Posthive Agent · MCP + CLI

Let your AI agent
run your socials.

Claude, Cursor, OpenClaw, or any agent that can call a tool or run a shell command can now schedule posts across multi platforms with you approving everything before it goes live.

Set up Posthive MCPUse the API
API live · 99.98% uptimeAGPL-3.0 · open source
Claude Code
G
Schedule a post about our launch for tomorrow at 9am.
Draft created for tomorrow, 9:00 AM. Review before it goes live.
SCHEDULED · JUN 14 · 09:00DRAFT

Shipped v2.0 today — drag-to-reschedule calendar, per-platform overrides, and first-comment on every platform. Try it free →

bsky.appthreads.netlinkedin.commastodon.social4 accountsapprove →
posthive.create_post · 128 tokens
Draft-first · nothing publishes without you
13
Platforms
10
Tools · commands
2
Install paths
100%
Draft-first
Setup

Connect in three steps.

One-click for the hosted apps that ship an MCP UI — Claude, ChatGPT, Cursor, Windsurf, Codex, VS Code. Config-file or shell for everything else.

Hosted apps
ClaudeClaude
Connect Posthive to Claude
Grants read + write access to your Posthive posts, accounts, and templates. Revoke anytime.
  1. 01
    Open Connectors
    Go to claude.ai → sidebar Customize → Connectors → Add → Add custom connector.
  2. 02
    Paste the MCP URL
    Enter the Posthive MCP URL and click Add.
    https://api.posthive.co/mcp
  3. 03
    Sign in with Posthive
    Claude opens a browser window. Sign in and click Approve & connect.
For Claude Desktop, Cursor & Windsurf

One config block,
full MCP access.

Drop posthive-mcp into any MCP client config and your agent instantly has tools to list accounts, create posts, approve drafts, and manage the queue — no server to run or maintain, npx handles it.

  • ✓list_accounts— connected accounts and IDs
  • ✓create_post— draft or schedule directly
  • ✓approve_draft— promote a draft to scheduled
  • ✓update_post / duplicate_post / delete_post
  • ✓list_templates / create_from_template
claude_desktop_config.jsonCOPY
{
  "mcpServers": {
    "posthive": {
      "command": "npx",
      "args": ["posthive-mcp"],
      "env": {
        "POSTHIVE_API_KEY": "ph_yourkey",
        "POSTHIVE_API_URL": "https://api.posthive.co"
      }
    }
  }
}
terminal
$ export POSTHIVE_API_KEY=ph_yourkey$ npx posthive-cli accounts:list$ npx posthive-cli posts:create \    --content "Shipped v2.0 today" \    --accounts acc_bluesky,acc_threads \    --first-comment "Full changelog: example.com"# → saved as a DRAFT - review it in Posthive → Posts$ npx posthive-cli posts:list --status draft
For OpenClaw, Claude Code & shell agents

A CLI that any agent can run.

posthive-cli mirrors the same API as a plain shell command. Every output is structured JSON — perfect for agents that call shell tools rather than speak MCP directly, like OpenClaw or custom pipelines.

It ships with a bundled skill file so agents that support Claude Code-style skills discover the full command set, platform character limits, and best practices automatically.

Install
npx posthive-cli
npm i -g posthive-cli
npx skills add AstaBlackClove/posthive
Self-documenting

Agents learn the tool
by reading it.

No prompt engineering required. The bundled SKILL.md describes every command, required flags, platform quirks (Instagram needs media, X has no links, LinkedIn rewards links in the first comment), and the draft-first workflow — so any capable agent figures out the right call on its own.

Try prompts like: “Schedule a LinkedIn post about our launch for tomorrow 9am with the link in the first comment” — the agent handles the rest.

SKILL.md
---
name: posthive
description: Schedule social posts across multiple platforms
---

# Workflow
1. accounts:list — get valid account IDs
2. posts:create — saved as DRAFT by default
3. posts:approve — promote a draft to scheduled

Ships inside posthive-cli. Works with Claude Code,
OpenClaw, and any agent that runs shell commands.
Safety by default

Nothing goes live
without your approval.

Drafts, not publishes

Every post an agent creates lands as a draft in your review queue. Nothing goes live until you or the agent explicitly approves it.

Explicit scheduling only

An agent can only skip the draft step by passing a future timestamp and being told to schedule directly — never by default.

Scoped API keys

Agents authenticate with a revocable ph_ API key, never your login. Rotate or revoke access from Settings anytime.

FAQ

Frequently asked questions

What is posthive-mcp?+

posthive-mcp is an MCP (Model Context Protocol) server you run with npx. It gives Claude Desktop, Cursor, Windsurf, and any MCP-compatible client tools to list accounts, create posts, and manage your Posthive queue.

What is posthive-cli and how is it different from the MCP server?+

posthive-cli is a shell command (`posthive`) that mirrors the same API. It's built for agents that run shell commands rather than speak MCP directly — OpenClaw, Claude Code skills, and custom automation pipelines. Every command outputs structured JSON.

Will an AI agent publish posts without my approval?+

No. Every post created by an agent is saved as a draft by default and appears in Posthive's review queue. Nothing publishes unless a human approves it, or the agent is explicitly told to schedule directly with a future timestamp.

Do I need a paid plan to use the agent tools?+

The MCP server requires a Pro or Team plan. The CLI works with any valid API key, which is also available on Pro and Team.

Wire up your agent
in minutes.

Generate an API key, drop it into your MCP config or export it for the CLI, and your agent is ready to draft posts across multi platforms.

Read the docsStart free trial