Hacker News Agent

An AI agent built while learning from HuggingFace's Agents Course. It analyzes trending Hacker News discussions — fetches top threads, explains why they're popular, and extracts key insights from community comments.

HN Agent UI

What it does

• Fetch top HN stories with engagement metrics (score, comments, score/comment ratio)
• Extract and summarize top comments from any thread
• Answer follow-up questions about specific stories

Getting Started

Install dependencies

pip install -r requirements.txt

Configure your LLM models

Gemini (via LiteLLM):

MODEL_PROVIDER=gemini
GEMINI_API_KEY=your_key_here
MODEL_ID=gemini/gemini-2.5-flash

OpenAI:

MODEL_PROVIDER=openai
OPENAI_API_KEY=your_key_here
MODEL_ID=gpt-4o-mini

HuggingFace Inference:

MODEL_PROVIDER=hf_inference
HF_TOKEN=your_token_here
MODEL_ID=Qwen/Qwen2.5-72B-Instruct

Run

python scripts/run_gradio.py

Open http://localhost:7860 in your browser.

How to Use

Just ask questions in natural language:

• "What's trending on Hacker News right now?"
• "Fetch the top 3 stories and explain why they're popular"
• "What are people discussing in the top HN thread?"
• "Give me a quick summary of today's top 5 stories"

The agent figures out which tools to call on its own. It fetches stories, analyzes engagement, and pulls comments as needed.

How It Works

Built on the smolagents CodeAgent, which follows a Thought -> Action -> Observation loop. The agent reasons about your question, picks the right tool, reads the output, and formulates a response.

Two tools:

• fetch_top_stories — Gets top N stories from HN with metadata (story ID, title, score, comments, URL, engagement ratio)
• extract_comment_insights — Pulls top comments for a story by its numeric ID and identifies discussion themes

Resources

• smolagents Documentation

License

MIT License