Schedchie Logo
Schedchie
Start free trial

Schedchie in OpenAI Responses APIAPI

Add social posting to your OpenAI agent

The OpenAI Responses API reaches Schedchie’s MCP server through its remote MCP tool with a connection token in the authorization field, and gets the posting, media and analytics tools. Schedchie is a flat-price social media scheduler for 20 platforms: Pro is €9.99 a month, with MCP included.

  • A tools entry of type mcp: server_label, server_url and the token in authorization.
  • The authorization value is not stored: send it with every request.
  • A connection token carries posting, media and analytics. The API runs no sign-in.
Start free trial

14-day free trialCancel anytime, no questions asked

See the 4 setup steps

Setting up OpenAI Responses API

The Responses API calls remote MCP servers for you. Give it Schedchie’s URL and a connection token, and your agent can schedule and report on posts.

Requirements

Access
An OpenAI API key. The Responses API takes remote servers over Streamable HTTP or HTTP/SSE and runs no sign-in, so you pass a token.

  1. OpenAI Responses APIConfig

    Add the MCP tool to the request

    POST this body to /v1/responses with your OpenAI key in the Authorization header. The Schedchie token goes in the tool’s authorization field.

    Config
    {
  "model": "gpt-6-astra",
  "input": "List my Schedchie accounts.",
  "tools": [{
    "type": "mcp",
    "server_label": "schedchie",
    "server_url": "https://mcp.schedchie.com/mcp",
    "authorization": "<connection token>",
    "require_approval": "never"
  }]
}

  2. OpenAI Responses APIConfig

    Send the token every time

    The Responses API does not store the authorization value. Include it in every request that uses Schedchie.

  3. OpenAI Responses APIPrompt

    Read the output

    The output lists the Schedchie tools in an mcp_list_tools item and each call in an mcp_call item. With require_approval set to always, or left out, answer each mcp_approval_request with an mcp_approval_response first.

Official docs: developers.openai.com/api/docs/guides/tools-connectors-mcp

Three prompts for OpenAI Responses API, with the tools they call

  • A request that turns a product update into an Instagram caption and a LinkedIn post, and schedules both for Friday at 12:00.

    list_accounts, schedule_posts

  • List the connected accounts and what each one has scheduled next week.

    list_accounts, list_posts

  • Which posts are waiting for my approval? Approve the LinkedIn one.

    list_pending_approvals, approve_post (Ultra)

Connection details

Same MCP server, permission groups and approvals as the Schedchie web app.

Tool entry
type mcp, server_label, server_url, authorization, require_approval, and allowed_tools to narrow the list.
Approvals
Left out, require_approval asks for approval before Schedchie gets any data: answer each mcp_approval_request with an mcp_approval_response.
Two credentials
Your OpenAI key in the Authorization header of the request; the Schedchie token in the tool’s authorization field.
Not stored
OpenAI does not keep the authorization value. Every request that uses Schedchie carries it again.
Sign-in
Your app handles OAuth itself if it wants one. The connection token skips that and carries posting, media and analytics.
Transport
Streamable HTTP or HTTP with SSE. Schedchie serves Streamable HTTP.
Only what you tick
Connect OpenAI Responses API to Schedchie. Your assistant gets the posting, media and analytics tools across 20 platforms, and only the permission groups you tick at sign-in.
Video from your phone
Your assistant can hand you an upload link for a video up to 200 MB and 15 minutes, then set your own cover on Instagram, Facebook, TikTok, Pinterest and YouTube.

More about Schedchie in OpenAI Responses API

What does a connection token give access to?

Posting, media and analytics: schedule, edit and retry posts, upload media, read analytics. Team, approvals, library and engagement need the browser sign-in. A token lasts 30 days; generate a new one in Dashboard, AI assistants.

Where does the token go: headers or authorization?

In the tool’s authorization field, as OpenAI’s guide shows. The Authorization header of the HTTP request is your OpenAI API key.

Should require_approval be never or always?

never lets the model call Schedchie straight away. always, the default when you leave it out, returns an mcp_approval_request first, which your app answers with an mcp_approval_response in the next request. Use always, or allowed_tools, if your agent should not post on its own.

Does an OpenAI client get ChatGPT’s smaller tool set?

No. ChatGPT’s 14-tool core set is for clients that sign in with OAuth under an OpenAI name. A connection token gets the posting, media and analytics tools.

Is the MCP connection billed on top of the plan?

No. It is included on every plan and in the 14-day free trial. Pro is €9.99 a month, flat. Posting to X is paid in X credits, as in the web app.

Are there limits?

Each connection can make up to 120 tool calls a minute. On Pro it gets 300 tool calls an hour, plus 30 session opens that make no call; on Ultra four times that, 1,200 calls and 120 opens.

14 days on your own accounts, then €9.99 a month

The assistant connection is included on every plan and in the trial.

Start free trial

14-day free trialCancel anytime, no questions asked

Other assistants on the same server

OpenAI Responses API and its logo belong to OpenAI. Schedchie is not affiliated with or endorsed by OpenAI.