How joaovitoramaral27/moth works
The tree suggests Moth is a desktop conversational assistant, with interface, orchestration, model inference, memory, actions, and voice components. It also includes tokenizer and model-training code, so those are shown as a separate development capability. The README is empty and the provided source excerpts contain no implementation text; relationships and runtime entry points therefore cannot be verified. The graph shows the named components without inferring call flows.
Components 28
Assistant app
- Application entry
main.py - Desktop window
interface/window.py - Chat interface
interface/chat.py - Settings interface
interface/settings.py - Assistant core
core/moth_core.py - Orchestrator
core/orchestrator.py - Conversation context
core/context.py - Personality
core/personality.py
Model runtime
- Inference engine
brain/inference/engine.py - Transformer model
brain/model/transformer.py - Attention
brain/model/attention.py - Model layers
brain/model/layers.py - Embeddings
brain/model/embeddings.py - Tokenizer
brain/tokenizer/tokenizer.py
Memory
- Memory database
memory/database.py - Long-term memory
memory/long_term.py - Short-term memory
memory/short_term.py - Moth database
storage/mothdb.db
Assistant capabilities
- App actions
actions/apps.py - File actions
actions/files.py - Reminders
actions/reminders.py - Wake phrase
voice/call_word.py - Speech input
voice/hearing.py - Voice output
voice/voice.py
Model development
- Training dataset
brain/training/dataset.py - Model trainer
brain/training/trainer.py - Model evaluation
brain/training/evaluation.py - Tokenizer training
brain/tokenizer/train_tokenizer.py