TrendRadar

TrendRadar MCP Server lets you run AI-powered analysis and conversations over locally stored TrendRadar data using MCP (Model Context Protocol). It provides a programmable endpoint for querying, analyzing, and interacting with your news heat data via MCP clients, enabling natural language queries, topic trend analysis, and cross-client compatibility while keeping your data on your own infrastructure.

How to use

You access TrendRadar through MCP clients by connecting to either the HTTP MCP endpoint or a local stdio MCP server. With an MCP client, you can ask questions like which topics are trending, request trend analyses, or run structured queries against your locally stored news data. Use the HTTP endpoint for remote connections and the stdio setup for local, tightly controlled deployments.

How to install

Prerequisites: You need a machine with container runtime or Python tooling depending on how you plan to run the MCP component. If you want the easiest path, use the Docker-based setup described in the deployment steps below. If you prefer running the MCP server locally, follow the stdio setup commands.

# Option 1: Docker-based deployment (recommended)
# 1) Prepare project directory and environment (as you would for TrendRadar)
# 2) Use docker-compose to start both push service and MCP
# 3) Expose port 3333 for MCP if you want HTTP access
# 4) Connect MCP clients to http://<host>:3333/mcp

# Option 2: Local STDIO MCP server via uv (requires Python environment and UV runtime)
# 2.1 Ensure uv is available, then run the MCP server
uv run python -m mcp_server.server --transport stdio

MCP server configurations

The MCP server can be run in two ways. The HTTP configuration connects to a remote or local MCP endpoint via HTTP. The STDIO configuration runs a local MCP server process that exposes MCP endpoints through standard input/output.

{
  "mcpServers": {
    "trendradar_http": {
      "type": "http",
      "name": "trendradar_http",
      "url": "http://localhost:3333/mcp",
      "args": []
    },
    "trendradar_stdio": {
      "type": "stdio",
      "name": "trendradar_stdio",
      "command": "uv",
      "args": ["--directory","/path/to/TrendRadar","run","python","-m","mcp_server.server"],
      "env": [{"name":"TZ","value":"Asia/Shanghai"}]
    }
  }
}

Security and environment considerations

MCP services expose endpoints that can be accessed by configured clients. If you are in a shared or public environment, restrict access to localhost or to trusted networks. For Docker deployments, consider binding the MCP service to a dedicated interface and using network segmentation. In all cases, protect any tokens or webhook URLs used by other TrendRadar components.

Examples of client integration (tools and endpoints)

You can use various MCP clients to interact with TrendRadar. Basic endpoints you may expect to access include tools for data retrieval and analysis such as querying latest news, analyzing topic trends, and generating summaries. See the tools section for a complete list of capabilities.

Troubleshooting and notes

If the MCP HTTP endpoint is not reachable, confirm the host and port are correct and that the MCP service is running. For stdio mode, ensure the directory path and command line are correct and that the UV runtime is installed. When using multiple MCP clients, verify that the serverIs reachable at the expected URL or port and monitor logs for any authentication or connection errors.

AI capabilities and MCP tooling

TrendRadar MCP supports 13 MCP tools for natural language queries, trend analysis, data insights, sentiment analysis, and more. You can issue questions like: retrieve the latest topics, compare platform impacts, analyze sentiment, and generate summaries. These tools enable you to derive deeper insights from your local TrendRadar data.

Notes on deployment with MCP

When you deploy, you can choose between a remote HTTP MCP endpoint and a local stdio MCP server. Combining both approaches is possible if you want to expose a public API while also running a local, offline analysis engine. Always test the MCP connection with a small query to verify that the client and server can communicate successfully before integrating into broader workflows.

Available tools

get_latest_news

Retrieve the most recent news items from the stored data set.

get_news_by_date

Fetch news items for a specific date range or date.

get_trending_topics

Analyze and return currently trending topics across sources.

search_news

Search for news items matching given keywords or phrases.

search_related_news_history

Find and relate past news items to current topics.

analyze_topic_trend

Analyze the trend of a topic over a specified period.

analyze_data_insights

Generate data-driven insights from stored news data.

analyze_sentiment

Perform sentiment analysis on identified news items.

find_similar_news

Identify news items similar to a given topic or article.

generate_summary_report

Create a concise summary report of identified hotspots.

get_current_config

Return the current MCP server configuration.

get_system_status

Provide the status of the MCP server and components.

trigger_crawl

Manually trigger a data crawl for fresh news.