Social Fetch MCP
Lets coding agents fetch real social media and web data from platforms like TikTok, Instagram, YouTube, and more, directly inside editors like Cursor and VS Code.
README
<p align="center"> <img src="https://api.socialfetch.dev/mcp/icon.svg" alt="Social Fetch MCP" width="96" height="96" /> </p>
<h1 align="center">Social Fetch MCP</h1>
<p align="center"> Let your coding agent fetch real social media and web data while it builds. </p>
<p align="center"> <a href="https://www.socialfetch.dev/docs/integrations/mcp">Documentation</a> | <a href="https://www.socialfetch.dev/docs/api">API reference</a> | <a href="https://www.socialfetch.dev/docs/credits">Credits & billing</a> </p>
<p align="center"> <strong>Server URL:</strong> <code>https://api.socialfetch.dev/mcp</code> </p>
What Is This?
Social Fetch MCP connects Cursor, VS Code, Claude, Windsurf, and other MCP clients to the Social Fetch API. Once installed, your agent can look up creators, fetch posts, search platforms, read web pages, check your balance, and read the Social Fetch docs from inside your editor.
Use it when you want your agent to answer questions like:
- "Find this TikTok creator and add their latest videos to my app."
- "Pull the transcript for this YouTube video and summarize the sponsorships."
- "Fetch this Instagram post, inspect the response shape, and wire it into my backend."
- "Read the Social Fetch docs and choose the right endpoint for Reddit search."
The MCP server itself is hosted by Social Fetch:
https://api.socialfetch.dev/mcp
There is no local server to clone, build, or run from this repository. Clients connect to the hosted Streamable HTTP endpoint and authenticate with OAuth in the browser.
Why Use It?
Agents are better when they can work with the same API your app uses. This gives them a direct, authenticated route to Social Fetch instead of making you bounce between docs, dashboards, scripts, and pasted JSON.
- Less setup: call Social Fetch tools directly from your agent.
- Better answers: the agent can search and read the live docs before choosing an endpoint.
- Real data in the loop: test with TikTok, Instagram, YouTube, X/Twitter, Reddit, Spotify, web, and more from inside your editor.
- No API keys in editor config: sign in once with OAuth.
- Same API everywhere: MCP tools map to the public REST API, SDK, and docs.
What You Can Fetch
Social Fetch MCP exposes endpoint tools for:
- TikTok, Instagram, YouTube, X/Twitter, Telegram, Facebook, Threads, LinkedIn, Reddit, and Spotify data.
- Profiles, posts, videos, comments, transcripts, search results, and platform-specific metadata.
- Web markdown generation and AI-assisted web page questions.
- Account identity and credit balance.
- Social Fetch docs through
docs_searchanddocs_read.
Endpoint tools are metered like the REST API. Docs tools, auth_whoami, and billing_balance_get do not consume lookup credits.
Requirements
- A Social Fetch account.
- Credits for metered API routes.
- An MCP client with remote Streamable HTTP support and OAuth, such as Cursor, VS Code, Claude Code, Claude Desktop, or MCP Inspector.
One-Click Install
- <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=socialfetch&config=eyJ1cmwiOiJodHRwczovL2FwaS5zb2NpYWxmZXRjaC5kZXYvbWNwIn0%3D"><img src="https://www.socialfetch.dev/integrations/mcp/cursor.svg" width="20" height="20" alt="" valign="middle" /> Install in Cursor</a>
- <a href="vscode:mcp/install?%7B%22name%22%3A%22socialfetch%22%2C%22url%22%3A%22https%3A%2F%2Fapi.socialfetch.dev%2Fmcp%22%2C%22type%22%3A%22http%22%7D"><img src="https://www.socialfetch.dev/integrations/mcp/vscode.svg" width="20" height="20" alt="" valign="middle" /> Install in VS Code</a>
- <a href="cursor://anysphere.cursor-deeplink/mcp/install?name=socialfetch&config=eyJ1cmwiOiJodHRwczovL2FwaS5zb2NpYWxmZXRjaC5kZXYvbWNwIn0%3D" title="Windsurf uses the same MCP install flow as Cursor"><img src="https://www.socialfetch.dev/integrations/mcp/windsurf.svg" width="20" height="20" alt="" valign="middle" /> Install in Windsurf</a>
- Open the hosted install page
Complete the OAuth browser flow after installing. If a one-click link does nothing - your editor is not installed, blocks custom URL schemes, or has no handler - use the manual setup below.
