Threads MCP: Connect ChatGPT, Claude and Gemini

Threads MCP connects an AI app to tools that can work with your Threads account. With Postory as the MCP server, you can ask an AI assistant to prepare a post, check its limits, schedule it, or publish it after you approve the preview. This guide shows the connection steps and the decisions that still belong to you.

What a Threads MCP connection actually does

An MCP connection gives an AI app named tools rather than your Threads password. Postory holds the Threads connection and exposes actions such as listing accounts, creating a draft, checking a proposed post, scheduling a delivery, and reading its status. The assistant uses those tools in the conversation; you can see which account and draft it selected.

That distinction matters if you use more than one social account. Ask the app to list your connected accounts first, then name Threads and the account you want. Otherwise a tool that permits a default account may choose a different one. Save the account's connection_id from the tool response when you want an exact target in a longer workflow. Do not copy an ID from another person's setup guide: connections belong to the signed-in Postory user.

Postory's MCP feature guide lists the current tools and client setup paths. Its publishing flow is a draft followed by a preflight preview and your approval. You can stop after drafting if you are not ready to send anything.

Connect Threads to ChatGPT, Claude, or Gemini

First, sign in to Postory and connect your Threads account on the Accounts page. Then add Postory's remote MCP endpoint, https://postory.io/mcp, to the AI app. The app opens Postory sign-in and consent; approve the scopes you want that app to use. An app connection alone does not connect a Threads account, so finish both parts before testing a prompt.

ChatGPT on the web: Open Plugins, add a custom MCP server, enter the Postory URL, and complete OAuth. Install and enable the Postory plugin, then select it in the chat. OpenAI's custom MCP server instructions describe the current ChatGPT controls. Workspace restrictions may affect whether you can add a custom server.

Claude on the web or desktop: Open Customize → Connectors, add a custom connector with the Postory URL, authorize it, then enable Postory in the conversation's Connectors menu. Anthropic's remote connector guide explains that account and organization settings can affect availability.

Gemini: Eligible Gemini Apps users can add Postory under Settings & help → Connected Apps → Custom apps, then select @Postory in a prompt. Google's custom app help lists the eligibility and language requirements. If you use Gemini CLI, run gemini mcp add --transport http --scope user postory https://postory.io/mcp, authorize with /mcp auth postory, and check /mcp for its tools; see the Gemini CLI MCP guide. Client tool use can vary, so confirm the assistant actually returns the requested result.

Draft and schedule Threads posts from your AI chat

Start with a safe request: “List my Postory accounts and tell me which Threads account is ready. Do not publish anything.” The reply should identify the platform, account, and connection state. If it asks you to reconnect, do that in Postory before trying a publish prompt.

Then ask: “Create a Threads draft for that account about our product update. Keep it concise and show me the saved draft.” This creates a saved Postory item; it does not send it to Threads. Read the text and account in the response. Ask for a revision while it is still a draft, or archive it if you do not want it.

For a scheduled post, give the assistant a time with a time zone and ask it to run preflight_publish before scheduling. The preflight checks the selected account, each part's text and media, Threads limits, timing, and the cost that applies at the charge step. Postory allows up to 500 characters and five links per Threads post. Its preview should show the final copy and delivery time. Only after checking those details should you explicitly approve the schedule. You can later ask it to list, reschedule, or cancel a queued delivery.

The same pattern works for a multi-part Threads post: draft the parts, read every part in order, preflight the entire thread, then approve publication or scheduling. The assistant should not treat “write a thread” as permission to publish it. If a delivery is pending, ask for its status instead of sending the draft again.

Limits and checks that prevent the wrong post

Account identity: A tool response for another platform is not a Threads result. Re-run list_accounts and specify the Threads connection_id if the app picked the wrong account. If you have several Threads accounts, use the exact one returned by your own grant.

Publishing approval: A draft can be created without sending a social post. Publishing and scheduling require the preview and a separate yes in chat, or approval on Postory's page when you turned on manual approval there. A prompt such as “show me what you would post” should end at the preview. Check the account, text, media, time, and quoted cost at that point.

Platform scope: Threads follower and following rosters are not available through Postory's MCP tools. Postory can work with imported Threads posts and replies, but those reads need an import first. A Threads shadowban check accepts a handle; the Profile Audit, Following check, and recovery scan remain X tools. For X-specific setup, use the separate X/Twitter MCP guide.

If the AI app displays an error, check whether it actually invoked the Postory tool and whether its answer contains the result. A successful connection screen or server call does not by itself prove that the app showed the requested content. Retry a small read such as list_accounts, and use the MCP setup page to confirm that client’s current path.

Start publishing to Threads with Postory

Connect Threads in Postory, add the MCP server to your AI app, and try one draft before scheduling a real delivery. The Postory MCP guide gives you the setup path for each client and the tools you can use after connecting.

Try Postory free to create a draft and review the publish preview from your AI chat.