Skip to content

0.1.15#58

Merged
eren23 merged 4 commits intomainfrom
feat/update-52
Mar 6, 2026
Merged

0.1.15#58
eren23 merged 4 commits intomainfrom
feat/update-52

Conversation

@eren23
Copy link
Owner

@eren23 eren23 commented Mar 6, 2026

No description provided.

eren23 added 4 commits March 5, 2026 20:37
- Added `vision_analyze` tool for image analysis using vision-capable LLMs, supporting URLs, file paths, and base64 data.
- Introduced `OpenRouterPreferences` dataclass for managing upstream provider routing preferences.
- Implemented model capabilities caching with `is_vision_capable()` and `get_cached_capabilities()` functions.
- Refactored code to reduce duplication in OpenAI and OpenRouter providers.
- Improved security for vision tools, including path traversal and SSRF protections.
- Fixed issues with image handling in various providers and improved error handling for OpenRouter preferences.
- Bumped version to 0.1.14.
Add OSS Demo Playbook to documentation

- Included a new entry for the OSS Demo Playbook in both mkdocs.yml and index.md.
- The playbook covers Big-3 agent benchmarking on large OSS repositories using code-intel tools.
- Added support for warning messages when images are dropped due to provider limitations in the ProductionAgent.
- Implemented a method to retrieve and clear the last image warning.
- Updated the ZAI provider to strip images and log warnings when images are detected, as it does not support inline images.
- Enhanced the image extraction utility to handle macOS drag-and-drop paths with backslash-escaped spaces.
- Improved the TUI to redirect pasted images and display warnings for stripped images.
- Refactored message building to ensure user-provided images are not restricted by working directory.
- Added unit tests for new image handling features and provider behavior.
- Added vision support with the `vision_analyze` tool for image analysis using LLMs.
- Introduced inline image support in TUI, allowing drag-and-drop or paste functionality.
- Enhanced image path extraction utility for better handling of various input formats.
- Updated ZAI provider to handle image blocks more effectively, including stripping and placeholder management.
- Improved security measures for vision tools and refined caching mechanisms for model capabilities.
- Added extensive tests for vision features and other functionalities to ensure robustness.
@eren23 eren23 merged commit 57c0955 into main Mar 6, 2026
3 of 4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant