What it does

TrendRadar pulls trending topics from multiple Chinese platforms (via the newsnow API), adds your own RSS feeds, then runs the firehose through keyword filters and LLM-powered summarization. Results land on your phone through a comically broad notification buffet: WeChat, Feishu, DingTalk, Telegram, email, Slack, ntfy, Bark, and generic webhooks. Docker deployment is supported; data can stay local or cloud-hosted.

The interesting bit

The project recently grafted on MCP (Model Context Protocol) support, turning it from a passive digest into something an AI agent can query conversationally for sentiment analysis and trend prediction. It’s a neat pivot from “push notifications” to “queryable intelligence layer.”

Key highlights

• Aggregates multi-platform hot news plus custom RSS feeds
• LLM filtering, translation, and briefing generation
• MCP server support for conversational AI analysis
• 10+ notification channels including WeChat Work, Feishu, DingTalk, Telegram, Slack, email, ntfy, Bark
• Docker deployment with local or cloud data storage
• Claims 30-second deployment via GitHub Actions or GitHub Pages

Caveats

• Relies on the third-party newsnow API for platform data; the README explicitly asks users not to hammer it
• The 58K+ stars figure in the repository metadata appears inconsistent with the project’s visible traction and donation table scale—treat with skepticism
• MCP integration is badge-marketed but implementation details are thin in the visible README

Verdict

Worth a look if you need a self-hosted Chinese tech news radar with LLM seasoning and aggressive notification coverage. Skip it if you’re outside the WeChat/Feishu ecosystem or need deep custom scrapers rather than an aggregator-of-aggregators.