Let an AI agent run your social posting
Connect Claude Code, OpenClaw, or Cursor to posthell. Enable publishing and your agent posts to your networks on its own. Only X is metered, and your credit wallet caps the spend. Pick a network to see how.
Your agent posts. You stay in control.
Give your AI agent a posthell key, turn on publishing, and it drafts and posts to your networks on its own. Only X spends credits, and your wallet caps it. Set it up once and tell your agent what you shipped.
What agent automation means here
posthell runs an MCP server, so an AI agent you already use, Claude Code, Cursor, OpenClaw or any MCP client, can draft and schedule posts through your account. You give the agent an API key, it writes posts into your queue, and you approve them.
Publishing without a human tap is off by default and is a per-key opt-in, not an account-wide switch. Even with it on, your credit wallet is the hard cap: an agent that runs out of X credits simply fails that post rather than running up a bill.
Who this is for
- Developers already living in an agent. If Claude Code or Cursor is where your day happens, posting from there beats opening another tab.
- Anyone automating a build log. Ship something, have the agent draft the post from the commit or the changelog, approve it later.
- Not the default path. If you just want your posts written for you, the app does that without any of this. The agent surface is a side door for technical users.
How to set it up
- Create an API key. In the dashboard, under Agent access. Keys are shown once, and you can hold up to 5 active.
- Point your client at the MCP server. One `claude mcp add` line, or the equivalent in Cursor or VS Code. There is also a `posthell-cli` npm package if you would rather script it.
- Choose the trust level. Leave the key on drafts-only and every post waits for your approval, or flip that one key to publishing and let it post inside your wallet cap.
Questions people ask
Can an AI agent post to social media for me?
Yes. posthell runs an MCP server, so Claude Code, Cursor or any MCP client can draft posts into your queue with an API key. Publishing without your approval is off by default and is a per-key opt-in, capped by your credit wallet.
What stops an agent running up a bill?
The credit wallet is a hard cap. Only X is metered, and when an agent runs out of X credits that post fails rather than being charged. Assistant includes 400 X posts a month, Team 600 and Department 1,000.
Do I need to be technical to use posthell?
No. The agent and CLI surface is optional. The normal path is the app: you tell posthell what happened in your business and approve the drafts it writes, with no API key involved.
Let it write this week's posts
Tell posthell what happened in your business this week. It writes the posts, tailors them per network and schedules them. Nothing publishes until you approve it.
7 days free, then from $37/mo billed yearly. Cancel anytime.