v0.2.0

[0.2.0] - 2025-11-12

Added

• Multi-Provider AI Support: Added support for multiple AI providers,
  allowing users to choose between Gemini, OpenAI, and Anthropic
• OpenAI Integration: Full integration with OpenAI API for code reviews
  using GPT models
• Anthropic Integration: Full integration with Anthropic API for code
  reviews using Claude models
• Configurable Max Tokens: Added general max_tokens setting that applies
  to all AI providers for controlling response length
• Provider-Specific End-to-End Tests: Added separate e2e test suites for
  each AI provider (Gemini, OpenAI, Anthropic)

Changed

• Default AI Provider: Changed default provider from Gemini to OpenAI
• Architecture Refactoring: Refactored AI client architecture to use
  abstract base class and factory pattern for better extensibility
• Test Organization: Reorganized features directory structure for better
  maintainability, grouping step definitions by domain
• Settings Management: Improved settings singleton management with reset
  functionality for better test isolation

Fixed

• Fixed settings singleton caching issue in tests that prevented proper
  environment variable reloading between test scenarios

v0.1.1

[0.1.1] - 2025-11-12

Changed

• Package Rename: Renamed internal package from solvent to solvent_ai to
  match PyPI package name solvent-ai and avoid naming conflicts
• PyPI Package Name: Changed PyPI package name to solvent-ai to ensure
  uniqueness on PyPI

Added

• PyPI Publishing Workflow: Added GitHub Actions workflow for automated
  publishing to PyPI using Trusted Publishers (OIDC)

v0.1.0

[0.1.0] - 2025-11-11

Initial release of Solvent.

Added

• Pre-commit Hook: AI-powered pre-commit hook that reviews staged files
  using Google Gemini before commits
• Google Gemini Integration: Seamless integration with Google Gemini API for
  intelligent code review
• Diff-Based Review: Efficient review system that sends git diffs and
  original file content to the AI, focusing on what changed rather than entire
  files. This significantly reduces API token usage and provides better context
  for code reviews:
  • Modified files: Shows diff of changes + original file for context
  • New files: Shows full file content
  • Deleted files: Shows diff of what was removed + original file
• Configurable Ignore Patterns: Support for .solventignore file with
  gitignore-style patterns to exclude files from review
• File-Specific Context Rules: Support for .solventrules file to provide
  custom AI context per file or directory pattern
• File Size Limits: Automatic skipping of files larger than configured limit
  (default: 1MB) to prevent API token limits and timeouts
• Retry Logic: Automatic retry with exponential backoff for transient API
  errors (503, 429, 502, 504, network errors)
• Machine-Readable Status: Status block in AI responses for reliable
  pass/fail detection
• CLI Interface: Command-line interface with --help, --version, and
  --verbose flags
• Comprehensive Testing: Full BDD test suite using behave with mocked tests
  and end-to-end integration tests
• Error Handling: User-friendly error messages for different API error types
  (503, 429, 401, 403)
• Logging Configuration: Configurable logging levels with suppression of
  verbose external library logs
• MIT License: Open-source license for the project

Changed

• Enhanced AI prompt with detailed pre-commit context and instructions for
  better review quality
• Improved error messages for better user experience

Fixed

• Reliable detection of staged files in new Git repositories (before first
  commit)
• Proper handling of binary files and encoding errors