TrendRadar: Self-Hosted AI Trend Monitor with Multi-Platform Aggregation, RSS, and Smart Alerts

By Prahlad Menon 3 min read

Information overload is the ambient condition of building in tech right now. Tracking what matters — across platforms, in real time, with signal separated from noise — requires either a full-time analyst or infrastructure that does it for you.

TrendRadar is the open-source, self-hosted version of that infrastructure.

It aggregates trending topics from dozens of platforms, runs AI filtering and analysis, and pushes smart briefings to whatever notification channel you actually use. Docker deploy, local data, MCP-compatible for AI agent integration.

What It Does

Multi-platform aggregation. TrendRadar pulls trending topics from dozens of sources via the NewsNow API: Hacker News, GitHub Trending, Reddit, Product Hunt, and major Chinese platforms (Weibo, Zhihu, Bilibili, and more). Add any RSS feed to extend coverage to your specific domain.

AI filtering. Define keyword filters. The AI scores each trending item for relevance, discards noise, and surfaces what actually matters to you. No more scrolling 200 HN submissions to find the 3 relevant ones.

AI translation + analysis. Non-English trending content gets translated and summarized automatically. A briefing arrives in your notification channel — not a link dump, but a synthesized summary of what’s happening and why it matters.

Smart push alerts. Configured alerts fire when monitored keywords hit a threshold of trending activity. Channels supported: Telegram, Slack, WeChat, DingTalk, Feishu/Lark, email, ntfy, Bark (iOS).

MCP server. TrendRadar exposes its data via MCP — so Claude Code, Cursor, or any MCP-compatible AI assistant can query live trend data as a tool call. Ask “what’s trending in AI right now” from inside your coding context and get a real answer.

Deploy in 5 Minutes

# Clone
git clone https://github.com/sansan0/TrendRadar.git
cd TrendRadar

# Configure
cp config.example.yaml config.yaml
# Edit config.yaml:
# - Add your LLM API key (OpenAI, Anthropic, or local Ollama endpoint)
# - Add Telegram bot token + chat ID (or Slack webhook, etc.)
# - Set your keyword filters

# Start
docker compose up -d

Local LLM (Ollama) setup:

llm:
  base_url: "http://host.docker.internal:11434/v1"
  api_key: "ollama"
  model: "llama3.2"

Telegram bot quickstart:

notifications:
  telegram:
    bot_token: "YOUR_BOT_TOKEN"
    chat_id: "YOUR_CHAT_ID"
    enabled: true

Data is stored locally in SQLite by default. Push frequency, filtering thresholds, and source selection are all configurable.

MCP Integration

For AI agent workflows, TrendRadar’s MCP server is the interesting piece. Add it to your MCP config:

{
  "mcpServers": {
    "trendradar": {
      "command": "npx",
      "args": ["-y", "trendradar-mcp"],
      "env": {
        "TRENDRADAR_URL": "http://localhost:8080"
      }
    }
  }
}

Now your AI assistant can:

  • what's trending in AI today → live aggregated results
  • any GitHub repos trending this week related to agents → filtered, summarized
  • set up an alert for mentions of [keyword] → configure via natural language

This closes the loop between monitoring and action — the agent that helps you act on trends can also see what those trends are in real time.

Why Self-Hosted Matters Here

Trend monitoring is an area where the content of what you’re tracking is sensitive. Competitive intelligence, early signals on topics you care about professionally, keyword-based monitoring of your own brand or products — you probably don’t want that flowing through a SaaS provider’s servers.

TrendRadar keeps everything local: the data, the AI analysis, the alert history. The LLM calls go to whichever endpoint you configure — local Ollama, your own Azure OpenAI instance, or the public APIs with your own keys.

The Stack

  • Backend: Python, lightweight, designed for low resource footprint
  • Data storage: SQLite (local) or configurable cloud backend
  • AI: Any OpenAI-compatible API — GPT-4o, Claude, Gemini, Llama via Ollama
  • Deployment: Docker Compose
  • Notifications: Telegram, Slack, WeChat, DingTalk, Feishu, email, ntfy, Bark
  • MCP: Exposes trend data as Model Context Protocol tools

Resources