Skip to main content

​
Working with her in ChatGPT

ChatGPT is where a lot of content gets written. With PostQueen connected it is also where content gets shipped: ask for two posts about the new menu and she writes the copy, generates the images and puts both on your calendar while you keep chatting.
ChatGPT with the PostQueen connector: it generates two post images and schedules them to Instagram and X
There are two routes in, and they solve different problems:
  • The connector. One URL in ChatGPT settings, and her 11 MCP tools show up in your chats. This is the route most people want.
  • A Custom GPT action. You are building a GPT of your own and want it to call her public API directly, with no MCP in the middle.

​
Route 1: the connector

ChatGPT reaches her over MCP. You paste one URL under Settings, then Connectors, and she becomes a tool the model can call:
Your key lives in the app at app.postqueen.ai: open Settings > Developers > Public API and click Reveal.
Your API key travels inside that connector URL, which makes the URL itself a secret. Anyone holding it can post from your accounts. Keep it out of screenshots, and if the key leaks, regenerate it in Settings > Developers > Public API and update every client that uses it.
Custom connectors sit behind ChatGPT’s developer mode, which OpenAI offers on the paid web plans. The full walkthrough covers the developer mode toggle, the connector form and enabling her per chat:

ChatGPT connector setup

Step by step, from the developer mode switch to your first tool call
Once she is switched on in a conversation, ask for your connected channels. If ChatGPT calls integrationList and hands back your accounts, you are connected.

​
Route 2: a Custom GPT action

Building your own GPT? Give it an action that talks to her public API. The endpoint you want is the one that creates posts:
Authentication is a single header carrying the raw key, with no Bearer prefix:
A few things worth knowing before you wire the action up:
  • type decides the outcome. Use schedule with a date, now to publish immediately, or draft to save the post on the calendar without scheduling it.
  • settings.__type tells the backend which platform schema to validate against. Each provider has its own required fields, listed in Create Post.
  • Integration IDs come from GET /public/v1/integrations, so a useful GPT usually gets that action too.
  • Images have to be uploaded through her first. Each entry in image is an object with an id and a path that came back from an upload, because providers reject raw file paths and outside links. The endpoint: Upload File.
The create-post endpoint allows 30 requests per hour. One request can carry several posts, so a GPT that batches a week into a single call gets a lot done inside that budget. Self-hosting? Point the action at https://postqueen.example.com/api/public/v1 and tune the limit with the API_LIMIT env var.

​
👑 Ask her for two posts

You describe what you want, and she takes care of the rest:
You: Create 2 posts about our cafe’s new dessert menu with photos, schedule them for this week.
ChatGPT: On it. Generating two images and scheduling them with PostQueen. generateImageTool running… “Latte week starts Monday. #coffee #newmenu” “The raspberry cake is here. #dessert #comehungry” 2 posts scheduled: Instagram + X, Thursday 10:00 AM. Both are on your calendar if you want to tweak them first.
Behind that one sentence sit two tools: generateImageTool produces each visual and hands back a hosted URL, then integrationSchedulePostTool writes the calendar entries. Bringing your own image from somewhere else? uploadFromUrlTool pulls it into her media library first, because attachments have to live on a PostQueen URL.

​
Staying in control

She runs on your terms. Leave the schedule alone and she is a true autopilot: whatever lands on the calendar goes out on time without another word from you. Want a closer look first? Every post is visible at app.postqueen.ai before it publishes, so you can edit or delete it, and a post she saves as a draft stays unscheduled until you publish it there.

​
FAQ

No. She works on her own through the dashboard, the CLI and the API. ChatGPT is one of several ways to reach her.
Everything her 11 MCP tools cover. She can list your channels and read what each one accepts, and she can generate images, upload media and schedule posts across your 30+ networks.
Yes. Ask for drafts and each post sits on your calendar unscheduled until you publish it there.
No. Agent access uses the same API key and plan limits as your regular PostQueen account, on any plan. ChatGPT’s own developer mode does require one of OpenAI’s paid plans.
The connector if you want her in your everyday chats with no build step. The action if you are shipping a GPT to other people and want to control exactly which calls it can make.

​
Learn more

Agents overview

Every way to put an agent in front of PostQueen, from chat apps to your terminal

MCP tools

The 11 tools ChatGPT calls once the connector is on

Create Post API

The endpoint behind a Custom GPT action, with every provider’s settings schema

MCP examples

Real prompts and the tool calls each one triggers