Releases: johnbean393/Sidekick
0.0.11
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added ability to generate images (Requires macOS 15.2+)
Bug Fixes
- Fixed error where generated images would be overwritten by new prompts
- Added dialog for user to chose response type (image or text) if classifier has insufficient confidence
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.10
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added support for speculative decoding, which can deliver up to a 40% increase in performance
- Added add files button to expose temporary resources capability
Incremental Improvements
- Better system prompts to improve output
Bug Fixes
- Fixed error where the model used for speculative decoding cannot be selected
- Fixed jumbled output when using speculative decoding
- Fixed excessively long waits for the server before switching to local inference
- Fixed issue where graphs would be tiny when the message contains LaTeX
- Fixed pause in playing audio when a message is sent
- Fixed error where LaTeX was not rendering properly
- Extended timeout to accommodate reasoning models like QwQ-32B-Preview
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.9
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added inline commands in 3rd party apps
Bug Fixes
- Fixed fatal errors when adding a command
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.8
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added ability to use remote LLM servers with an OpenAI compatible endpoint
Bug Fixes
- Fixed the issue where some inference settings would not appear
- Removed Sidekick restart requirement when toggling whether the server is used
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.7
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added ability to render LaTeX
Bug Fixes
- Fixed error with inline LaTeX
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.6
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added ability to generate charts
- Added model selection sheet
- Improved default model selection; no more under-utilizing system resources
Bug Fixes
- Fixed error where numbers were not recognized
- Fixed list reordering
- Fixed toolbar appearance
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.5
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Added ability to load temporary resources into context in full.
- Added tip to complement resource loading
- Optimized CPU usage during generation
- Improved chat UI
Bug Fixes
- Fixed JSON parsing error
- Fixed text field would receive dropped files as a POSIX path
- Added missing translations
- Fixed issue where first 10 characters of generated text is not shown during generation
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.4
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Tips are displayed to ease the learning curve and aid feature discovery
Bug Fixes
- Missing translations were added
- Fixed error where parts of response containing non-references JSON would be hidden.
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.3
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- Users can now directly view content fed to the chatbot
- Users can now export conversations in .png format
Bug Fixes
- Raised auto scroll threshold to reduce chance of scrolling error
- Fixed bug where toggling "Use Search Result Context" did nothing
- Fixed bug where prompt field would not clear after message is sent.
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.
0.0.2
Update
Features
- Local LLM
- File, folder, website indexing and context
- Web search
New Features
- On-device transcription for speech input using Apple's
Speechframework - Users can now optimize search results in Settings
Bug Fixes
- Fixed missing Chinese, Simplified translation during
searchandprocessingstages - Fixed issue where lengthy tasks popover was missing on macOS 14
Installation
You can download the package (.pkg) and launch it. Note that since it is not notarized, you will need to enable it in System Settings.