# AIHOT [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/KKKKhazix/AIHOT  
**GitHub Stars:** 7634  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/aihot

## Description
Chinese AI news: search topics, explore hot events, and read daily, weekly and monthly briefs.

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "aihot": {
    "command": "npx",
    "args": ["-y","aihot"],
    "env": {
      "LLM_BASE_URL": "",
      "LLM_MODEL": "",
      "LLM_EXTRA_JSON": "",
      "LLM_REASONING_TOKENS": "",
      "ADMIN_PASSWORD": "",
      "GROUP_REVIEW_MODEL": ""
    }
  }
}
```

**Requires environment variables:** `LLM_BASE_URL`, `LLM_MODEL`, `LLM_EXTRA_JSON`, `LLM_REASONING_TOKENS`, `ADMIN_PASSWORD`, `GROUP_REVIEW_MODEL` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What AIHOT does

The AIHOT MCP server gives agents access to content produced by an AIHOT news site. Its documented content includes topic and full-text search, current hot events, and daily, weekly, and monthly briefs. The underlying application is designed for creating an industry-specific news site: operators choose sources, categories, topics, prompts, selection thresholds, and branding.

AIHOT is not limited to an AI-news feed. The repository includes example sources and configuration intended to be replaced with sources for another sector, such as legal, human resources, finance, or precious metals. The MCP endpoint is one of several agent-facing outputs; the same site can also provide RSS feeds, a public API, Agent Markdown, and `llms.txt`.

## How it works

A configured source enters a processing pipeline that removes duplicates, applies a preliminary filter, and scores potentially relevant items twice using independent evaluations. Selected material receives a Chinese title, summary, recommendation reason, categories, and tags. Foreign-language articles can also be translated.

Related reports are grouped into events. Candidate matches are found using title and summary vectors, or text overlap when no vector service is configured, then reviewed by a model. Event popularity is based on independent sources within a recent time window, rather than counting every duplicate article. Daily editions select one item per event and follow-up coverage only when there is new progress; weekly and monthly editions are compiled from daily editions.

## Setup and configuration

Run the project with Docker and Node.js 24. Initialization requires an API key for an OpenAI-compatible model service. The documented quick-start flow clones the repository, runs `node scripts/init-env.ts --llm-key <key>`, and starts the stack with `docker compose up -d --build`. The local site is then available at `http://localhost:3000`, with administration at `/admin`.

Model settings can be adapted for providers such as DeepSeek, Qwen, or Zhipu by changing `LLM_BASE_URL`, `LLM_MODEL`, and `LLM_EXTRA_JSON` in `.env`. Reasoning models may also require `LLM_REASONING_TOKENS`. The administrator password is configured with `ADMIN_PASSWORD`. After startup, `/agent` provides the available MCP, RSS, and API connection details.

## Tools and capabilities

The AIHOT MCP server is suited to agent workflows that need to:

- Search article titles, summaries, and full text.
- Inspect topics and current hot events.
- Read daily, weekly, and monthly briefings.
- Consume curated content from a customized industry site.
- Use the same publication data through MCP, RSS, or the public API.

The exact MCP tool names are not specified in the provided material, so integrations should obtain the connection instructions from the running site's `/agent` page.

## Limitations and notes

AIHOT depends on the sources and model configuration supplied by the operator. The repository contains 18 public overseas AI sources as examples, not the project's complete production source list or operating data. The selection prompts and thresholds are editable, and their quality depends on the chosen standards and calibration samples.

The AIHOT MCP server is part of the site's engine rather than a separately documented hosted service. The provided material does not specify a standalone package command, transport details, authentication scheme, or exact MCP tool names. The AIHOT name and logo are not covered by the MIT code license, and the repository asks operators to use their own branding.

_Full upstream README: https://allmcps.com/mcp/aihot/readme_