Releases: microsoft/genaiscript
v1.95.3
π Enhancements
- Model Encoding Support: Introduced
modelEncodings
in configuration, enabling custom model-to-encoding mappings for greater flexibility and compatibility. - Dynamic Encoder Loading: Improved token encoder resolution by dynamically selecting encodings based on the
modelEncodings
configuration.
π οΈ Improvements
- Config Management: Enhanced configuration handling with a new
config
getter, ensuring reliable access to loaded configurations. - Test Host Updates: Added a stubbed
config
getter toTestHost
for consistency across host implementations.
π Sample Config Update
- Added example
modelEncodings
entries ingenaiscript.config.json
to demonstrate usage.
v1.95.2
- π Enhanced Development Experience: Added new devcontainer configurations, improving local development setups with pre-installed tools and features like
common-utils
,git
, andffmpeg
. - ποΈ File Caching: Introduced a robust file caching mechanism, enabling efficient storage and reuse of processed files across runs.
- π οΈ Schema Validation Upgrade: Integrated JSON schema validation using external schemas, enhancing configuration reliability and error handling.
- π₯ AI Kitchen Script: Added a new
ai-kitchen
script showcasing video analysis with transcript generation, frame extraction, and YouTube-ready content creation. - π JSON Schema Inference: Leveraged
@stoplight/json-schema-generator
for dynamic schema generation from objects, simplifying workflows. - πΌοΈ Improved Image Handling: Enhanced MarkdownTrace to support embedded images with better file and URL handling.
- π Web Scraping Support: Expanded browser locator capabilities with new methods for handling multiple elements.
v1.95.1
π Enhanced GitHub Copilot Chat Integration
- Replaced "client" language model with
github_copilot_chat
for seamless integration. - Improved error messages for better clarity when models are unavailable.
π οΈ Schema Improvements
- Added advanced JSON Schema validation and inference capabilities.
- Enhanced tooling for stricter schema generation and validation.
π Copy Code Button
- Introduced a "Copy" button for code snippets in the web UI, making it easier to share and reuse code.
πΉ GitHub Copilot Chat Demo
- Added a sample video for showcasing GitHub Copilot Chat capabilities.
π§ Bug Fixes and UI Tweaks
- Addressed minor UI inconsistencies in Markdown rendering and trace panels.
- Improved error handling and logging for GitHub-related operations.
v1.95.0
What's New in 1.95.0 π
- Enhanced Security π: Introduced a robust Content Security Policy (CSP) for webviews, ensuring stricter control over resources like scripts, styles, and frames.
- Improved Webview Experience π₯οΈ: Added a persistent webview panel with seamless reveal functionality, providing a smoother user interface for error handling and output display.
- Streamlined Codebase π οΈ: Replaced custom nonce generation with a secure
randomHex
utility, enhancing maintainability and reliability.
v1.94.2
π Enhanced Model Management: Improved error handling and added support for listing models across multiple providers, ensuring a smoother user experience when managing language models.
π§ Client Language Model Refactor: Transitioned the client language model to runtimeHost
for better modularity and maintainability.
π Web Compatibility: Introduced support for web-browser scenarios, including enhanced CSP handling and iframe-based views for seamless integration.
πΉ Video Blogifier Update: Enhanced video transcription and processing with improved prompt handling and dynamic topic resolution for better YouTube optimization.
π οΈ Dependency Updates: Refreshed dependencies across packages to maintain stability and compatibility.
π Bug Fixes: Addressed issues with client builds, environment configurations, and context menu visibility to enhance overall reliability.
v1.94.1
π Enhanced File Handling: Added support for resolving data URIs directly in files, enabling seamless processing of base64-encoded data.
π Improved Token Configuration: Refined token handling logic for better compatibility with bearer tokens across various providers.
π Documentation Update: Updated Azure AI Foundry documentation for clarity and consistency.
π οΈ Simplified Validation: Removed redundant version checks in Azure environment parsing for a cleaner configuration process.
v1.94.0
What's New in 1.94.0
π Enhanced Video Processing
- Introduced keyframe extraction and scene detection using FFmpeg, enabling better video analysis and integration.
- Streamlined video transcription and screenshot capabilities for improved LLM input preparation.
π οΈ Improved CLI Functionality
- Added
--out-output
flag for better output management. - Centralized AI request dispatch logic for consistent and efficient processing.
π Advanced Data Parsing
- Added support for JSONL file parsing, expanding data format compatibility.
π€ AI Enhancements
- Introduced a new
classify
prompt for text and image categorization with entropy and logprob analysis. - Improved handling of Hugging Face token prefixes for multi-token scenarios.
π¨ Developer Experience
- Enhanced trace handling and output management for better debugging and insights.
- Refactored internal logic for clarity and maintainability.
π₯ Additional Video Tools
- Added support for high-detail image scaling and scene-based frame extraction.
This release brings significant improvements to video processing, CLI tools, and AI integration, ensuring a smoother and more powerful experience for developers.
v1.93.0
What's New in GenAIScript 1.93.0
π Keyframe Extraction and Scene Detection
Introducing enhanced video processing capabilities! Extract keyframes or detect scenes with adjustable thresholds using the new --keyframes
and --scene-threshold
options in the CLI.
π£οΈ Local Whisper ASR Support
Seamlessly integrate the Whisper ASR Webservice for transcription tasks. Configure it as a model provider and enjoy local transcription support.
π Improved Notebook Execution
Notebook frontmatter now supports additional run options like models and providers, enabling more flexible and powerful script execution.
π§ Enhanced FFmpeg Integration
Optimized frame extraction with wildcard-based output handling, scene-specific frame selection, and improved caching for smoother workflows.
π Bug Fixes
- Resolved issues with parsing notebook frontmatter.
- Fixed minor inconsistencies in transcription and frame extraction logic.
v1.92.3
What's New in GenAIScript 1.92.3
π Enhanced Video and Audio Processing
- Added
probe
command to extract detailed metadata from video/audio files. - Improved
extract-audio
andextract-frames
commands with new options for transcription, frame count, size, and format.
π₯ Large Media Support
- Optimized frame extraction for large images and videos, ensuring better performance and flexibility.
π Transcription Improvements
- Revamped transcription caching and error handling for seamless speech-to-text conversion.
βοΈ FFmpeg Integration
- Introduced FFmpeg feature installation in dev environments for enhanced media processing.
π οΈ Developer Experience
- Simplified CLI workflows and added comprehensive tests for video processing tasks.
Enjoy smoother media handling and powerful new tools!
v1.92.2
What's New in 1.92.2
π Enhanced Webview Handling
Improved handling of external URIs for webview assets, ensuring seamless integration and better external resource management.
π₯ Video Processing Improvements
- Added support for caching during audio extraction with FFmpeg, optimizing repeated operations.
- Updated video transcription to leverage caching for alt-text generation.
π Cleaner Repositories
Introduced .gitignore
for large local test video files, keeping repositories tidy.
π οΈ Bug Fixes & Optimizations
- Resolved issues with optional parameters in FFmpeg operations.
- Adjusted input handling to ensure consistent directory management.