@posteverywhere/mcp — Social Media MCP Server for Claude, ChatGPT, Cursor, Codex & AI Agents

👁 npm version
👁 npm downloads
👁 License: MIT
👁 GitHub stars
👁 Glama quality score

Official Model Context Protocol server for PostEverywhere — let Claude Code, Claude Desktop, ChatGPT (via the hosted connector), Cursor, OpenAI Codex, and other MCP-compatible AI clients schedule and publish social media posts to Instagram, TikTok, YouTube, LinkedIn, Facebook, X (Twitter), Threads, and Pinterest using natural language.

💡 Building a programmatic integration? Use the companion @posteverywhere/sdk Node.js SDK instead — full TypeScript types, retry handling, error classes.

🔗 Quick Links

Related MCP server: Sociona MCP Server

What is MCP?

Model Context Protocol (MCP) is an open standard from Anthropic that lets AI assistants connect to external tools and data sources. Once you connect this MCP server, your AI assistant can:

• Schedule posts across all your connected social accounts

• Generate AI images and attach them to posts

• List, edit, retry, and delete scheduled or published posts

• Surface per-platform publishing results including failure reasons

…all from natural-language prompts inside Claude, Cursor, or any MCP-compatible client.

Quick Start

1. Get an API key

Sign up free at posteverywhere.ai/signup (7-day trial), connect your social accounts, then create an API key at Settings → Developers. Choose your scopes:

• Read — list accounts, posts, and media

• Write — create, edit, delete posts and media

• AI — generate images using AI models

2. Add the MCP server to your AI client

Claude Code

claude mcp add posteverywhere -- npx -y @posteverywhere/mcp

Then set POSTEVERYWHERE_API_KEY in your environment, or pass it as an argument:

claude mcp add posteverywhere -e POSTEVERYWHERE_API_KEY=pe_live_your_key_here -- npx -y @posteverywhere/mcp

📖 Claude Code MCP docs →

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
 "mcpServers": {
 "posteverywhere": {
 "command": "npx",
 "args": ["-y", "@posteverywhere/mcp"],
 "env": {
 "POSTEVERYWHERE_API_KEY": "pe_live_your_key_here"
 }
 }
 }
}

Restart Claude Desktop. The PostEverywhere tools will appear in the 🔨 menu.

📖 Claude Desktop MCP docs →

Cursor

In Cursor, open Settings → MCP → Add new MCP server and paste:

{
 "posteverywhere": {
 "command": "npx",
 "args": ["-y", "@posteverywhere/mcp"],
 "env": {
 "POSTEVERYWHERE_API_KEY": "pe_live_your_key_here"
 }
 }
}

📖 Cursor MCP docs →

Other MCP clients

This package exposes a standard stdio MCP server. Any MCP-compatible client can connect using the same configuration shape.

3. Try it out

Once connected, ask your AI assistant things like:

"List my connected social accounts"

"Schedule a post to all my accounts for tomorrow at 2pm: Just shipped a new feature 🚀"

"Show me my recent posts and their publishing status"

"Generate an image of a sunset over mountains in 16:9 and post it to Instagram and X"

"Retry the failed destinations on my latest post"

"What are my last 10 posts on TikTok?"

Available Tools

Accounts

• list_accounts — list all connected social accounts and their health status

• get_account — get details for a specific account (token expiry, can-post status)

Posts

• list_posts — list posts filtered by status (scheduled, published, failed, draft) or platform. Drafts include their target accounts + per-platform content so you can review them before publishing.

• get_post — get full details for a single post including all destinations (for drafts, the saved accounts + per-platform content)

• create_post — publish now, schedule for later, or save as a draft for human review (draft: true)

• schedule_post — publish or schedule a draft (the approval step — pass scheduled_for, or publish_now: true)

• update_post — modify a scheduled or draft post (content, schedule time, accounts)

• delete_post — delete a scheduled or draft post

• get_post_results — per-platform publishing results, errors, and live URLs

• retry_failed_post — retry every failed destination of a post

🧑‍💼 Human-in-the-loop drafts — let the agent draft posts for you to approve before anything goes live:

1. "Draft a LinkedIn post about our launch — don't publish it yet" → create_post(draft: true)

2. "Show me my drafts" → list_posts(status: "draft") / get_post

3. "Looks good — schedule it for Tuesday 9am" → schedule_post(scheduled_for: …) (or "publish it now" → publish_now: true)

Drafts are saved in your PostEverywhere account and never publish until you schedule them — review them in the app or via the agent.

