Run your socials from Claude.
Connect post mate to Claude Desktop, Claude Code or claude.ai over the Model Context Protocol. Draft, schedule, publish and read analytics across all 13 networks — just by asking. No SDK, no API to wire up.
On it — scheduling to both accounts now.
Done — both are queued for Thursday 09:00, with #newdrop #fyp #spring added. Want me to duplicate it for next Thursday as well?
Three ways to connect
Choose your Claude product and get set up in under a minute.
$ npx -y post-mate-mcp setup
claude mcp add post-mate -- npx -y post-mate-mcp$ claude mcp add post-mate --transport http https://mcp.post-mate.com --header "Authorization: Bearer <your-key>"https://mcp.post-mate.comThe server URL never changes: https://mcp.post-mate.com. Replace pm_••• with a key you mint in Settings → MCP. Keys are shown once and can be rotated or revoked any time.
Three steps. About three minutes.
Mint a key
Open Settings → MCP and generate a key. It's shown once — copy it somewhere safe.
Add the server
Paste the config block for your client (Desktop, Code or claude.ai) and drop your key in place of pm_•••.
Just ask
Restart the client and say what you want. Claude calls the right tool; posts land as drafts for you to approve.
Eight tools. Read and write.
Not a read-only bot. Claude can actually schedule, edit, cancel, fetch metrics and write captions — all under the same workspace auth as your dashboard.
Schedule a post to one or many networks at a given time.
Copy an existing post to another time or set of networks.
List upcoming scheduled posts for the next N days.
Move an existing post to a new date and time.
Cancel a scheduled post before it publishes.
Generate per-platform caption variants for a topic.
Fetch views, likes, comments and saves for a post.
Return every connected social account in the workspace.
Talk to it like a teammate.
Drop any of these into Claude with the post mate connector attached, and it just runs. Click to copy.
Claude can post. Within reason.
Letting a model publish to your audience is only useful if it's safe. Every key respects per-workspace guardrails you control in Settings.
Manage keys & railsApproval before publish
Tool calls land as drafts in the same queue as your team's. Nothing goes live until a human approves. On by default.
Rate limited
Per-key caps on posts per hour and per network — 12/hour out of the box — so an agent can never run away with your feed.
Full audit log
Every tool call is recorded with its prompt, arguments and result for 90 days. Filter by key, agent or account.
Revocable keys
Scoped to one workspace. Rotate or kill any key instantly from Settings — the old one stops working immediately.
Same server, three front doors.
Questions, answered.
Do I need a paid plan?
The MCP server is on the Pro plan and up. Any number of agents and keys are included — keys are capped at 10 per workspace, which you can rotate freely.
Is it safe to let Claude post?
Yes. Approval-before-publish is on by default, so tool calls become drafts you review. Add per-key rate limits and read the 90-day audit log any time. Turn approval off only if you trust the agent.
Which networks are supported?
All 13 that post mate supports — Instagram, TikTok, X, LinkedIn, Threads, YouTube, Pinterest, Bluesky and more. Claude picks accounts by handle or network.
Does my prompt get sent to post mate?
No. Your Claude client talks to Anthropic directly. We only receive the structured tool calls and their arguments — never the raw conversation.
What happens if a key leaks?
Revoke it from Settings → MCP and it stops working instantly. Generate a new one and update your config. Each key is scoped to a single workspace.
Can I use it from my own agent?
Yes — it's the open Model Context Protocol. Anything that speaks MCP (Cursor, Windsurf, a custom agent) connects with the same URL and a Bearer key.
Tell Claude what to post.
We'll handle the rest.
Mint a key, paste one block of config, and your whole content workflow becomes a conversation.