Releases: johnbean393/Sidekick
Releases · johnbean393/Sidekick
1.0.0 Release Candidate 16
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Deep Research
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
feat: Implement GraphRAG
feat: Encourage querying expert resources using tools
feat: Overall index progress
feat: Revamped chat attachments
feat: Paste to add attachments
Fixes and Performance Improvements
fix: Reorder inference settings
fix: GLM 4.6 reasoning on OpenRouter
fix: Crash during indexing
fix: GraphRAG indexing indicator
fix: Improve table rendering performance
fix: Track token usage
fix: Fix resources indexing limbo state
fix: Fix toolbar text color
fix: Resume indexing
fix: Prompt field scrolling
fix: Fix cancelled requests
fix: Delete attachments for deleted conversations
fix: Increase extraction speed
fix: Improve attachments UI
fix: Reduce startup latency
fix: Stop worker model toggling vision
fix: Remove provider ranking param
1.0.0 Release Candidate 15
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Deep Research
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- feat: Add
run_pythontool - feat: Added Google as a search provider
- feat: Add todo tools
- feat: Context compression for long tool call chains
- feat: Improved model picker
- feat: Show model capabilities in picker
- feat: Allow toggling functions by group
- feat: Auto collapse code blocks and messages
Fixes and Performance Improvements
- fix: Improve tool calling stability
- fix: Improve reliability with malformed tool calls
- fix: Parallel tool calls
- fix: Fixed Claude 4 Thinking and Moonshot models via OpenRouter
- fix: Fix repeated models
- fix: Fix model name in selector
- fix: Scroll to position on generation end
- fix: Model identifiers with remove providers
- fix: Color
Codableconformance - fix: Greedy user message view
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 14
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Deep Research
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- feat: Support
gpt-oss - feat: Support
glm-4.5 - feat: Updated app icon
- feat: Compile for macOS 26
Fixes and Performance Improvements
fix: Improved chat rendering performance
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 13
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Deep Research
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- feat: Add Homebrew to PATH if installed
- feat: llama.cpp native function calling
- feat: Improved HTML conversation export
- feat: Allow adjusting completion threshold
- feat: Add Claude 4
- feat: Add Mistral's Magistal
- feat: Add MiniMax-M1
- feat: Prioritize provider throughput
- docs: Add Deep Research
Fixes and Performance Improvements
- fix: add creation of marp target directory to move binary
- fix: Fix streaming with high throughput providers
- fix: UI render increment
- fix: Lower completions memory usage
- fix: Recognize models with
:thinkingsuffix as reasoning capable
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 12
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Deep Research
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- feat: Add card for Deep Research tool
- feat: Use Tavily Extract if available
- feat: Updated embedded docs
- feat: Add xAI to list of providers
- feat: Add Deep Research button
- feat: Deep Research
- feat: Improve inline Markdown image
- feat: Use Tavily advanced search for Deep Research
- feat: Drag to export Deep Research diagrams
- feat: Support llama-server vision
- feat: Allow skipping tutorial
- feat: Added inference records
- feat: Add dashboard
- feat: Adapt to dashboard selection
- refactor: Mermaid diagram rendering
Fixes and Performance Improvements
- fix: Inline writing asisstant repeating
- fix: Align capsule button menu to leading edge
- fix: Deep Research preview progress updating
- fix: Uncomment research
- fix: Improved propagation between research sections
- fix: Preview showing after Deep Research is interrupted
- fix: Add DeepSeek to known reasoning models
- fix: Mitgated blank text in flowchart SVGs
- fix: Repeated generation in Deep Research
- fix: Repeated input tokens in Deep Research
- fix: Table not updating with filter
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 11
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- Implemented router for hybrid reasoning model
- Update default completions model
- UI for common advanced parameters
- Improve long function chain execution
- Search on specific site
Fixes and Performance Improvements
- Add Qwen3-32B to list of known hybrid reasoning models
- Set thinking to false for conversation titles
- Update llama-server docs link
- Avoid auto extracting code into snapshot
- Enlarged prompt bar buttons click area
- Return icon not appearing
- onKeyPress not working
- Reason button animation glitch
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 10
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- Add support for Qwen3 style hybrid reasoning (tested with leaked quants)
Fixes and Performance Improvements
- Added error iteration and few shot prompting to Diagrammer
- Runaway error pipe file handle
Reasonbutton default state with local models- Remove
-instructsuffix from Qwen3 model names - Function schema not injected into system prompt with local models
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 9
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
Fixes and Performance Improvements
- Fixed prompt bar for logographic languages
- Allow deletion of all local models
- Trim
web_searchcontent if too long - Make
web_searchcontent trimming greedy - Fixed quick prompts
- Text cursor jumps to end
- Prompt bar animations
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 8
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Memory
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant & Completions
- Detector
New Changes
- Added
planfunction & improved rendering performance - Added
get_user_selectionfunction - Improved temperature slider
- Use Command + Return to send
- Support multiple tool calls
- Improved web search with time sensitive queries
- Update onboarding & translations
- Added example
Tutorialexpert to onboarding - Diagrammer uses mermaid instead of d2 to allow more diagram types
- Added memory
- Add
Forgetmemory button - Button to open memories window
Fixes and Performance Improvements
- Tool calling with local models
- Function use defaults to false on local models
- Prompt field insertion index crash
- Tool parsing for OpenAI style calls
- Improved prompt field contrast
- Update function permissions
- Prompt field causes hang when typing quickly
- No selectable model names if
/v1/modelsfails - Fix prompt bar animations
- Improve title editor responsiveness
- Improve message rendering
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.
1.0.0 Release Candidate 7
Update
Features
- Local LLM & remote VLMs with an OpenAI compatible API
- File, folder, website indexing and context
- Web search
- Function calling
- Canvas
- Image generation
- Extensions
- Diagrammer
- Slide Studio
- Inline Writing Assistant
- Detector
New Changes
- Added 9 functions, including functions for calendar, reminder and location operations
- Added plan function & improved rendering performance
- Added support for web search with DuckDuckGo
- Add progress indicator shimmer
Fixes
- Show function calling progress for OpenAI compatible tool calling
- Support for o4-mini
- Increase file text extraction quality
- Fixed incorrect translations
- Tool calling with local models
Installation
Download the disk image (.dmg), mount it, then drag the app into the Applications folder.