How kunal0040/kaitu works
Kaitu is a personal AI assistant whose primary workflow accepts keyboard or voice input, routes it to local replies, memory handling, commands, calculations, or the AI brain, and returns a response. Voice modes add speech input and spoken output; commands cover Windows applications, web search, and reader/calculator skills. Memory management classifies facts before reading or updating persistent memory. The README documents these capabilities, but several implementation files were not sampled, so uncertain internal links are omitted.
Main entry points: router.py Input router · command_engine.py Command engine · voice_talk.py Voice modes · provider_manager.py Provider selection
Components 24
Interaction and routing
- CLI entry
main.py - Input router
router.py - Local responses
local_responses.py
Assistant intelligence
- AI brain
brain.py - Provider selection
provider_manager.py - Ollama provider
providers/ollama_provider.py - Groq provider
providers/groq_provider.py - Gemini provider
providers/gemini_provider.py
Persistent memory
- Memory manager
memory_manager.py - Memory and history
memory.py
Commands and skills
- Command engine
command_engine.py - Command aliases
aliases.py - App discovery
app_discovery.py - Calculator skill
skills/calculator.py - Reader skill
skills/reader.py
Voice interaction
- Voice modes
audio/voice_talk.py - Voice controls
audio/voice_processing.py - Speech input
audio/speech_input.py - Speech output
audio/speech_output.py
Other components
- User
- Model services
- Windows applications
- Web browser
- Speech devices
Connections 32
- User → CLI entry: enters text
- CLI entry → Input router: routes input
- CLI entry → Voice modes: starts voice mode
- Input router → Local responses: checks replies
- Input router → Memory manager: handles memory
- Input router → Command engine: detects commands
- Input router → Calculator skill: checks calculations
- Input router → AI brain: asks assistant
- AI brain → Provider selection: selects provider
- Provider selection → Ollama provider: dispatches requests
- Provider selection → Groq provider: dispatches requests
- Provider selection → Gemini provider: dispatches requests
- Ollama provider → Model services: calls model
- Groq provider → Model services: calls model
- Gemini provider → Model services: calls model
- Memory manager → Memory and history: reads and writes
- Memory manager → Model services: classifies facts
- Ollama provider → Memory and history: reads context
- Groq provider → Memory and history: reads context
- Command engine → Command aliases: normalizes targets
- Command engine → App discovery: finds or closes apps
- App discovery → Windows applications: controls apps
- Command engine → Web browser: opens or searches
- Command engine → Calculator skill: runs calculation
- Command engine → Reader skill: runs reader
- Reader skill → Speech output: speaks text
- Voice modes → Speech input: captures speech
- Speech input → Speech devices: uses recognition
- Voice modes → Input router: routes recognized input
- Voice modes → Voice controls: checks voice controls
- Voice modes → Speech output: speaks responses
- Speech output → Speech devices: synthesizes speech