Server data from the Official MCP Registry
Facebook & Instagram data for AI agents: pages, posts, groups, ads, Marketplace, profiles, reels.
About
Facebook & Instagram data for AI agents: pages, posts, groups, ads, Marketplace, profiles, reels.
Remote endpoints: streamable-http: https://mcp.socialapis.io/mcp
Security Report
Valid MCP server (2 strong, 4 medium validity signals). 3 known CVEs in dependencies Package registry verified. Imported from the Official MCP Registry.
12 tools verified · Open access · 4 issues found
Security scores are indicators to help you make informed decisions, not guarantees. Always review permissions before connecting any MCP server.
Permissions Required
This plugin requests these system permissions. Most are normal for its category.
What You'll Need
Set these up before or after installing:
Environment variable: SOCIALAPIS_API_KEY
Environment variable: MCP_PROXY_URL
How to Install & Connect
Available as Local & Remote
This plugin can run on your machine or connect to a hosted endpoint. during install.
Documentation
View on GitHubFrom the project's GitHub README.
SocialAPIs MCP Server

Facebook & Instagram data for AI agents — local (npx) or hosted (remote MCP)
Website • Documentation • Discord • npm
Official SDKs: Python • JavaScript / TypeScript • Go
🚀 Quick Start
AI agents (Cline, Claude Code, Cursor): follow
llms-install.mdfor step-by-step install and verification.
One-click: use the Install in Cursor button above — then replace YOUR_API_KEY in Cursor's MCP settings with your key. Also listed on Smithery, cursor.directory, mcp.so and the official MCP Registry (io.github.SocialAPIsHub/social-media-api).
Option A — Hosted (remote MCP, no install)
Point any client that supports remote MCP servers (Streamable HTTP) at:
https://mcp.socialapis.io/mcp
Authenticate with your SocialAPIs API key in a header:
Authorization: Bearer YOUR_API_KEY
(x-api-token: YOUR_API_KEY also works.) Tool listing works without a key; tool calls need one. Example config for clients that take a URL + headers:
{
"mcpServers": {
"socialapis": {
"url": "https://mcp.socialapis.io/mcp",
"headers": { "Authorization": "Bearer YOUR_API_KEY" }
}
}
}
Option B — Local (npx, stdio)
Installation
npm install -g @socialapis/mcp
Configuration
Add to your Claude Desktop config:
macOS:
nano ~/Library/Application\ Support/Claude/claude_desktop_config.json
Windows:
notepad %APPDATA%\Claude\claude_desktop_config.json
🔧 Configuration
Method 1: Command Line Argument (Recommended for Claude Desktop)
{
"mcpServers": {
"socialapis": {
"command": "npx",
"args": ["-y", "@socialapis/mcp", "YOUR_API_KEY"]
}
}
}
Method 2: Environment Variable
# Set environment variable
export SOCIALAPIS_API_KEY=your_api_key_here
# Run without argument
npx @socialapis/mcp
Method 3: .env File (For Development)
# Copy example file
cp .env.example .env
# Edit with your values
nano .env
.env file:
SOCIALAPIS_API_KEY=your_api_key_here
MCP_PROXY_URL=https://mcp.socialapis.io
Environment Variables
| Variable | Description | Default |
|---|---|---|
SOCIALAPIS_API_KEY | Your SocialAPIs API key | None (required) |
MCP_PROXY_URL | MCP proxy server URL | https://mcp.socialapis.io |
PORT | HTTP server port | 3001 |
API_BASE_URL | Backend API URL | https://api.socialapis.io |
Get API Key
- Sign up at socialapis.io
- Go to Dashboard
- Copy your API key
- Replace
YOUR_API_KEYin config
Test It
Restart Claude Desktop and ask:
Get Nike's Facebook page details
📋 Features
- 🌐 One API key - Facebook and Instagram behind a single interface
- 🤖 AI-First - Built for Claude, Cursor, and AI agents
- 📊 Rich Data - Posts, comments, engagement metrics
- 🔍 Advanced Filtering - Time ranges, pagination
- 🎯 Simple Auth - No OAuth complexity
- 🧩 Structured output - every tool declares an output schema and returns
{ result, creditsCharged, creditsRemaining }, so agents can track credit spend directly - 🛡️ Read-only annotations - all tools are marked
readOnlyHint, so clients can skip confirmation prompts - ⚡ Fast - Global edge network
- 🔒 Secure - API keys stay local
🛠️ Available Tools
12 tools covering 47 actions across Facebook and Instagram. Each tool groups one resource and takes an action argument; every action maps to one REST endpoint on api.socialapis.io. Tool descriptions list each action, the parameters it needs and its credit cost.
