v0.12.1

Full Changelog: v0.12.0...v0.12.1

v0.12.1

Improvements

• Fixed handling of the -include-files flag: The tool now correctly processes explicitly included files without performing unnecessary codebase access checks. This makes the behavior more intuitive when you want to analyze specific files using the -include-files parameter.

This is a minor bugfix release that improves the reliability of file inclusion functionality.

The change is relatively small but important for users who specifically want to analyze certain files using the -include-files flag, as it now handles this case more efficiently and correctly.

v0.12.0

Full Changelog: v0.11.8...v0.12.0

v0.12.0

New Features & Improvements

Enhanced Error Handling for File Modifications

• Added automatic retry mechanism when file modifications fail
• The tool will now attempt up to 3 retries when encountering errors during file modifications
• Provides clearer error messages and feedback about failed operations

XML Output Format Improvements

• Improved the structure and consistency of code map outputs
• Better handling of file content formatting

Model Updates

• Added support for the latest Claude 3.5 Sonnet model (claude-3-5-sonnet-20241022)

Bug Fixes

• Fixed various edge cases in file handling and code parsing
• Improved error reporting and user feedback

This release focuses on making the tool more reliable and user-friendly, particularly when handling file modifications and errors. The automatic retry mechanism helps ensure that your requested changes are applied successfully on a best effort basis.

v0.11.8

Full Changelog: v0.11.7...v0.11.8

v0.11.8 Release Notes

New Features

• Added support for multiple .cpeignore files throughout the directory hierarchy. The tool now reads and combines ignore patterns from all .cpeignore files found in parent directories, similar to how .gitignore works.

Improvements

• Enhanced file selection reliability by validating that selected files actually exist before processing them. This prevents the tool from attempting to analyze non-existent or hallucinated files that might be suggested by the LLM.
• Improved error handling and retry logic when analyzing code maps, making the tool more robust when dealing with invalid file selections.

Bug Fixes

• Fixed an issue where the tool could attempt to process directories as files during analysis.

These changes make the tool more reliable and flexible when working with larger codebases, especially those that need different ignore patterns at different directory levels.

v0.11.7

Full Changelog: v0.11.6...v0.11.7

v0.11.7

New Features

• Added support for ignoring files and directories using .cpeignore file (similar to .gitignore)
• Added retry mechanism for file analysis to improve reliability when the model returns malformed responses
• Increased request timeout from 2 minutes to 5 minutes to better handle larger codebases

Improvements

• Better token usage reporting with a cleaner, more consistent format
• Command line options have been reorganized into a dedicated package for better maintainability
• Improved handling of tool calls and responses across different model providers (Anthropic, OpenAI, Gemini)

Bug Fixes

• Fixed handling of empty responses in Gemini provider
• Fixed error handling in tool calls to provide better feedback when analysis fails

The primary focus of this release is on improving reliability and user experience when analyzing larger codebases, with the addition of file ignore capabilities and more robust error handling.

v0.11.6

Full Changelog: v0.11.5...v0.11.6

v0.11.5

Full Changelog: v0.11.4...v0.11.5

v0.11.4

Full Changelog: v0.11.3...v0.11.4

v0.11.3

Full Changelog: v0.11.1...v0.11.3

v0.11.1

Full Changelog: v0.11.0...v0.11.1

v0.11.0

Full Changelog: v0.10.0...v0.11.0