknown option keys are

but the same slot is returned to every caller until something occupies it. For a single post prefer when="next_slot" — the server resolves it at write time, so a batch never lands on the same minute.

1. Queue slots are computed, not reserved. getqueueslots shows a human real times,

Reversibility

cancelpost and deletepost soft-delete, and restorepost brings a post back. A post whose time has passed is restored as a draft rather than re-armed, so restoring never publishes anything by itself. publishnow and retry_post send immediately and cannot be undone — confirm with the human first.

Registry

Shortened. The full README is on GitHub.