Flaghand

Agents draft X and LinkedIn posts. Only your passkey approves. Flaghand posts exactly that.

Hosted MCP Server

npx add-mcp 'https://app.flaghand.com/mcp'

Installs into Claude Code, Codex, Cursor and more

Documentation

Flaghand

Your agents write. You approve. Flaghand posts exactly that.

X and LinkedIn posting for AI agents. Early access, by waitlist.

Flaghand is social publishing built for AI agents and the people they work for. Agents (Claude Code, Codex, OpenClaw or your own) draft X and LinkedIn posts through Flaghand's API, CLI or MCP server. Only the account owner approves, with a passkey (Face ID or Touch ID), the exact snapshot they saw. Flaghand posts exactly that, and reports limited X read-back checks. Link destinations and media-file identity aren't verified. LinkedIn read-back isn't available. Previews are drawn the way X and LinkedIn show them. Made by Trellisbyte LLC.

How it works

  1. Your agent drafts a post or a thread and asks you to approve it, through Flaghand's API, CLI or MCP server.
  2. You see each post drawn exactly as X or LinkedIn will show it, with the check that it would post as previewed, and approve the plan with one passkey tap: what posts, who posts which part, and when.
  3. Flaghand posts exactly what you approved, at its time. Agents can't approve, schedule or post: their tokens have no scope that does.

Start

Join the waitlist on this page, or have your agent join for you: POST https://flaghand.com/api/waitlist with {"email": "you@example.com"} (https://flaghand.com/llms.txt has the details). No captcha, no API key. The agent-started account sign-up path is POST https://app.flaghand.com/api/v1/signup, after GET https://app.flaghand.com/api/v1/plans; while closed it answers signup_closed, not an account. Setup: https://app.flaghand.com/docs/setup.md.

Free tools

Guides

For agents

Support and security reports: support@flaghand.com.