unitsauce 0.1.1

AI-powered test failure analysis and fix suggestions

UnitSauce

AI-powered test failure analysis and auto-fix for Python projects.

UnitSauce analyzes failing tests, identifies bugs in your code changes, and generates fixes using Claude AI. Works as a CLI tool or GitHub Action.

---

Features

• Automatic failure detection — Runs pytest and identifies failing tests
• AI-powered analysis — Uses Claude to understand the bug from git diff
• Smart fixes — Generates minimal code or test fixes
• Verification — Confirms the fix actually works before reporting
• PR comments — Posts fix suggestions directly to your pull request

---

Quick Start

CLI Usage

pip install unitsauce

# Auto-detect whether to fix code or test
unitsauce /path/to/project

# Force fix code only
unitsauce /path/to/project --mode code

# Force fix test only
unitsauce /path/to/project --mode test

# Output as markdown
unitsauce /path/to/project --output markdown

GitHub Action

Add to your workflow (.github/workflows/test.yml):

name: Tests
on:
  pull_request:

permissions:
  pull-requests: write

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 2
      
      - uses: actions/setup-python@v5
        with:
          python-version: '3.11'
      
      - name: Install dependencies
        run: pip install pytest
      
      - name: Run tests
        run: pytest
      
      - name: UnitSauce Analysis
        if: failure()
        uses: zanstarasinic/unitsauce@main
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}

---

Configuration

CLI Options

Option	Description	Default
--mode	auto, code, or test	auto
--output	console, markdown, or json	console

Environment Variables

Variable	Required	Description
ANTHROPIC_API_KEY	Yes	Your Anthropic API key
GITHUB_TOKEN	For PR comments	Provided automatically by GitHub Actions

---

Requirements

• Python 3.10+
• Git repository
• Anthropic API key
• pytest

---

How It Works

1. Detects failures — Runs pytest and collects failing tests
2. Analyzes changes — Gets git diff to see what changed
3. Identifies affected code — Maps failures to modified functions
4. Generates fix — Sends context to Claude, gets minimal fix
5. Verifies — Applies fix, runs test again to confirm
6. Reports — Shows diff or posts PR comment

---

Example PR Comment

UnitSauce Analysis

Found 1 failing test(s), fixed 1.

---

test_calculator.py::test_add

Error: assert 6 == 5

Fixed by: Updating test in test_calculator.py

- assert add(2, 3) == 5
+ assert add(2, 3) == 6

---

Local Development

git clone https://github.com/zanstarasinic/unitsauce.git
cd unitsauce
pip install -e .

---

License

MIT