sansan0/TrendRadar

TrendRadar pulls hot topics from multiple platforms plus arbitrary RSS feeds, runs keyword filtering, and then passes the results through an LLM to produce translated summaries, sentiment analysis, and trend briefings — all pushed to whatever notification channel your team already uses. The self-host path with Docker is the real differentiator here. Most hosted social-listening tools charge per seat and per keyword, and they keep your data on their servers. This runs on your own infra and integrates with MCP for natural-language querying on top of the collected data. The multi-channel push support (Telegram, Slack, WeChat, Feishu, DingTalk, email, and more) means it fits into existing ops workflows rather than requiring a new dashboard habit. Reservation: the codebase leans Chinese-market-first in its default sources, so founders targeting English-language audiences will need to configure their own feed list carefully. -> Best for: SaaS team of 2-5 that needs competitive or market-signal monitoring without paying enterprise SaaS prices