Commands Reference

April 29, 2026 · View on GitHub

Agent CLI provides multiple commands, each designed for a specific purpose.

Voice & Audio Commands

Command	Purpose	Use Case
`transcribe`	Speech-to-text	Record voice → get text in clipboard
`transcribe-live`	Continuous transcription	Background service with VAD
`diarize-live-session`	Retroactive diarization	Label speakers in saved `transcribe-live` sessions
`speakers`	Speaker profiles	List, rename, and merge persistent diarization identities
`speak`	Text-to-speech	Read text aloud
`voice-edit`	Voice-powered editor	Edit clipboard text with voice commands
`assistant`	Wake word assistant	Hands-free voice interaction
`chat`	Conversational AI	Full-featured voice chat with tools

Text Processing Commands

Command	Purpose	Use Case
`autocorrect`	Grammar & spelling	Fix text from clipboard

AI Services Commands

Command	Purpose	Use Case
`rag-proxy`	RAG server	Chat with your documents
`memory`	Long-term memory	Persistent conversation memory
`server`	ASR & TTS servers	Local Whisper and Kokoro/Piper with TTL-based memory management
`daemon`	Service manager	Install/manage servers as system daemons

The server command provides local ASR (speech-to-text) and TTS (text-to-speech) servers with unique advantages over standalone alternatives:

Dual-protocol - Both OpenAI-compatible API and Wyoming protocol from the same server
TTL-based memory management - Models load on-demand and unload after idle periods, freeing RAM/VRAM
Multi-platform acceleration - MLX Whisper on Apple Silicon, Faster Whisper on Linux/CUDA
Unified configuration - Consistent CLI, environment variables, and Docker setup

Installation Commands

These commands help set up Agent CLI and its services:

Command	Purpose
`install-services`	Install all AI services (Ollama, Whisper, Piper, OpenWakeWord)
`install-hotkeys`	Set up system-wide hotkeys
`install-extras`	Install optional Python dependencies (rag, memory, vad, etc.)
`start-services`	Start all services in a Zellij terminal session

Development Commands

Command	Purpose	Use Case
`dev`	Git worktree manager	Parallel development with AI agents

Configuration Commands

Command	Purpose
`config`	Manage configuration (init, show, edit)

Configuration - Config file keys and defaults
Architecture - How the system fits together

Common Options

Most commands support these options (audio/text agents and servers). Installation and config commands have their own flags. Use agent-cli <command> --help to see the exact options.

Option	Description
`--help`, `-h`	Show help for the command
`--config PATH`	Use a specific config file
`--log-level LEVEL`	Set logging level (DEBUG, INFO, WARNING, ERROR)
`--log-file PATH`	Write logs to a file
`--quiet`, `-q`	Suppress console output
`--print-args`	Show resolved arguments including config values

Provider Options

Most commands support multiple providers:

LLM Providers (`--llm-provider`)

ollama - Local LLM via Ollama (default)
openai - OpenAI-compatible API
gemini - Google Gemini API

ASR Providers (`--asr-provider`)

wyoming - Local Whisper via Wyoming (default)
openai - OpenAI-compatible Whisper API
gemini - Google Gemini API

TTS Providers (`--tts-provider`)

wyoming - Local TTS via Wyoming protocol (Kokoro or Piper, default)
openai - OpenAI-compatible TTS API
kokoro - Local Kokoro TTS (direct, without Wyoming)
gemini - Google Gemini TTS API

Process Management

Commands with background capabilities support:

Option	Description
`--stop`	Stop a running background process
`--status`	Check if a background process is running
`--toggle`	Toggle the background process on/off

Example:

# Start transcription in background
agent-cli transcribe &

# Check status
agent-cli transcribe --status

# Stop it
agent-cli transcribe --stop