Skip to content

Releases: natetowsley/DiffAid

v0.2.1 - .env discovery fixes

11 Jan 05:51

Choose a tag to compare

DiffAid v0.2.1

Fixes

  • Reliable .env discovery for CLI installs
    • DiffAid now loads .env starting from the user’s current working directory (and searches upward).
    • Fixes cases where .env existed in the project folder but GEMINI_API_KEY was still reported missing.

Notes

  • No new flags or behavior changes beyond .env loading reliability.
  • If you previously had to run $env:GEMINI_API_KEY=... each session, .env should now work as expected when running diffaid from your repo.

v0.2.0 - Enhanced AI Code Review

11 Jan 04:36

Choose a tag to compare

🎉 DiffAid v0.2.0 — Enhanced AI Code Review

Excited to release DiffAid 0.2.0 with powerful new features for smarter, more flexible code reviews!

What’s New?

✅ New Command Flags

  • --json — Machine-readable JSON output for CI/CD pipelines and automation
  • --strict — Treat warnings as errors (exit code 1) for stricter quality gates
  • --detailed / -d — Comprehensive line-by-line analysis with all suggestions
  • --version / -v — Quickly check your installed version

⚙️ Configuration Improvements

  • .env file support — Set your API key once per project, no more repetitive exports!
# Just create a .env file
echo "GEMINI_API_KEY=your-key" > .env
  • Better error messages — Clearer guidance when the API key is missing or a .env file exists but is misconfigured

🧠 Smarter AI Reviews

  • Dual review modes: Default mode gives you 5-10 key findings; detailed mode analyzes everything

  • Severity guidelines — AI now consistently classifies issues as error / warning / note

  • Descriptive messages — AI includes function/class/variable names instead of vague “fix this” messages

  • Quality examples — AI trained on good vs bad message patterns

🔧 Breaking Changes

  • Removed line numbers from findings — they were unreliable with git diffs. DiffAid now focuses on descriptive messages with specific function/variable names instead.

📊 Usage Examples

Quick overview (default):

diffaid

Detailed analysis:

diffaid --detailed

Strict CI/CD mode:

diffaid --strict --json

Check version:

diffaid --version

🐛 Bug Fixes

  • Fixed inconsistent severity classification between review modes

  • Improved schema validation with explicit lowercase severity requirements

  • Better .env file error handling

📦 Installation

pip install --upgrade diffaid

Or for new users:

pip install diffaid

📖 Full Changelog

See CHANGELOG.md for detailed changes:
https://github.com/natetowsley/diffaid/blob/main/CHANGELOG.md

🙏 Thank You

Always open to new feedback and criticism, feel free to share!

Upgrade today and catch more bugs before they reach production!

DiffAid v0.1.0 - Initial Release

07 Jan 01:41

Choose a tag to compare

First Release!

DiffAid is an AI-assisted git diff review CLI that catches bugs before you commit.

Features

  • Smart Analysis - Uses Google Gemini AI to review code changes
  • Three Severity Levels - Errors, warnings, and notes
  • CI Integration - Exit codes for automated workflows (0=success, 1=errors, 2=tool failure)
  • Fast Reviews - Analyzes staged changes in seconds
  • Clean Terminal Output - Color-coded findings with file locations

Installation

pip install diffaid

Quick Start

  1. Get a free Gemini API key: https://aistudio.google.com/apikey
  2. Set your API key:

Mac/Linux:

   export GEMINI_API_KEY="your-key-here"

Windows (PowerShell):

   $env:GEMINI_API_KEY="your-key-here"
  1. Stage your changes and run:
   git add .
   diffaid

What's Included

  • Core CLI functionality
  • Gemini AI integration
  • Comprehensive test suite
  • Full documentation

Requirements

  • Python 3.10+
  • Git
  • Gemini API key (free tier available)

Known Limitations

  • Currently supports only Gemini AI (more providers coming)
  • Requires internet connection for AI analysis
  • API rate limits apply (free tier: 15 requests/min)

Acknowledgments

Powered by Google Gemini AI


Full documentation: https://github.com/natetowsley/diffaid/blob/master/README.md
PyPI: https://pypi.org/project/diffaid/