TrendRadar

TrendRadar MCP Server provides an AI-powered analysis backend that lets you run local, private analyses on your news data. It exposes a programmable interface so you can query trending topics, perform topic analysis, and obtain structured insights using MCP-compatible clients. This server is designed to work with your own data and to integrate with a variety of client tools for natural language interaction and advanced analytics.

How to use

Connect to TrendRadar MCP Server from any MCP-enabled client. You have two main options: run the MCP server locally (stdio) or run it as a remote HTTP service. The HTTP option exposes a REST-like endpoint at the MCP URL you configure, while the stdio option lets a client launch the server as a local process.

How to install

Prerequisites: you need Docker installed if you want the Docker-based setup, or you can run the MCP server as a local process using the stdio approach.

Option 1: Run MCP as a remote HTTP service (Docker recommended)

# 1) Get the latest TrendRadar repository (recommended method)
git clone https://github.com/xhh-im/trendradar.git
cd trendradar

# 2) Start the MCP service in a container (recommended for isolation and ease of setup)
# This runs the MCP server in a separate container and exposes port 3333
# You may customize environment variables as needed in docker/.env

docker compose up -d trend-radar-mcp

# 3) Verify the MCP HTTP endpoint
curl http://127.0.0.1:3333/mcp

Option 2: Run MCP as a local stdio process (direct command)

# 1) Ensure Python is installed (recommended), and install dependencies if needed
# 2) Run MCP server via stdio mode using uv (as shown in documentation examples)
uv run python -m mcp_server.server --directory /path/to/TrendRadar run

Note: If you run through the stdio path, the command is typically launched by an MCP client with a complete directory reference. The key starting point is the uv command plus the path to TrendRadar, then the mcp_server module. If you are following a documented flow, ensure the path to TrendRadar is correctly provided.

## Additional sections

MCP server endpoints and commands are described in the configuration sections below. The HTTP method is the preferred approach when you want a remote MCP API, while the stdio method is suitable for local development and testing. Ensure you expose the correct port and manage security considerations for remote access.

## Configuration and usage notes

The MCP server supports multiple connection modes and environments. If you plan to run both the HTTP and stdio variants, you can configure both entries in your MCP client configuration so you can switch between local testing and remote deployment as needed.

## Security and maintenance

Keep MCP-related endpoints isolated to trusted networks. Use authentication where possible and rotate secrets regularly. When running in Docker, prefer the Docker-based workflow for easier lifecycle management and better reproducibility.

## Troubleshooting tips

If the MCP HTTP endpoint is not responding, check container status, confirm port 3333 is open, and verify that the MCP server process is running. For stdio mode, ensure the uv executable is accessible and that the directory path to TrendRadar is valid.

## Tools exposed by the MCP server

The MCP server exposes a suite of analytical tools that you can call from MCP clients. These tools enable you to fetch latest news, query by date, explore trending topics, perform advanced searches, and run topic and data analyses. See metadata for the complete list and descriptions.

## Available tools

### get\_latest\_news

Fetches the latest news items based on configured keywords and platforms.

### get\_news\_by\_date

Retrieves news items for a specific date range or day.

### get\_trending\_topics

Returns topics that are currently trending across configured platforms.

### search\_news

Performs advanced search across stored news data with filters.

### search\_related\_news\_history

Finds related historical news for a given topic.

### analyze\_topic\_trend

Analyzes the trend of a topic over time, including velocity and momentum.

### analyze\_data\_insights

Generates data insights and cross-platform comparisons.

### analyze\_sentiment

Performs sentiment analysis on news headlines and articles.

### find\_similar\_news

Finds news items similar to a given piece of content.

### generate\_summary\_report

Creates a summarized report of detected hotspots and trends.

### get\_current\_config

Returns current MCP and server configuration.

### get\_system\_status

Provides current status of the MCP system and components.

### trigger\_crawl

Manually trigger a data crawl of configured platforms.