# Team approval before an AI agent posts on X and LinkedIn To have your team approve an AI agent's posts before they go live, keep three jobs apart: the agent drafts, your team or client reviews an exact preview, and one owner approves with a passkey. In Flaghand the agent can't do the last step at all. No agent token can approve, schedule or post, so a post goes out only after the owner approves that exact text, media, account and time. Any later change to the post needs a new approval. Checked on November 3, 2026, against Flaghand's approval docs, linked under [Sources](#sources). ## The workflow ### 1. The agent drafts the plan and asks for approval Your agent (Claude Code, Cursor, Codex, ChatGPT or your own script) writes the posts, uploads media with alt text and proposes a time for each, all through Flaghand's API or MCP server. Flaghand checks every post against X's and LinkedIn's rules before anything is requested. The agent then requests approval with a note that says what the plan is, such as "Launch thread on X Tuesday 9:30 AM, LinkedIn post Wednesday 10 AM, nothing to do by hand." The owner gets the approval link from the agent, and a push notification on any device where they've turned notifications on. ### 2. Your team or client sees the exact posts Make a **client preview link**: in the desk it's **Client preview › Copy client link** on a draft, or the agent makes one for the whole plan. Anyone with the link sees each post drawn the way X or LinkedIn will show it, under its account and time, without a Flaghand account. The page has no approve, edit or post controls, isn't indexed by search engines, and can be revoked at any time. If a post changes after you shared the link, the page says "Changed …, after this link was shared", so nobody signs off on a version that's gone. Preview links need a card on file. Collect feedback wherever your team already talks, and give it to the agent to apply. ### 3. The owner approves, sends back or denies The owner opens the approval sheet on a phone or computer and sees each post with the agent's note above it. Then: - **Approve** with Face ID, Touch ID or a security key. One tap covers the whole plan. - **Send back** with a one-line note, such as "Lead with the result, not the feature list." The agent reads it, revises and asks again, and the sheet shows what changed since. - **Deny** a post that shouldn't go out, with an optional reason. A denied draft can't be requested again; the agent makes a new one. Only a workspace owner can approve, send back or deny. Agents, API tokens and preview links can't. ### 4. Flaghand posts exactly what was approved Approving freezes each post: text, media, alt text, the account and the time, signed by the owner's passkey. Flaghand posts that snapshot at its time and checks the signature again right before posting. If anyone edits the post afterwards, the approval is void and the owner approves the new version. For X, Flaghand reads the post back and compares it with the approval; LinkedIn doesn't let Flaghand read posts back, so you get the post's link. ## What Flaghand doesn't do Flaghand is built for one owner and their agents, not for multi-stage agency review. It has no reviewer stages, no "everyone must approve" rule and no comments on the preview page: one owner's passkey is the decision. If you need staged client and legal sign-off inside the tool, [Flaghand vs Planable](/docs/guides/flaghand-vs-planable) covers where Planable fits better, and [the social media approval workflow guide](/docs/guides/social-media-approval-workflow) compares Planable, Later, Hootsuite and Agorapulse. ## Questions **Can an AI agent approve its own posts?** No. No Flaghand token has a scope that approves, schedules or posts, and the server refuses an agent that tries. Approval takes the owner's passkey. **Can a client see the post before it's approved?** Yes. A client preview link shows each post exactly as it will appear, with its account and time, read-only and without an account. It notes any change made after the link was shared. **What happens if someone edits a post after approval?** The approval is void, and the post doesn't go out until the owner approves the new version. **Can one approval cover a week of posts?** Yes. The agent requests the whole plan at once, and the owner approves every post in it with one passkey tap. **What if the owner sends a post back?** The agent gets the note, revises the post and requests approval again. The approval sheet shows what changed since it was sent back. ## Get started [Start your free trial](/schedule), or ask your agent to set it up from [set up and connect](https://app.flaghand.com/docs/setup.md). You add a card when you approve your first post ([current terms](/pricing)). To connect an agent, see [Cursor](/docs/guides/schedule-x-and-linkedin-posts-from-cursor), [ChatGPT or Codex](/docs/guides/schedule-linkedin-posts-with-chatgpt-or-codex) and [Claude Code](/docs/guides/connect-linkedin-and-x-to-claude-code). The checks for any tool are in [how to let an AI agent post safely](/docs/guides/let-an-ai-agent-post-safely). ## Sources All read on November 3, 2026. - Flaghand: [approvals](https://app.flaghand.com/docs/approvals), [client preview links](https://app.flaghand.com/docs/approvals#client-preview-links), [notifications](https://app.flaghand.com/docs/notifications), [schedule and results](https://app.flaghand.com/docs/schedule-and-results) LinkedIn and X are trademarks of their owners. Flaghand isn't affiliated with or endorsed by them. [All guides](/docs/guides.md)