Xpoz MCP Server
Enables AI agents to search Twitter, Instagram, Reddit, and TikTok using natural language queries, with CSV exports up to 500K rows and no API keys needed.
README
Xpoz MCP Server
Search Twitter, Instagram, Reddit & TikTok from any AI agent. Billions of posts indexed. Natural language queries. CSV exports up to 500K rows. No API keys needed.
Remote MCP server — no local installation required. Connect via Streamable HTTP and authenticate with OAuth 2.1 (sign in with Google, no API keys to manage). Direct probes to
https://mcp.xpoz.ai/mcpreturn401 Unauthorized— that's expected; your MCP client handles the OAuth flow.
Quick Start
Claude Desktop / Cursor / Windsurf
Add to your MCP config:
{
"mcpServers": {
"xpoz": {
"type": "streamable-http",
"url": "https://mcp.xpoz.ai/mcp"
}
}
}
Claude Code, Codex, Gemini CLI
# Claude Code
claude mcp add --transport http --scope user xpoz https://mcp.xpoz.ai/mcp
# Codex / Gemini CLI: see https://www.xpoz.ai/integrations
OpenClaw
clawhub install xpoz-social-search
Then authenticate via the OAuth link when prompted. Eight more pre-built skills are available — see clawhub.ai/u/atyachin.
Try it without signing up
Want to kick the tires first? Mint a free trial token — no sign-in required, valid for 5 days. It returns a limited set of results so you can see the server in action:
# Get a trial token (starts with "TRIAL")
curl -X POST https://api.xpoz.ai/api/trial/token
# -> { "success": true, "data": { "accessKey": "TRIAL...", "expiresInSeconds": 432000 }, ... }
Skip the OAuth flow by passing the trial token as a bearer header in your MCP config:
{
"mcpServers": {
"xpoz": {
"type": "streamable-http",
"url": "https://mcp.xpoz.ai/mcp",
"headers": { "Authorization": "Bearer TRIAL..." }
}
}
}
For full results, grab your own API key or just connect via OAuth as shown above.
SDKs & CLI
If you'd rather call Xpoz directly from code or a terminal instead of through an MCP client:
TypeScript / Node.js
npm install @xpoz/xpoz
Source: XPOZpublic/xpoz-ts-sdk.
Python
pip install xpoz
Source: XPOZpublic/xpoz-python-sdk.
CLI
| Platform | Install command |
|---|---|
| macOS / Linux (Homebrew) | brew install XPOZpublic/xpoz/xpoz-cli |
| macOS / Linux (one-liner) | curl -fsSL https://raw.githubusercontent.com/XPOZpublic/xpoz-cli/main/install.sh | sh |
| Windows (PowerShell) | iwr -useb https://raw.githubusercontent.com/XPOZpublic/xpoz-cli/main/install.ps1 | iex |
| Windows (winget) | winget install Xpoz.XpozCli |
| Any platform with Python ≥ 3.10 | pip install xpoz-cli |
Source: XPOZpublic/xpoz-cli.
All four channels speak the same MCP protocol against https://mcp.xpoz.ai/mcp under the hood, so any tool you see below is reachable from every surface. Get an API token at xpoz.ai/get-token.
