使用说明

🐳 Hacker News Sentiment Analysis for Docker

A multi-agent AI system powered by Docker cagent that fetches Docker-related discussions from Hacker News and classifies comments into positive and negative sentiment.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      Docker cagent                               │
│                  (Multi-Agent Orchestrator)                      │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌──────────────┐    ┌───────────────────┐    ┌──────────────┐  │
│  │  HN Fetcher  │───▶│ Sentiment Analyzer │───▶│   Report     │  │
│  │    Agent     │    │      Agent         │    │  Generator   │  │
│  └──────┬───────┘    └───────────────────┘    └──────┬───────┘  │
│         │                                            │          │
│         ▼                                            ▼          │
│  ┌──────────────┐                           ┌──────────────┐    │
│  │mcp-hackernews│                           │   Output     │    │
│  │  MCP Server  │                           │    JSON      │    │
│  └──────────────┘                           └──────────────┘    │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

🚀 Quick Start

Prerequisites

Docker Desktop 4.49+ (includes cagent)
One of: OpenAI API Key, GitHub Token, or Docker Model Runner

Choose Your Model Provider

Provider	Config File	Cost	Setup
OpenAI	`cagent.yaml`	Pay per token	`export OPENAI_API_KEY=sk-...`
GitHub Models	`cagent-github-models.yaml`	Free tier	`export GITHUB_TOKEN=ghp_...`
Local (DMR)	`cagent-local.yaml`	$0	Just Docker!
Multi-Agent	`cagent-multi-agent.yaml`	Varies	Based on models used

Run the Agent

# Clone the repo
git clone https://github.com/ajeetraina/hackernews-sentiment-analysis.git
cd hackernews-sentiment-analysis

# Option 1: OpenAI (best quality)
export OPENAI_API_KEY=your_key_here
cagent run cagent.yaml

# Option 2: GitHub Models (free tier)
export GITHUB_TOKEN=your_github_pat
cagent run cagent-github-models.yaml

# Option 3: Local models (free, private)
docker model pull ai/llama3.3:70B-Q4_0
cagent run cagent-local.yaml

# Option 4: Multi-agent workflow
cagent run cagent-multi-agent.yaml

📊 What It Does

Fetches Docker-related stories from Hacker News
Collects all comments from high-engagement discussions
Analyzes sentiment (Positive/Negative/Neutral)
Extracts key themes from the community
Generates a structured JSON report

🎯 Sentiment Classification

Positive Themes

Ease of use / Developer experience
Portability / Cross-platform
DevOps workflow improvement
Community and ecosystem
Reproducibility

Negative Themes

Performance overhead
Resource consumption
Docker Desktop pricing
Complexity for simple use cases
Security concerns
macOS/Windows issues

📁 Project Structure

.
├── cagent.yaml                 # OpenAI version
├── cagent-github-models.yaml   # GitHub Models version (free)
├── cagent-local.yaml           # Docker Model Runner (local, free)
├── cagent-multi-agent.yaml     # Multi-agent with sub-agents
├── docker-compose.yaml         # Run with web dashboard
├── templates/
│   └── index.html              # Web dashboard UI
└── output/
    ├── dashboard.json          # Dashboard data
    ├── positive_comments.json  # Positive comments
    └── negative_comments.json  # Negative comments

🔧 Configuration Options

Change Search Queries

Edit the instruction in any cagent YAML file:

instruction: |
  Search Hacker News for Docker-related stories using these queries:
    - "Docker"
    - "Docker security"      # Add custom queries
    - "Docker alternatives"
    - "Podman vs Docker"

Switch Models

# Override model at runtime
cagent run cagent.yaml --model openai/gpt-4o-mini
cagent run cagent.yaml --model anthropic/claude-sonnet-4-5
cagent run cagent.yaml --model dmr/ai/qwen3:8B-Q4_0

🌐 Web Dashboard (Optional)

Run with Docker Compose to get a web UI:

docker compose up -d
open http://localhost:8080

📈 Example Output

{
  "summary": {
    "total": 156,
    "positive": 67,
    "negative": 52,
    "neutral": 37,
    "sentiment_score": 0.18
  },
  "top_positive_themes": [
    {"theme": "Ease of Use", "count": 24},
    {"theme": "Portability", "count": 19}
  ],
  "top_negative_themes": [
    {"theme": "Performance Overhead", "count": 18},
    {"theme": "Docker Desktop Pricing", "count": 11}
  ]
}

🔗 Related Resources

🤝 Contributing

Contributions welcome! Feel free to:

Add new analysis features
Improve sentiment classification
Add more visualization options
Create integrations (Slack, Discord, etc.)

📝 License

MIT License - Feel free to use and modify!

Built with ❤️ by Ajeet Singh Raina using Docker cagent and mcp-hackernews