* 'chore: WIP'
* fix(chat): remove unnecessary focus state and fix recording border styling
- Remove unused isFocused state management from ChatTextArea
- Fix recording border to use inline style with proper error foreground color
- Simplify border logic by removing redundant focus-based styling
- Update Storybook mocks with missing extension state properties
- Add currentApiConfigName and listApiConfigMeta to ChatView stories
* chore: add kilocode change markers and clean up unused speech types
- Add // kilocode_change comments to track custom modifications in type definitions
- Remove deprecated speech recognition message types from WebviewMessage interface
- Add descriptive comments to STT-related component imports
- Mark experiment IDs and message types with change tracking comments
* chore(changeset): add changeset file for version tracking
* Add docs
* Update i18n strings
* bot fixes
* refactor(stt): consolidate FFmpeg detection with path caching
Move FFmpeg availability checking from sttConfig.ts into FFmpegCaptureService
as a static method with global path caching and absolute path resolution.
* fix(stt): cache FFmpeg not-found result to avoid repeated checks
Set cachedFFmpegPath to null when FFmpeg is not found to prevent
repeated filesystem path checks on subsequent availability queries.
Updated the Chinese translation for system notifications feature documentation
and made improvements to the English documentation including better formatting
of troubleshooting solutions and clearer descriptions of notification setup
processes across different operating systems
- Add helper functions to detect and parse dynamic MCP tool names
- Normalize dynamic tools to standard 'use_mcp_tool' format
- Use triple underscores (___) as separator to allow underscores in tool names
- Update tool name generation in mcp_server.ts
- Add comprehensive test coverage for dynamic tool parsing
- Extract toolInputProps and convert to JSON string for MCP server compatibility
* feat(activate): enable auto-launch configuration for tasks
Introduces `.kilocode/launchConfig.json` to automatically start tasks with specified prompts, AI profiles, and modes upon workspace opening.
This allows you to easily run the same prompt against multiple models or multiple profiles by just launching multiple copies of VS Code in the directories with the launch configurations.
* Updated changeset
---------
Co-authored-by: Chris Hasson <noreply@example.com>
Joshua Lambert