PlugWiki
TrendRadar logo

TrendRadar MCP Server: Setup for Claude Code, Cursor and Other Apps

CommunityLocal

Lets Claude, Cursor and other assistants search and analyse trending news and RSS feeds that TrendRadar collects, then send alerts.

TrendRadar collects trending headlines from many platforms plus your RSS feeds and pushes alerts to chat apps and email. Its MCP server lets your assistant search that data, compare periods, rate sentiment and generate summary reports.

The server runs on your computer or in Docker next to TrendRadar, and it reads the data TrendRadar has already saved. You clone the repository and run it yourself, so setup takes a few steps. The README and tool descriptions are mostly in Chinese.

Good for

  • Analysts tracking public opinion and trending topics on Chinese platforms
  • PR and marketing teams watching how a brand or topic is covered
  • Researchers comparing coverage across platforms and RSS feeds
  • Self-hosters who already run TrendRadar

Works in

12 should work

Tap an app to see how to set it up and any known problems.

Install TrendRadar MCP

Claude Code

  1. 1

    Copy this prompt.

    Add the TrendRadar MCP server to Claude Code. Run this command:
    
    claude mcp add trendradar -- uv --directory /path/to/TrendRadar run python -m mcp_server.server
    
    Then check that it loaded and tell me if I need to sign in or restart.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires uv, Cloned TrendRadar repo.

Replace /path/to/TrendRadar with the folder you cloned. The HTTP option needs the server running first, with start-http.sh or the Docker image.

Built from the package data we hold. Not yet tested by us.

Tools

  • Turns phrases like last week into exact dates

    Converts a natural phrase such as this week or last 7 days into a precise date range on the server, so every model gets the same dates. The server recommends calling it first.

    Try: “What are the exact dates for last week?”

    Read · resolve_date_range

  • Shows the newest trending headlines

    Returns the latest batch of crawled hot-list news, optionally for chosen platforms such as Zhihu or Weibo, up to 1000 items.

    Try: “What is trending on Weibo and Zhihu right now?”

    Read · get_latest_news

  • Gets headlines from a past day or range

    Returns news for a date or date range, for history and comparison.

    Try: “Show me the headlines from yesterday.”

    Read · get_news_by_date

  • Counts the hottest topics

    Counts topic frequency, either for your saved watch words or by pulling frequent words from headlines automatically.

    Try: “Which topics are hottest today?”

    Read · get_trending_topics

  • Shows new items from your RSS feeds

    Returns recent items from your RSS feeds, up to 30 days back. RSS data is stored apart from hot-list news.

    Try: “Show me the latest items from Hacker News and 36kr.”

    Read · get_latest_rss

  • Searches your RSS feeds

    Finds RSS articles that contain a keyword, over up to 30 days.

    Try: “Search my feeds for machine learning.”

    Read · search_rss

  • Shows which RSS feeds have data

    Lists your configured RSS feeds and how many items each has today.

    Try: “Which RSS feeds are working?”

    Read · get_rss_feeds_status

  • Searches headlines by keyword, fuzzy text or name

    Searches the hot lists and RSS in keyword, fuzzy or entity mode, for a date range and chosen platforms.

    Try: “Search for news about Tesla this week.”

    Read · search_news

  • Finds news similar to a headline

    Finds other headlines related to a title you give, today or from history, with a similarity cut-off you can set.

    Try: “Find news related to this headline.”

    Read · find_related_news

  • Tracks how a topic's attention rises and falls

    Analyses one topic over time: trend, life cycle, sudden spikes or a prediction.

    Try: “Analyse how much attention Tesla got over the last 7 days.”

    Read · analyze_topic_trend

  • Compares platforms and keyword patterns

    Compares how platforms cover a topic, how active each platform is, or which keywords appear together.

    Try: “Which platforms talk most about AI?”

    Read · analyze_data_insights

  • Rates the mood of news about a topic

    Analyses sentiment and popularity of news about a topic over a date range. Your AI client reads the data and judges the sentiment.

    Try: “How positive is the news about Bitcoin today?”

    Read · analyze_sentiment

Read from the source code

Permissions

medium risk · 4 things to watch

Medium risk

The assistant can send messages to your configured alert channels and start crawls, and article reading goes through the Jina AI Reader service. Leave alert channels unconfigured and keep the HTTP port on localhost if you want it to only look.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • It can send messages to your real channels

    send_notification posts to every channel you set up, such as Telegram, Slack or email. Leave channels unconfigured if the assistant should not send anything. github.com

  • Article reading goes through a third-party service

    read_article and read_articles_batch send article addresses to Jina AI Reader. Do not use them for private or internal pages. github.com

  • Crawling and syncing write files

    trigger_crawl can save results to your output folder and sync_from_remote copies data from cloud storage to disk. github.com

  • The HTTP mode listens on a local port

    Run with uv, the server's --host default is 0.0.0.0, so HTTP mode listens on every network interface unless you pass --host 127.0.0.1. The Docker setup in the README publishes port 3333 on 127.0.0.1. Keep the port off public networks. github.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Talks toJina AI Reader (read_article). Notification services you configure. GitHub (version check). News platforms (trigger_crawl)
Sign-in and keysNo account for the server itself. Notification channels and storage keys live in your config.yaml or .env.

Questions

Is TrendRadar an MCP server or something else?

Both. TrendRadar is mainly a news and trend monitor with push alerts. The MCP server is an optional part that lets an assistant analyse the data it collects.

Do I need an API key?

The MCP server itself needs none. Alerts need the webhooks or tokens of the channels you use, and its AI features may need a key for the model you pick in TrendRadar's own settings.

Does it need data before it works?

Yes. The README's troubleshooting says tools fail when no crawl data exists. Run TrendRadar's crawler first, or have the assistant start a crawl.

Can I run it with Docker?

Yes. The README names the image wantcat/trendradar-mcp, which serves the MCP endpoint on port 3333 and reads your config and output folders read-only. The docker compose setup starts it as its own service.

Which clients does it work with?

The README gives setup for Cursor, Cline, Continue and the MCP Inspector, plus a generic template for any MCP client. Use the install box for the stdio and HTTP options.

Can I use it read-only?

The README does not describe a read-only switch. You can lower the risk by not configuring any alert channels, since sending is the main write action.

How many tools does it have?

The server code registers 27 tools. The README's inspector check mentions 17, which is the older count, so trust the list on this page.

Information

updated Nov 2, 2026

Publisher
Community
Transport
stdio, Streamable HTTP
License
GPL-3.0
Last commit
Oct 29, 2026
Open issues
30

WebsiteGitHubIssues

Sources

Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.

Show all 2
  1. github.com · github.com
  2. github.com · github.com

Facts from the official MCP registry, GitHub, npm, PyPI and the Docker catalog. Updated Nov 2, 2026.