githai 0.1.1

A CLI tool that combines GitHub's API with AI to help you understand and analyze repositories

GITHAI CLI

A CLI tool that combines GitHub's API with the LLM API to help you understand, analyze, and work with GitHub repositories and issues more effectively.

Quick Start

# Install the CLI
pip install githai

# Set up your GitHub token (required for GitHub API access)
ghai keys set GITHUB_TOKEN

# Analyze a GitHub issue with AI
ghai explain-issue --issue-url https://github.com/owner/repo/issues/123

# List available commands
ghai --help

API Keys and Model Configuration

GHAI supports multiple AI models from different providers. You can configure API keys for the following services:

Supported Models and Keys

• GitHub Models

  • github/gpt-4o-mini
  • github/claude-3-haiku
  • etc...
  • Requires: GITHUB_TOKEN

• OpenAI Models

  • gpt-4
  • gpt-3.5-turbo
  • etc...
  • Requires: OPENAI_API_KEY

• Anthropic Models

  • claude-3-sonnet
  • claude-3-haiku
  • etc...
  • Requires: ANTHROPIC_API_KEY

Setting API Keys

# GitHub token (required for all GitHub API access)
ghai keys set GITHUB_TOKEN

# OpenAI API key (for OpenAI models)
ghai keys set OPENAI_API_KEY

# Anthropic API key (for Anthropic models)
ghai keys set ANTHROPIC_API_KEY

# List all stored keys
ghai keys list

# Show the path to your keys file
ghai keys path

Default Model Configuration

Set a default model to use across all commands:

# Set default model (recommended: use free GitHub models)
ghai keys set-default-model github/gpt-4o-mini

# View current default model
ghai keys get-default-model

# Clear default model setting
ghai keys clear-default-model

If no default model is set, commands will use their built-in defaults.

Key Management

# Delete a specific key
ghai keys delete OPENAI_API_KEY

# View all available key management commands
ghai keys --help

Example Commands and Usage

Explain Issue

Analyze GitHub issues with AI to get detailed explanations, context, and insights:

# Analyze a GitHub issue using the default prompt
ghai explain-issue --issue-url https://github.com/owner/repo/issues/123

# Use a custom prompt file
ghai explain-issue --issue-url https://github.com/owner/repo/issues/123 --prompt custom_prompt.md

# Short form
ghai explain-issue -u https://github.com/owner/repo/issues/123 -p custom_prompt.md

Progress Update

Get progress updates on GitHub issues and their related sub-issues:

# Get progress update using default prompt
ghai progress-update --issue-url https://github.com/owner/repo/issues/123

# Use a custom prompt file
ghai progress-update --issue-url https://github.com/owner/repo/issues/123 --prompt progress_prompt.md

# Short form
ghai progress-update -u https://github.com/owner/repo/issues/123 -p progress_prompt.md

Custom Prompts

Most commands support custom prompts to tailor the AI analysis to your specific needs.

Using Custom Prompts

1. Create your prompt file in Markdown format
2. Use the --prompt option with any command:

   ghai explain-issue -u <issue-url> --prompt your-custom-prompt.md
   

3. Path resolution: Prompts can be specified with:
   • Absolute paths: /full/path/to/prompt.md
   • Relative paths: ./prompts/custom.md

Use prompt best practices.

Project Structure

ghai/
├── src/
│   └── ghai/
│       ├── __init__.py
│       ├── cli.py              # Main CLI entry point
│       ├── github_api.py       # GitHub GraphQL API client
│       ├── llm_api.py          # AI/LLM integration
│       ├── commands/           # Command modules
│       │   ├── __init__.py
│       │   ├── explain_issue.py    # Issue analysis command
│       │   ├── snippet_update.py   # Code snippet updates
│       │   ├── progress_update.py  # Progress reporting
│       │   └── keys.py             # Key management
│       ├── types/              # Type definitions
│       │   └── github_api_types.py
│       └── util/
│           ├── __init__.py
│           └── key_util.py  # Key utility functions
├── prompts/                    # AI prompt templates
│   ├── explain_issue_prompt.md
│   ├── initiative_prompt.md
│   └── snippet_prompt.md
├── pyproject.toml
└── README.md

License

MIT License