Platforms & Tools
Twitter / X (14 tools)
| Tool | Description |
|---|---|
searchTwitterUsers |
Find users by name, bio, or keywords |
getTwitterUser |
Get profile details (followers, bio, metrics) |
getTwitterUsersByKeywords |
Discover users actively posting about topics |
getTwitterUserConnections |
Get followers or following list |
getTwitterPostsByKeywords |
Search tweets by keywords, hashtags, mentions |
getTwitterPostsByAuthor |
Get tweets from a specific user |
getTwitterPostsByIds |
Fetch specific tweets by ID |
getTwitterPostComments |
Get replies to a tweet |
getTwitterPostRetweets |
Get retweets of a tweet |
getTwitterPostQuotes |
Get quote tweets |
getTwitterPostInteractingUsers |
Users who liked/retweeted a post |
countTweets |
Count tweets matching a query (hourly/daily buckets) |
checkOperationStatus |
Poll async export operations |
cancelOperation |
Cancel a running operation |
Instagram (9 tools)
| Tool | Description |
|---|---|
searchInstagramUsers |
Find users by name or username |
getInstagramUser |
Get profile details and metrics |
getInstagramUsersByKeywords |
Discover users posting about topics |
getInstagramUserConnections |
Get followers or following list |
getInstagramPostsByKeywords |
Search posts by keywords or hashtags |
getInstagramPostsByUser |
Get posts from a specific user |
getInstagramPostsByIds |
Fetch specific posts by ID |
getInstagramPostInteractingUsers |
Users who liked/commented on a post |
getInstagramCommentsByPostId |
Get comments on a post |
Reddit (9 tools)
| Tool | Description |
|---|---|
searchRedditUsers |
Find users by name |
getRedditUser |
Get user profile and karma |
getRedditUsersByKeywords |
Discover users active in topics |
getRedditPostsByKeywords |
Search posts across subreddits |
getRedditPostWithCommentsById |
Get a post with its comment tree |
getRedditCommentsByKeywords |
Search comments by keywords |
searchRedditSubreddits |
Find subreddits by name |
getRedditSubredditWithPostsByName |
Get a subreddit's metadata and recent posts |
getRedditSubredditsByKeywords |
Discover subreddits by topic |
TikTok (7 tools)
| Tool | Description |
|---|---|
searchTiktokUsers |
Find creators by name or username |
getTiktokUser |
Get creator profile (followers, likes, bio, verification) |
getTiktokUsersByKeywords |
Discover creators by topic in their post history |
getTiktokPostsByKeywords |
Search videos by keywords or hashtags |
getTiktokPostsByUser |
Get videos posted by a specific creator |
getTiktokPostsByIds |
Fetch specific videos by ID |
getTiktokCommentsByPostId |
Get comments on a specific video |
Tracking & Monitoring (3 tools)
For continuous brand monitoring and lead generation — Xpoz indexes new matching content as it appears.
| Tool | Description |
|---|---|
addTrackedItems |
Subscribe to ongoing monitoring of a query, user, or subreddit across any platform |
getTrackedItems |
List your active tracked items with last-update timestamps |
removeTrackedItems |
Stop monitoring tracked items by ID |
Key Features
- Natural language queries — describe what you want, Xpoz optimizes the search
- CSV exports — up to 500K rows in a single download via async operations
- Server-side pagination — handle large result sets efficiently
- Field selection — request only the fields you need
- Cache control —
forceLatest: truebypasses cache for real-time data - Async operations — long-running exports with status polling
- OAuth 2.1 — dynamic client registration with Google as the upstream identity provider, no API keys to manage
Use Cases
- Lead generation — find people discussing problems your product solves
- Influencer discovery — identify creators by what they post, not just follower count
- Brand monitoring — track mentions, sentiment, and competitor activity
- Market research — analyze trends and conversations at scale
- Expert finding — discover domain authorities across platforms
- Security intelligence — monitor vulnerability discussions before CVEs publish
Discovery surfaces (for AI agents and scanners)
These public, machine-readable endpoints describe Xpoz to crawlers and agentic clients:
| Surface | URL |
|---|---|
| MCP manifest | https://www.xpoz.ai/.well-known/mcp.json |
| MCP tool catalog (descriptions + JSON-Schema parameters) | https://www.xpoz.ai/.well-known/mcp/tools.json |
| Agent Skills index (agentskills.io v0.2.0) | https://www.xpoz.ai/.well-known/agent-skills/index.json |
llms.txt |
https://www.xpoz.ai/llms.txt |
llms-full.txt |
https://www.xpoz.ai/llms-full.txt |
discovery.txt |
https://www.xpoz.ai/discovery.txt |
| Webhooks contract | https://www.xpoz.ai/webhooks |
| Rate limits & 429 contract | https://www.xpoz.ai/rate-limits |
Links
- 🌐 xpoz.ai
- 💰 Pricing
- 📦 MCP Registry
- 🛠️ ClawHub Skills
- 📖 Documentation
- 🐙 XPOZpublic on GitHub — TS SDK, Python SDK, CLI, agent skills, cookbooks
- 🍺 Homebrew tap
License
This repository contains configuration and documentation only. The Xpoz MCP server is a hosted service — see Terms of Service.
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。