| Tool | Actions |
|---|---|
facebook_page | details (followers, contact info, category; exact_followers_count=true costs 5 credits), id, posts (limit 3-9, date filters), videos (limit 6-12), reels |
facebook_group | details, id, posts (limit 3-9, date filters), videos |
facebook_post | details, details_extended (view counts, video URLs, audio metadata), id, attachments (5 credits), video, comments, comment_replies |
facebook_search | pages, people, posts, videos, locations (location UIDs for the other searches) |
facebook_ads | search (ads by keyword, page, country, status), keywords (advertisers for a keyword), page_details, archive_details, countries |
facebook_marketplace | search (price, location, category, condition filters), vehicles, rentals, listing, seller, categories, city_coordinates |
facebook_media_download | Download URLs for Facebook images and videos |
instagram_profile | details, id, posts, reels, highlights, highlight_details |
instagram_post | details, id (shortcode from a post URL) |
instagram_reels | feed, by_audio |
instagram_search | Popular users, hashtags and places for a keyword |
instagram_location | posts (recent or top), nearby |
Example call: facebook_page with {"action": "posts", "link": "https://www.facebook.com/nike", "limit": 9}.
Upgrading from 1.x: the 47 earlier tool names (facebook_get_page_details, instagram_get_profile_posts, …) still work when called directly, so existing prompts and scripts keep running. They're no longer listed. If your client auto-approves tools by name, add the new names.
Not supported yet
Only Facebook and Instagram are live today. TikTok, X / Twitter, LinkedIn and YouTube are on the roadmap with no committed dates — tools for them will appear here only once they ship.
Track the platform roadmap at socialapis.io/api-sources.
💡 Usage examples
Each prompt below is a real Claude Desktop session. Some of these are single-tool-call patterns ("get me X"); some require Claude to chain multiple calls + aggregate the results (noted where).
Single-call patterns (fast, cheap)
What's Nike's follower count on Facebook?
→ Uses facebook_page action=details (1 credit)
Get the latest 9 posts from facebook.com/EngenSA
→ Uses facebook_page action=posts with limit=9 (1-3 credits depending on actual returned count)
Show me the Meta ads currently running for "Apple Vision Pro" in Germany
→ Uses facebook_ads action=search (1 credit)
Multi-call patterns (Claude orchestrates these — but it's slower + more expensive)
Compare engagement on Nike vs Adidas's last 9 Facebook posts
→ Claude calls facebook_page action=posts twice (~2-6 credits total),
aggregates reactions/comments/shares per post, returns a comparison.
What are people saying in the comments on Coca-Cola's last 3 posts?
→ Claude calls facebook_page action=posts (1 credit) then
facebook_post action=comments 3 times (3 credits) and summarizes.
Show me marketplace listings for "PlayStation 5" under $400 in Berlin
→ Claude calls facebook_marketplace action=city_coordinates (1 credit) +
facebook_marketplace action=search with filters (1 credit).
What this MCP server does NOT do
Some queries look natural in a chat ("compare engagement over the last month") but require aggregations the API doesn't expose as a single tool yet. Claude can still answer them, but it'll fan out into many tool calls — which is slow + expensive.
| Query shape | Why it's hard |
|---|---|
| "Engagement rate over the last 30 days" for a page | Requires fetching every post in the date range (paginated, limit capped at 9 per call) and computing engagement per post. Hits the LLM tool-call budget on busy pages. |
| "Compare engagement rates between Brand A, B, C over the last month" | Same problem, 3× — one paginated fetch per brand, then comparison math. Works for small windows; slow for "last month" on high-volume pages. |
| Historical archive older than what Facebook itself serves | We surface what Facebook makes publicly visible. Posts that scrolled off Facebook's visible feed aren't retrievable. |
| Server-side time-series (daily engagement, weekly growth) | Not yet — on the roadmap as a future engagement-stats endpoint with built-in aggregation. |
If your use case maps to one of these patterns and you want the aggregation pre-computed instead of LLM-orchestrated, contact support with the specific query — we're prioritizing the aggregation endpoint based on customer demand.
🏗️ Architecture
Claude Desktop
↓
@socialapis/mcp (local MCP client)
↓
https://mcp.socialapis.io (global proxy)
↓
https://api.socialapis.io (data API)
Why this architecture?
