TrendRadar
by sansan0
TrendRadar's MCP server puts 24 tools for cross-platform trend search, sentiment analysis, and alerting in front of an agent, on top of a crawler watching 35+ platforms.
Add to Claude Code
claude mcp add trendradar -- uv --directory /path/to/TrendRadar run python -m mcp_server.server
About
TrendRadar is a public-opinion and trend-monitoring crawler that aggregates hot topics from 35+ platforms plus RSS feeds, then applies AI filtering, translation, and sentiment analysis before pushing alerts to channels like WeChat, Feishu, DingTalk, Telegram, Slack, and email. Its mcp_server module (python -m mcp_server.server, built on FastMCP) puts that same pipeline behind roughly 24 MCP tools, so an agent can ask about trends conversationally instead of you reading a generated report.
The tool surface splits cleanly by task: date-range resolution for natural-language queries like 'last 7 days,' news and RSS search, trend/sentiment analysis and period comparisons, batched article reading (capped at 5 per call), and direct notification dispatch to any of the 9 supported channels. Because the crawler and the MCP server are the same codebase, tool calls reflect exactly what's configured in the project's own config.yaml, with no separate data pipeline to keep in sync.
Key features
- Aggregates trending topics from 35+ platforms plus custom RSS feeds
- ~24 MCP tools covering date parsing, news/RSS search, trend analysis, and article reading
- AI-driven keyword filtering, translation, and sentiment analysis on top of raw aggregated data
- Direct notification dispatch to 9 channels (WeChat, Feishu, DingTalk, Telegram, Slack, email, and more)
- Batched article reading capped at 5 per call to keep responses manageable
- Docker-deployable with locally or cloud-persisted data
Use cases
- Asking an agent to summarize what's trending across platforms over a given date range
- Running sentiment or trend comparison between two time periods without building a custom script
- Having an agent push a generated news digest straight to a team's Feishu or Slack channel
- Searching aggregated RSS and platform data by keyword as part of a research workflow
Available tools
resolve_date_range
Converts a natural-language date phrase like 'last 7 days' into a standard date range.
search_news
Searches aggregated cross-platform news by keyword.
analyze_sentiment
Runs sentiment analysis over a set of aggregated news or RSS items.
generate_summary_report
Produces a synthesized summary report from aggregated news and trend data.
send_notification
Dispatches a message to one of the 9 supported notification channels.
Frequently asked questions
Do I need to run the crawler separately from the MCP server?
No. The mcp_server module lives in the same TrendRadar repo and reads from the same config.yaml as the core crawler, so one setup covers both.
Can it push results somewhere other than chat?
Yes. send_notification dispatches directly to channels like WeChat, Feishu, DingTalk, Telegram, Slack, email, ntfy, or Bark, independent of whatever client you're chatting in.