- Updated timeout settings for various tools to improve responsiveness and prevent blocking the agent loop.
- Enhanced error handling in tools to provide clearer feedback on invalid paths and execution failures.
- Implemented abort signal support in multiple tools, allowing for graceful cancellation of long-running operations.
- Adjusted the `walk` function to respect abort signals and limit the number of files processed, preventing hangs during directory traversals.
- Improved the `grepTool`, `globTool`, and `fileSearchTool` to handle timeouts and abort signals effectively, enhancing overall tool performance.
- Updated the `semanticIndex` module to support GPU acceleration with ONNX Runtime, improving performance on compatible devices.
- Introduced new functions to determine preferred embedding devices and handle device status, enhancing the flexibility of the embedding process.
- Expanded the `IndexStatus` interface to include backend type, device information, and runtime details for better tracking of indexing status.
- Enhanced the UI in the settings panel to display runtime and device information, providing users with clearer insights into the indexing process and model usage.
- Added support for per-tool timeout overrides in seconds, allowing users to customize tool execution limits.
- Introduced `setToolTimeoutOverrides` function to apply user-defined timeout settings.
- Updated the `FeatureConfig` interface and default settings to include `toolTimeoutsSec`.
- Enhanced the settings UI to display and manage tool timeout values, ensuring a better user experience for configuring tool behavior.
- Introduced a new `indexWatch` module to handle debounced file watching for incremental updates to the semantic index.
- Updated `activate` function to initialize the index watcher and replace the previous build index call with a warm disk and incremental sync.
- Enhanced the `semanticIndex` module to support enabling/disabling indexing and improved index path normalization.
- Modified the settings panel and UI components to include an option for enabling/disabling indexing, reflecting changes in the feature store.
- Ensured that indexing operations respect the new `indexingEnabled` flag across various functionalities, including search and document indexing.
- Added per-workspace conversations with automatic migration from global conversations.
- Implemented auto-loading of GGUF models with a loading card in chat.
- Enhanced llama.cpp server to use random free ports with retry on bind failure.
- Improved composer dropdown positioning and shared model selection across composers.
- Fixed production error related to missing '@huggingface/hub' package.
- Removed MCP tool marketplace.
- Updated version in package.json to 0.0.2.
Pawan Osman