mcp-hn-tddschn 2.0.0

MCP Server for Hacker News

Hacker News MCP Server

Fork of https://github.com/erithwik/mcp-hn ,

Changes:

• uses https for everything
• uses curl_cffi
• better coverage of https://hn.algolia.com/api API

---

A Model Context Protocol (MCP) server that provides tools for fetching information from Hacker News.

Tools

• get_stories Fetching (top, new, ask_hn, show_hn) stories
• get_story_info Fetching comments associated with a story (comment_depth, num_comments supported)
• search_stories Searching for stories by query
• search_hn Generic HN search with tags, pagination, numeric filters, and Algolia parameters
• search_comments Search comments globally or within one story (story_id)
• get_item Fetch any HN item by id (story/comment/poll/job)
• get_user_info Fetching user info
• get_user_activity Search author activity across item types (story, comment, poll, job)

Example Usage

Use prompts like the following:

User: Get the top stories of today
  Output: Uses `get_stories` tool and returns a story about AI
User: What does the details of the story today that talks about the future of AI
  Output: Uses `get_story_info` tool based on the results of the previous tool
User: What has the user `pg` been up to?
  Output: Uses `get_user_info` tool and returns a summary of the user's activity
User: What does hackernews say about careers in AI?
  Output: Uses `search_stories` tool and returns a summary of the comments

A more detailed example with the puppeteer MCP server:

User: What are the top stories of today?
  Output: Uses `get_stories` tool and returns a story about AI
User: Can you use the puppeteer tool to read the article about <AI> and also use the hackernews tool to view the comments and give me a summary of what the main comments are about the article?
  Output: Uses puppeteer tool to read the article about AI and then uses the `get_story_info` hn tool to get the comments and returns a summary of the comments

Quickstart

Installing via Smithery

To install Hacker News MCP Server for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install mcp-hn-tddschn --client claude

Claude Desktop:

Update the following:

On MacOS: ~/Library/Application\ Support/Claude/claude_desktop_config.json

On Windows: %APPDATA%/Claude/claude_desktop_config.json

With the following for production:

{
  "mcpServers": {
    "mcp-hn": {
      "command": "uvx",
      "args": ["mcp-hn-tddschn", "--max-response-words", "10000"]
    }
  }
}

Response Trimming

Large tool outputs are trimmed to 10000 words by default to reduce context/token usage.

• CLI option: --max-response-words <int>
• 0 disables trimming

Example:

uvx mcp-hn-tddschn --max-response-words 4000