Media

• list_media — list files in your media library

• get_media — get media file details and processing status

• upload_media_from_url — import an image from any public URL into the library (one call, returns a media_id ready for create_post)

• delete_media — remove a media file

AI

• generate_image — generate an image from a text prompt (4 models, 7 aspect ratios)

📖 Full tool reference and parameters →

Example Prompts

These all work out of the box once the server is connected:

Scheduling

• "Schedule a Christmas post for December 25th at 9am: Merry Christmas to our amazing community 🎄"

• "Post this to Instagram and TikTok only: Behind-the-scenes of our latest feature drop"

• "Queue up a week of daily morning posts starting Monday at 9am"

Content generation

• "Generate a 9:16 image of a coffee shop at sunrise and schedule it as a TikTok post for tomorrow"

• "Create a square image of abstract green shapes for a LinkedIn carousel"

Monitoring

• "What posts failed in the last week and why?"

• "Show me the publishing results for my last Instagram Reel"

• "Which of my accounts need to be reconnected?"

Recovery

• "Retry the failed LinkedIn destination on post abc-123"

• "Delete all my draft posts older than 30 days"

Each natural-language prompt maps to one or more MCP tool calls — the agent figures out the right sequence.

Supported Platforms

All eight platforms work on every plan:

• Instagram — feed, reels, stories, carousels

• TikTok — videos, photo carousels

• YouTube — videos with thumbnails, descriptions, tags

• LinkedIn — text, images, video, document carousels

• Facebook — pages, video, reels, multi-image

• X (Twitter) — text, threads, media (tier-aware char limits)

• Threads — text and media posts

• Pinterest — pins to boards

Configuration

Rate Limits

Hitting a 429? The MCP server returns a clear rate_limit_exceeded code with retry_after seconds — the agent will know to wait and retry.

📖 Rate limit details →

How it compares

Documentation

• 📖 Full API Reference — every tool, every parameter

• 🔐 Authentication & Scopes

• ⚠️ Error Handling — including the retryable flag agents should respect

• ⏱️ Rate Limits — per-minute, per-hour, per-day caps

• 🖼️ Media Requirements — file size, format, aspect ratio per platform

• 🤖 Agent System Prompts — recommended prompts for AI agent integrations

• 🧠 About MCP — the open protocol behind this server

• 🚀 Quick Start Guide — first post in 60 seconds

• 🔗 Webhooks — receive publish events on your endpoints

• 🏷️ API Scopes — fine-grained permission control

• 📋 Changelog — what's new in the API

MCP & AI Tooling Resources

• 🧠 Model Context Protocol — Specification

• 🛠️ MCP Quickstart for Users

• 👨‍💻 MCP Quickstart for Developers

• 🗂️ Browse all MCP servers (community list)

• 💼 Anthropic — MCP announcement

• 🔧 Claude Code — Get started

• 🔧 Claude Code — MCP setup

• 🖥️ Claude Desktop

• 🖱️ Cursor — MCP docs

• 💬 Anthropic Claude

PostEverywhere Around the Web

• 🌐 PostEverywhere Homepage

• 🛠️ Developers Landing Page — overview of API, SDK, MCP, integrations

• 🤖 AI Agents Page — using PostEverywhere with Claude, ChatGPT, Cursor, and more

• 💵 Pricing — Starter ($19), Growth ($39), Pro ($79); 7-day free trial

• ✍️ Blog — guides, tutorials, product updates

• 📚 Help Center — guides, troubleshooting, FAQs

• 🎛️ Dashboard (sign in)

• ✨ Sign Up — 7-day free trial

• 🔑 Get an API key

• 📦 Node.js SDK (npm)

• 💻 Node.js SDK (GitHub)

• 📦 MCP Server (npm — this package)

• 💻 MCP Server (GitHub — this repo)

• 🏢 PostEverywhere on GitHub

Per-Platform Schedulers

PostEverywhere is the backend for all of these — every plan includes every platform:

• 📷 Instagram Scheduler — feed posts, reels, stories, carousels

• 🎵 TikTok Scheduler — videos, photo carousels, full content disclosure flags

• 📺 YouTube Scheduler — videos with thumbnails, tags, descriptions, privacy controls

• 💼 LinkedIn Scheduler — personal + Company Page posts, document carousels, video

• 👍 Facebook Scheduler — Pages, Reels, video, multi-image carousels

• 🐦 X (Twitter), Threads, Pinterest — all included on every plan

Related

• 📦 @posteverywhere/sdk — Node.js / TypeScript SDK (npm)

• 🌐 posteverywhere.ai — Web dashboard, AI Studio, calendar UI

• 📚 Help Center — Guides, troubleshooting, FAQ

• 💵 Pricing — From $19/mo, 7-day free trial

Support

• 📧 Email: support@posteverywhere.ai

• 🐛 Issues: github.com/posteverywhere/mcp/issues

• 💬 Help Center

License

MIT — see LICENSE.

Built by the team at PostEverywhere. The smarter way to schedule social media to Instagram, TikTok, YouTube, LinkedIn, Facebook, X, Threads, and Pinterest — now with native MCP support for AI agents.

Maintenance

Resources

• NPM Package
• GitHub Repository
Need Help?Related Servers

Tools

• bulk_create_postsA
• create_campaignB
• create_postA
• create_webhookA
• delete_campaignA
• delete_mediaA
• delete_postA
• delete_webhookB