Manual Setup
Cursor
Add this to ~/.cursor/mcp.json or a project .cursor/mcp.json:
{
"mcpServers": {
"socialfetch": {
"url": "https://api.socialfetch.dev/mcp"
}
}
}
Restart Cursor, open MCP settings, and complete the OAuth browser flow when prompted.
VS Code
Add this to settings.json:
{
"mcp": {
"servers": {
"socialfetch": {
"url": "https://api.socialfetch.dev/mcp"
}
}
}
}
Claude Code
claude mcp add --transport http socialfetch https://api.socialfetch.dev/mcp
Then run /mcp and authenticate Social Fetch to finish the OAuth flow.
Claude Desktop
Add a custom remote MCP connector with this URL:
https://api.socialfetch.dev/mcp
Authorize Social Fetch when Claude prompts you.
stdio-only Clients
If your client only supports stdio MCP servers, bridge to the hosted server with mcp-remote:
npx mcp-remote https://api.socialfetch.dev/mcp
Point your client at the stdio process that command starts.
Authentication
Social Fetch MCP uses OAuth 2.1. Your MCP client handles the flow:
- The client opens Social Fetch sign-in in your browser.
- You sign in or create an account and approve access.
- The client receives an access token and sends it with each MCP request.
Do not paste a Social Fetch API key into MCP config. Credits are charged to the signed-in account, just like REST API requests.
Tools
tools/list exposes endpoint tools plus docs tools.
Endpoint tool names match API operation IDs with dots replaced by underscores:
| Tool name | API operation |
|---|---|
tiktok_profile_get |
tiktok.profile.get |
instagram_post_get |
instagram.post.get |
web_markdown_generate |
web.markdown.generate |
auth_whoami |
auth.whoami |
billing_balance_get |
billing.balance.get |
Each endpoint tool's arguments mirror the matching API route parameters. See the API reference for the full endpoint catalog, argument schemas, response shapes, and pricing.
Two docs tools are also available:
| Tool name | Purpose |
|---|---|
docs_search |
Search Social Fetch docs across API, SDK, auth, errors, credits, and integrations. |
docs_read |
Read a full docs page as markdown by pathname, such as /docs/sdk. |
Recommended agent prompt:
Use the Social Fetch MCP docs to add TikTok profile lookup to this app with the TypeScript SDK.
Example: TikTok Profile
In Cursor, MCP Inspector, or another MCP client, call tiktok_profile_get with:
{ "handle": "nike" }
The result contains the same data and meta envelope returned by the REST API. Check meta.creditsCharged and meta.requestId after metered calls.
Troubleshooting
| Symptom | What to check |
|---|---|
401 on every call |
Reconnect the server and complete OAuth again. Your token may be missing or expired. |
| OAuth redirect loop | Finish signing in at Social Fetch, then approve access to return to your client. |
402 responses |
Your account is out of credits for the metered route. Top up in Credits & billing. |
| No tools listed | Reconnect the server. If it persists, restart your MCP client. |
| Client only supports stdio | Use npx mcp-remote https://api.socialfetch.dev/mcp. |
Repository Contents
server.json- MCP Registry metadata for the hosted remote server.examples/- Copy-paste MCP client configuration snippets.assets/socialfetch-mark.svg- Local copy of the Social Fetch mark (also served athttps://api.socialfetch.dev/mcp/icon.svg).
The implementation lives in Social Fetch's private API service, not in this public listing repository.
Related
License
MIT (c) Social Freak Ltd
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。