GitHub Issues AI Agent

A lightweight, AI-powered command line tool that fetches GitHub issues, prioritizes them using Azure OpenAI (GPT-5/GPT-5.1), and identifies duplicates.

Features

• Multi-repo support: Fetch issues from one or more repositories.
• Smart Prioritization: Uses LLM reasoning to assign P0 (Critical), P1 (High), or P2 (Normal) priority.
• Duplicate Detection: Identifies potential duplicate issues in the same analysis pass.
• Concise Reporting: Generates a clean Markdown report with priorities and reasoning.

Installation

Requires Python >= 3.11.

1. Clone the repository:

   git clone https://github.com/your-org/github-issues-agent.git
   cd github-issues-agent

2. Create & activate a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # macOS/Linux
   # On Windows: .venv\Scripts\activate

3. Install the package:

   pip install -e .

Configuration

Create a .env file with your credentials:

GITHUB_TOKEN=your_github_pat
AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/
AZURE_OPENAI_API_KEY=your_api_key
AZURE_OPENAI_DEPLOYMENT=gpt-5.1
AZURE_OPENAI_API_VERSION=2025-04-01-preview

Usage

Run the agent via the command line:

issues-agent --repos owner/repo --output report.md

Alternatively, you can still use:

python -m issues_agent --repos owner/repo --output report.md

Options

Argument	Description	Default
--repos	Comma-separated list of repositories (owner/repo)	Required
--output	Output markdown file path	report.md
--limit	Max issues to fetch per repo	100
--since	Fetch issues updated since (e.g. 2y, 6m, 30d)	2y
--model	LLM model name to use	gpt-5.1
--dry-run	Print report to stdout instead of writing to file	False

Examples

Analyze a single repo:

issues-agent --repos microsoft/vscode --limit 50

Analyze multiple repos with a specific model:

issues-agent --repos owner/repo1,owner/repo2 --model gpt-4o --output triage.md

Fetch only recent issues:

issues-agent --repos owner/repo --since 30d

Project Structure

github-issues-agent/
├── issues_agent/
│   ├── __init__.py
│   ├── __main__.py          # Entry point
│   ├── cli.py               # CLI entry point
│   ├── config.py            # Config loading
│   ├── github_client.py     # GitHub API client
│   ├── llm_client.py        # Azure OpenAI client
│   ├── analysis.py          # LLM analysis logic
│   ├── report.py            # Report generator
│   └── logging.py           # Logging setup
├── tests/                   # Tests
├── .env.example             # Env template
├── README.md
└── requirements.txt

License

MIT