v0.0.7 Release Notes

Overview

Key Features

🔄 Provider Architecture Improvements

• Shared image source handling across providers — Unified image processing utilities between Gemini and Vertex AI providers, eliminating code duplication and ensuring consistent behavior
• Shared video source handling across providers — Extracted video processing logic for HTTP URLs, local files, and YouTube videos, now available to both providers
• VertexAI HTTP/local file support — Vertex AI provider now supports HTTP URLs and local file uploads (previously Gemini-only)
• Provider env loading warnings — Better diagnostics when providers are misconfigured or missing credentials

📹 Video Processing Enhancements

• Size-aware video handling — Intelligent video processing based on file size
• 50MB inline threshold — Videos under 50MB are sent inline; larger videos use file upload APIs for better reliability
• Improved reliability — Enhanced error handling and security in file operations

🎨 Analysis Capabilities

• Conditional system prompts — Context-aware prompts for all analysis modes (palette, hierarchy, components)
• Dynamic prompt modes — analyze_image tool now supports multiple analysis modes for specialized outputs
• Updated default model — Switched to gemini-3.1-flash-lite-preview for better performance and cost efficiency

Technical Details

Breaking Changes

None — this is a fully backward-compatible release.

Dependencies

• Updated @google-cloud/storage handling for improved stability
• No new external dependencies added

Testing

All 10 test cases pass with both Gemini and Vertex AI providers:

• ✅ Image analysis (URL, local file, base64)
• ✅ Image comparison (2 and 3 images)
• ✅ Object detection (URL, local file)
• ✅ Video analysis (remote URL, local file, YouTube)

Migration Guide

No migration needed. Existing code continues to work without changes. To take advantage of new features:

1. Use Vertex AI for local files — Vertex AI now supports local file uploads like Gemini
2. Leverage dynamic prompt modes — Use options.mode in analyze_image for specialized analysis
3. Benefit from improved video handling — Large videos are now handled more reliably

What's Next

Future releases will focus on:

• Additional analysis modes and specialized outputs
• Performance optimizations for batch operations
• Enhanced error recovery and retry logic