hidrix-tools
MCP tool server that gives any AI agent the ability to search, scrape, and analyze content across the internet.
README
<h1 align="center">hidrix-tools</h1>
<p align="center"> MCP tool server that gives any AI agent the ability to search, scrape, and analyze content across the internet. </p>
<p align="center"> <a href="https://github.com/sonpiaz/hidrix-tools/blob/main/LICENSE"><img src="https://img.shields.io/github/license/sonpiaz/hidrix-tools" alt="License" /></a> <a href="https://github.com/sonpiaz/hidrix-tools/stargazers"><img src="https://img.shields.io/github/stars/sonpiaz/hidrix-tools" alt="Stars" /></a> <img src="https://img.shields.io/badge/MCP-compatible-purple" alt="MCP" /> <img src="https://img.shields.io/badge/Bun-TypeScript-black" alt="Bun" /> </p>
What is hidrix-tools?
AI agents like Claude, OpenClaw, Hermes, Cursor, and pi can read files and write code — but they can't search the web, read social media, or scrape data on their own.
hidrix-tools fixes that. It's an MCP server that provides tools for web search, social media search, Facebook scraping, content analysis, and more. Plug it into any agent that supports MCP — no vendor lock-in.
Your AI Agent hidrix-tools Internet
┌──────────┐ MCP protocol ┌──────────────┐ ┌─────────┐
│ Claude │ ◄─────────────────► │ web_search │ ◄────────────► │ Brave │
│ OpenClaw │ │ x_search │ │ X/Twitter│
│ Hermes │ │ reddit_search│ │ Reddit │
│ Cursor │ │ facebook_ │ │ Facebook│
│ pi │ │ scraper │ │ YouTube │
│ Codex │ │ content_ │ │ TikTok │
│ ... │ │ scorer │ │ Meta Ads│
└──────────┘ └──────────────┘ └─────────┘
Why use hidrix-tools?
| Without hidrix-tools | With hidrix-tools |
|---|---|
| Agent can only read local files | Agent can search web, read articles, scrape social media |
| "I don't have access to the internet" | "Here are the top 10 Reddit posts about your topic this week" |
| Manual copy-paste research | Agent autonomously gathers data from 7+ platforms |
| No competitive intelligence | Agent scrapes competitor FB pages, analyzes engagement patterns |
| One platform at a time | Cross-platform research in a single conversation |
Compatible agents
| Agent / Platform | Setup | Status |
|---|---|---|
| Claude Code / Claude Desktop | settings.json → mcpServers |
✅ Tested |
| pi | Native extension (no MCP) | ✅ Tested |
| OpenClaw | openclaw.json → mcp_servers |
✅ Compatible |
| Hermes Agent | config.yaml → mcp_servers |
✅ Compatible |
| Cursor | MCP settings | ✅ Compatible |
| Codex CLI | MCP config | ✅ Compatible |
| Any MCP client | stdio transport | ✅ |
Tools
Search
| Tool | Source | What it does |
|---|---|---|
web_search |
Brave Search | Search the web — titles, URLs, descriptions |
x_search |
GetXAPI / RapidAPI | Search X/Twitter with engagement data (likes, retweets, views) |
x_thread_reader |
GetXAPI | Read full X threads, articles, and tweet replies |
x_user_posts |
GetXAPI | Get recent posts from a specific X user with engagement |
reddit_search |
RapidAPI | Search Reddit with full text and engagement data |
reddit_thread_reader |
Reddit JSON API (free) | Read full post + comment tree |
reddit_subreddit_top |
Reddit JSON API (free) | Top posts from any subreddit by time range |
youtube_search |
RapidAPI | Search YouTube videos |
tiktok_search |
RapidAPI | Search TikTok videos |
Scrape & Fetch
| Tool | Source | What it does |
|---|---|---|
web_fetch |
Mozilla Readability | Fetch any URL → clean readable markdown |
facebook_scraper |
Apify + Meta API | Scrape FB groups, pages, keyword search, or Meta Ad Library |
Analyze
| Tool | Source | What it does |
|---|---|---|
content_scorer |
Built-in | Score and rank posts by weighted engagement + time-decay |
content_analyzer |
Built-in | Topic clusters, content patterns, timing trends, author leaderboard |
| Tool | Source | What it does |
|---|---|---|
linkedin_search |
Apify (no login) | Search LinkedIn posts by keyword with engagement data |
linkedin_profile |
Apify (no login) | Get recent posts from a LinkedIn profile |
Intel
| Tool | Source | What it does |
|---|---|---|
similarweb |
RapidAPI | Website traffic, rank, engagement analytics |
Storage & Automation
| Tool | Source | What it does |
|---|---|---|
data_store |
SQLite (built-in) | Save/query scraped posts, dedup, run history, storage stats |
Real-world use cases
🔍 Research agent
"Find top 10 posts about AI agents on Reddit this week, compare with X/Twitter, summarize the trends"
Agent uses: reddit_search → x_search → content_analyzer
📊 Content pipeline
"Scrape 5 Facebook groups about marketing VN, rank top 100 posts by engagement, analyze patterns"
Agent uses: facebook_scraper (group mode) → content_scorer → content_analyzer
🕵️ Competitive intelligence
"What ads is competitor X running on Facebook? What's their messaging strategy?"
Agent uses: facebook_scraper (ads mode) → Meta Ad Library API (free)
📰 Article research
"Read these 5 articles and summarize the key insights"
Agent uses: web_fetch (for each URL)
📈 Market research
"What are people saying about product X across Reddit, X, and YouTube?"
Agent uses: reddit_search + x_search + youtube_search → content_analyzer
🏆 Top content discovery
"Find the most viral posts in this Facebook group this month"
Agent uses: facebook_scraper → content_scorer (sort by engagement)
Install
Quick setup
git clone https://github.com/sonpiaz/hidrix-tools.git ~/.hidrix-tools
cd ~/.hidrix-tools && bun install && cp .env.example .env
# Add your API keys to .env
Connect to your agent
Claude Code — add to ~/.claude/settings.json:
{
"mcpServers": {
"hidrix-tools": {
"command": "bun",
"args": ["run", "/path/to/hidrix-tools/server.ts"]
}
}
}
OpenClaw — add to openclaw.json:
mcp_servers:
hidrix-tools:
command: "bun"
args: ["run", "/path/to/hidrix-tools/server.ts"]
Hermes Agent — add to ~/.hermes/config.yaml:
mcp_servers:
hidrix-tools:
command: "bun"
args: ["run", "/path/to/hidrix-tools/server.ts"]
pi (coding agent) — copy extension:
cp -r integrations/pi-extension ~/.pi/agent/extensions/hidrix-tools
# Then /reload in pi
See integrations/pi-extension/README.md for details.
Standalone (any MCP client):
bun run server.ts
API Keys
Tools are auto-skipped if their API keys are missing — only configure what you need.
| Key | Tools | Free Tier | Get it |
|---|---|---|---|
BRAVE_API_KEY |
web_search | 2,000 queries/month | api.search.brave.com |
GETXAPI_KEY |
x_search, x_thread_reader, x_user_posts | $0.001/call, no subscription | getxapi.com |
RAPIDAPI_KEY |
x_search (fallback), reddit_search, youtube_search, tiktok_search | Varies | rapidapi.com |
SUPADATA_API_KEY |
youtube_channel (primary) | Paid from $9/mo | supadata.ai |
APIFY_API_TOKEN |
facebook_scraper (group/page/search) | $5/month credit | console.apify.com |
META_ADS_ACCESS_TOKEN |
facebook_scraper (ads mode) | Free, unlimited | facebook.com/ads/library/api |
Add your own tool
Tools are auto-discovered — no changes to server.ts needed.
cp -r tools/_template tools/your-tool-name
# Edit tools/your-tool-name/index.ts
bun run server.ts
# [hidrix-tools] ✓ your_tool_name
See docs/adding-a-tool.md for the full guide.
Project structure
hidrix-tools/
├── server.ts — MCP server, auto-discovers tools
├── tools/
│ ├── _template/ — Copy this to create a new tool
│ ├── web-search/ — Brave Search
│ ├── web-fetch/ — URL → markdown
│ ├── x-search/ — X/Twitter search (GetXAPI + RapidAPI fallback)
│ ├── x-thread-reader/ — Read X threads & articles
│ ├── x-user-posts/ — User timeline with engagement
│ ├── reddit-search/ — Reddit search (upgraded)
│ ├── reddit-thread-reader/ — Full post + comment tree (free API)
│ ├── reddit-subreddit-top/ — Subreddit top posts (free API)
│ ├── youtube-search/ — YouTube search
│ ├── tiktok-search/ — TikTok search
│ ├── similarweb/ — Traffic analytics
│ ├── facebook-scraper/ — FB groups, pages, search, ads
│ ├── content-scorer/ — Engagement scoring + ranking
│ └── content-analyzer/ — Topic/pattern/trend analysis
├── lib/
│ ├── tool-registry.ts — Auto-discovery engine
│ ├── rapidapi.ts — Shared RapidAPI client
│ ├── getxapi.ts — GetXAPI client (X/Twitter)
│ ├── apify.ts — Apify REST client
│ ├── meta-ads.ts — Meta Ad Library API client
│ └── readability.ts — HTML → markdown extraction
└── docs/
├── adding-a-tool.md — Tool authoring guide
├── feedback-and-bug-reporting.md — Bug reporting standards
├── known-issues.md — Tracked limitations
└── ROADMAP.md — Development roadmap
Docs
| Doc | Description |
|---|---|
| Adding a tool | How to create a new tool |
| Feedback & bug reporting | Standards for detecting and reporting issues |
| Known issues | Tracked provider limitations and quirks |
| Roadmap | Development roadmap (X upgrade, Reddit upgrade, LinkedIn, extensions) |
Tech stack
| Technology | Purpose |
|---|---|
| Bun | Runtime |
| TypeScript | Language |
| MCP SDK | Agent protocol |
| Brave Search | Web search |
| Mozilla Readability | Content extraction |
| Apify | Facebook scraping |
| Meta Ad Library API | Ad intelligence (free) |
Contributing
Contributions are welcome. See CONTRIBUTING.md for guidelines.
When reporting bugs, follow docs/feedback-and-bug-reporting.md.
Related
- affiliate-skills — 45 AI agent skills
- content-pipeline — AI-powered LinkedIn content generation
- Kapt — macOS screenshot tool with annotation & OCR
- Yap — Push-to-talk dictation for macOS
License
MIT — see LICENSE for details.
<p align="center">Built by <a href="https://github.com/sonpiaz">Son Piaz</a></p>
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。