- ✅ Low latency (global edge network)
- ✅ High reliability (99.9% uptime)
- ✅ Auto rate limiting
- ✅ Smart caching
- ✅ Your API key stays local
🔧 Development
Local Setup
# Clone repository
git clone https://github.com/SocialAPIsHub/mcp-server.git
cd mcp-server
# Install dependencies
npm install
# Run MCP client
npm start YOUR_API_KEY
# Run HTTP proxy server
npm run serve
Project Structure
mcp-server/
├── src/
│ ├── tools.js # The 47 endpoint definitions (one per action)
│ └── groups.js # Groups them into the 12 listed tools
├── mcp-wrapper.js # MCP client (runs locally)
├── server.js # HTTP proxy server
├── package.json
├── Dockerfile
└── README.md
Testing
# Test MCP client locally
node mcp-wrapper.js YOUR_API_KEY
# Test HTTP proxy
curl http://localhost:3001/health
curl http://localhost:3001/tools
# Test specific tool
curl -X POST http://localhost:3001/proxy \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"tool":"facebook_page","arguments":{"action":"details","link":"https://facebook.com/nike"}}'
📊 Pricing
| Plan | Credits / month | Price |
|---|---|---|
| Free | 200 | $0 |
| Pro | 1,500 | $4.99 |
| Ultra | 30,000 | $49 |
| Mega | 120,000 | $179 |
| Enterprise | Custom | Contact us |
Most tools cost 1 credit per call; each tool description states its exact price. Current plans: socialapis.io/pricing.
🤝 Contributing
We welcome contributions! Please see CONTRIBUTING.md for details.
Quick Contribution Guide
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
📖 Documentation
- Full Documentation
- API Reference
- MCP Guide
- Examples
- Agent skill: teaches Claude Code, Cursor, Codex and other coding agents to use SocialAPIs (
npx skills add SocialAPIsHub/skills)
💬 Support
- 📧 Email: support@socialapis.io
- 💬 Discord: discord.gg/D5bQskrwV
- 🐛 Issues: GitHub Issues
- 📚 Docs: docs.socialapis.io
🗺️ Roadmap
Shipped:
- Facebook API support — 7 tools, 31 actions (Pages, Groups, Posts, Search, Ads Library, Marketplace, Media)
- Instagram support — 5 tools, 16 actions (Profiles, Posts, Reels, Highlights, Discovery / Locations)
- MCP server implementation
- HTTP proxy server
- npm package published —
@socialapis/mcp - MCP Registry listing — registry.modelcontextprotocol.io
- Python SDK —
socialapis-sdkon PyPI (50 endpoints, MIT) - JavaScript / TypeScript SDK —
socialapis-sdkon npm (Node 18+, Bun, Deno, browsers) - Go SDK — github.com/SocialAPIsHub/socialapis-go (idiomatic, zero deps)
- Agent skill — github.com/SocialAPIsHub/skills for Claude Code, Cursor, Codex and other coding agents
Upcoming:
- TikTok support
- X (Twitter) support
- LinkedIn support
- YouTube support
- Advanced analytics — server-side aggregation endpoints (engagement-over-time, brand comparisons) so multi-call patterns become a single tool call
- Real-time webhooks — push notifications on new posts / engagement thresholds
- LangChain integration
Platform priorities shift based on customer demand. The fastest way to push something up the queue is to email support@socialapis.io or DM @socialapis on Telegram with the use case.
📄 License
This project is licensed under the MIT License - see the LICENSE file for details.
🙏 Acknowledgments
- Built with Model Context Protocol
- Powered by Anthropic Claude
- Inspired by the AI agent community
📇 Listed on
Also on the Official MCP Registry, Smithery, LobeHub, mcp.so, mcpservers.org, cursor.directory and awesome-remote-mcp-servers.
🌟 Star History
Made with ❤️ by the SocialAPIs Team
Python SDK • JS SDK • Go SDK
Reviews
No reviews yet
Be the first to review this server!
More Content & Media MCP Servers
Toleno
Freeby Toleno · Developer Tools
Toleno Network MCP Server — Manage your Toleno mining account with Claude AI using natural language.
mcp-creator-python
Freeby mcp-marketplace · Developer Tools
Create, build, and publish Python MCP servers to PyPI — conversationally.
MCP Marketplace
Freeby mcp-marketplace · Developer Tools
Search and install MCP servers from inside your AI client.
FinAgent
Freeby mcp-marketplace · Finance
Free stock data and market news for any MCP-compatible AI assistant.
by Lharries · Communication
Read, search, and send WhatsApp messages through your AI assistant
Memory
Freeby Modelcontextprotocol · Productivity
Knowledge graph-based persistent memory across sessions
