MCP server for social media: let an AI agent post for you

An MCP server for social media is a Model Context Protocol endpoint that lets an AI assistant call tools to post. PostDodo runs one, so an agent can list your connected accounts, schedule or publish across all 10 networks, and read back a confirmed live link, with your API key gating every action.

What is an MCP server?

MCP, the Model Context Protocol, is an open standard for connecting AI assistants to external tools. An MCP server is the piece that exposes those tools. Once your assistant is pointed at one, it can call the server’s actions inside a normal conversation, the same way it reads a file or runs a search. A social-media MCP server exposes posting tools, so the assistant can draft, schedule, and publish to your accounts instead of you pasting text into each app by hand.

Why would you let an AI agent post to social media?

Because the work already happens in the chat. You brief an assistant on a launch, it drafts the copy, and with an MCP server it can put that copy straight into your schedule with no context-switch. Common uses:

The agent does the busywork; you keep the judgment.

The catch: an agent should not post publicly without your review

Publishing is public and permanent. Once a post goes out, your followers and the platform have already seen it, and deleting it later does not undo the impression. That is the real risk in handing posting to an agent: a wrong draft, an invented claim, or the wrong account, live before you looked. So keep the rule simple. An agent can draft and queue, but a human should approve what actually goes public. And remember the default: with create_post, leaving out the schedule time publishes immediately, so treat it like a real send button, not a sandbox.

How do you connect PostDodo’s MCP server?

PostDodo exposes its MCP server at one HTTPS endpoint, authenticated with a PostDodo API key. Three steps:

{
  "mcpServers": {
    "postdodo": {
      "url": "https://postdodo.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pd_live_YOUR_API_KEY"
      }
    }
  }
}

The key carries the same account access and billing gate as the app and the REST API, so keep it server-side and never ship it in client code. A missing or invalid key is rejected with 401. Prefer plain HTTP over an MCP client? The same key drives the social media scheduling API, and the developer docs list the endpoints.

What tools does the PostDodo MCP server expose?

Three, and only three. They mirror the app and the REST API, so an agent can run the whole loop: see your accounts, post, then check what went out.

ToolWhat it does
list_accountsLists the connected social accounts the agent can post to, each with its id, platform, handle, and whether it is active.
create_postPublishes now or schedules a post across one or more connected accounts. Omit the schedule time to publish immediately, or pass an ISO 8601 future time to schedule.
list_postsLists your recent posts, newest first, with per-platform delivery status and the live post URL for each.

The flow is straightforward. Call list_accounts to get the account ids, then pass those ids to create_post with your text. Leave out the schedule time and it posts now; include an ISO 8601 future time and it is queued. You can attach media by public URL, add a first comment, or chain a thread on the networks that support one (Bluesky, Mastodon, X, and Threads). Instagram needs at least one media item. Afterwards, list_posts reads back each target’s status and live link.

Which networks can the agent post to?

All ten PostDodo supports, from a single create_post call: Bluesky, Mastodon, Facebook, Instagram, Threads, X, LinkedIn, TikTok, YouTube, and Pinterest. Connect each account once in the dashboard, and the agent sees it in list_accounts and can target it by id. One call can fan a post out to several networks at once, and each target reports its own status and live link.

How does PostDodo keep an agent’s posts safe and auditable?

Two things keep this safe: a human stays in the loop, and every action leaves a receipt.

None of this removes your judgment. Posting is irreversible, so keep the approval step: let the agent draft and queue, review the queue, and let create_post publish only what you signed off on. The live-link receipt is what makes an agent’s actions auditable after the fact.

Want an AI agent that can actually post, without handing over the keys? PostDodo ships a real MCP server for all 10 networks, with a confirmed live-link receipt on every post, on flat pricing with no per-network fee. Generate an API key on any paid plan, point your MCP client at https://postdodo.com/api/mcp, and let the agent draft while you approve. The developer docs have the full contract.

Frequently asked questions

What is an MCP server for social media?

It is a Model Context Protocol server that exposes posting tools to an AI assistant. Once your assistant connects to it, it can list your connected accounts, then schedule or publish posts to them from a normal conversation. PostDodo runs one for all 10 networks, authenticated with your API key.

Can an AI agent post to social media on its own?

Yes, technically. With PostDodo's MCP server an agent can call create_post to publish immediately or schedule for later. But posting is public and irreversible, so the safer pattern is to let the agent draft and queue, and keep a human approval step before anything goes live.

How do I connect PostDodo's MCP server?

Get a PostDodo API key from Settings, then API, on a paid plan. Add the server to your MCP client config with the key as a bearer token, pointed at https://postdodo.com/api/mcp. Your assistant will then see the list_accounts, list_posts, and create_post tools.

What tools does the PostDodo MCP server expose?

Three: list_accounts returns your connected accounts and their ids, create_post publishes now or schedules across one or more accounts, and list_posts reads back recent posts with per-platform status and live links. They mirror the app and the REST API, so the agent can post and verify what went out.

How do I know what the agent actually posted?

Every post carries a receipt. create_post reports success only when the network confirms it, and list_posts returns each target's status and the platform's own live URL. So you can open the exact link the agent published, per network, and confirm it went out as intended.

Which social networks does the MCP server support?

All ten PostDodo supports, from one create_post call: Bluesky, Mastodon, Facebook, Instagram, Threads, X, LinkedIn, TikTok, YouTube, and Pinterest. Connect each account once in the dashboard, and the agent can target any of them by the id from list_accounts.