For AI buildersREST + MCP · free

Social media posting for AI agents

AI agents cannot post to social platforms on their own. Schedule & Chill is the layer that lets them: one REST API and a native MCP server that publish to LinkedIn, Bluesky, Mastodon and YouTube. Your agent calls one endpoint; the queue, OAuth, retries, and platform rules are handled.

Free · No credit card · No trial timer

Two ways to give an agent access
MCP
https://schedulenchill.com/mcp-oauthPoint the client at that URL and the agent gets 22 typed tools. No API key: it runs an OAuth flow and holds its own token. Best when the model decides what to do.
REST
POST /api/posts with content and account ids. Best when your code decides and the model only writes copy.
Both
Share the same quota and the same managed publishing queue. REST always uses an API key; MCP takes either that key or the OAuth URL above.
The problem

Publishing is the part nobody wants to build

Why agents can’t just post

  • Every platform has its own OAuth flow, token refresh cycle, and app review process — 10 platforms means 10 integrations to build and keep alive.
  • Media rules differ per network: aspect ratios, file size caps, video length, and whether photos and video can be mixed in one post.
  • Rate limits and transient failures mean posting needs a queue with retries, not a single HTTP call inside an agent loop.
  • Agents hallucinate. Publishing is irreversible, so the system needs a review window and a cancel path.

What Schedule & Chill handles

  • One OAuth flow per platform, completed once in the dashboard, then reusable by every agent.
  • A managed queue with retries — the agent fires and forgets, and the queue owns delivery.
  • Per-account delivery status, including partial publishes where one platform succeeded and another failed.
  • Platform media rules enforced before the post leaves, so the agent gets a clear error instead of a silent failure.
  • Scheduling with a future timestamp, so a human can review and cancel before anything goes live.
  • A media library the agent can search and attach from, instead of re-uploading the same asset.
The call

One call from anywhere

Same result whether it comes from a Python agent, a Node worker, or a shell script — and the MCP path below needs no code at all.

REST — your code decides

bash
curl -X POST https://schedulenchill.com/api/posts \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -d '{
    "content": "Shipped: per-account delivery status.",
    "social_account_ids": [789, 790],
    "scheduled_at": "2026-08-01T14:00:00Z"
  }'

MCP — the model decides

connector
# Paste this URL into Claude, ChatGPT,
# Cursor or any MCP client. No key.
https://schedulenchill.com/mcp-oauth

> "Schedule a LinkedIn post about our
  launch for Tuesday at 9am."

The agent gets 22 tools, 4 resources and 2 prompts, with schemas it can reason about — including reading the existing queue before adding to it.

Guardrails

Shipping agents safely

Publishing is irreversible. These three defaults keep an autonomous agent from becoming a public incident.

Schedule, don’t publish

Give agent-created posts a future timestamp. The review window costs nothing and cancel_post undoes anything wrong.

Fetch ids, never guess

Call get_accounts at the start of a run so the model picks from real accounts instead of inventing ids that fail or hit the wrong channel.

Read the queue first

list_scheduled tells the agent what is already planned, so it does not duplicate a post a human already wrote.

Questions

Agents and social posting

Yes, if you give it a posting API. Models cannot reach LinkedIn or Bluesky on their own. Schedule & Chill provides that layer — a REST API for code-driven agents and a native MCP server for tool-calling agents — covering LinkedIn, Bluesky, Mastodon and YouTube.

Not by itself, but connecting it takes one paste. Add https://schedulenchill.com/mcp-oauth under Settings → Connectors → Add and approve the consent screen — no API key. That needs a paid ChatGPT plan with developer mode on. Any other MCP-capable client or custom GPT that can call an HTTPS endpoint can use an API key instead.

Ship the agent, not the OAuth

Create a free account, connect a channel, and give your agent one URL. The queue, retries and platform rules are already handled.

Free · No credit card · No trial timer

Last updated September 4, 2026