Post to all your socials from your AI agent

Ten tools. It lists the channels you have connected, reads what a network requires before it will accept a post, writes and schedules the post, generates an image or a video, and uploads your own media. It is the same agent you get inside the app, with the same ten tools. Reading your numbers back is not one of them yet, so analytics stay in the dashboard.

Claude, Claude Code, ChatGPT, Codex, Cursor, Gemini CLI, OpenClaw and Hermes, each with a setup guide of its own. The real rule is broader: anything that speaks MCP or can run a shell command works, which covers n8n, Make.com, Zapier, your own cron jobs, your CI, and an agent you wrote yourself.

No. MCP is the open standard that lets an assistant call outside tools, and the only part you touch is one URL. Paste it into your assistant's connector settings, approve the tools it lists, and you are done. Connecting each social account takes about two minutes and needs no platform API keys of your own.

That is yours to set, one instruction at a time. Ask for drafts and you get drafts: they sit on the calendar with no publish attached until you give them a time. Ask for a scheduled post and it will go out at that time, which is the point of scheduling. Either way you can open it, rewrite it, move it or pull it back to draft, and pulling it back stops a publish that has already started.

Never, and there is nowhere to type one. Each network is connected once through its own official sign-in, which hands us a token rather than a password, and the tokens are encrypted where they are stored. Your agent only ever holds a postqueen API key. Revoke that key and the agent is cut off with your channels left connected.

Thirty, each through its official API: Instagram, X, TikTok, LinkedIn, YouTube, Facebook, Threads, Bluesky, Mastodon, Pinterest, Reddit, Discord, Telegram, Slack, Twitch and fifteen more, down to the blogs and the newsletter list. Each one has its own page here with what it can and cannot schedule, because the platforms genuinely differ.

One connected account is one channel. Three Facebook pages are three channels rather than one, and two accounts on the same network are two channels, which is allowed and common. Posts are unlimited on every plan. The only number a plan bounds is how many channels you connect at once, and you can disconnect and swap them whenever you like.

Every post travels through that network's own official API, the route those platforms built and document for scheduling tools. Nothing here drives a browser in the background or logs in as you. What arrives is a normal post on your own account, published at the time you picked. What a network's ranking does after that is its own business, and no scheduler can honestly promise otherwise.

Yes, in the dashboard, and over the CLI and the REST API too. Coverage follows the platforms: about ten of the thirty publish an insights API, and those are the channels with numbers against them. The rest are blank here for now. Your agent cannot read any of it back yet, so this is the one part of the job that stays in the app.

Yes. Channels group into customers, so each brand or client keeps its own channels and its own calendar inside one account, and you switch between them without logging out. Your agent can be pointed at a single group as well, which is how an agency keeps one client's week from landing on another client's channels.

Then do not. The calendar, the composer, the picture editor and the analytics are a normal application you can run entirely by hand. The CLI, the REST API and the n8n node are there for automation without a conversation in it. The agent is the fastest way in, not the only one, and nothing is locked behind it.

postqueen is open source under AGPL-3.0, and a self-hosted install with no payment keys configured runs without limits. Bring your own server and your own platform credentials and it costs nothing. Cloud plans are for people who would rather not run infrastructure, and every one of them includes the CLI, the REST API, the Node SDK, the n8n node and the MCP server. Plans are on the pricing page.