Commands Reference
Quantish Agent supports both CLI arguments and interactive slash commands.
CLI Commands
quantish
Start interactive chat mode. This is the main way to use the agent.
quantish init
Run the setup wizard to configure API keys and create your wallet.
quantish config
View your current configuration (keys are masked).
quantish config --export
Export configuration as environment variables for your bots.
quantish tools
List all available tools (local + MCP trading tools).
quantish -p "your prompt"
Run a single prompt in one-shot mode. Useful for scripts.
CLI Options
| Option | Description |
|---|---|
-p, --prompt |
Run a single prompt and exit |
-v, --verbose |
Show detailed tool call information |
--no-mcp |
Disable trading tools (MCP) |
--no-local |
Disable local tools (filesystem, shell) |
--version |
Show version number |
--help |
Show help message |
Interactive Slash Commands
While in interactive mode, use these commands:
/help
Show all available slash commands and keyboard shortcuts.
/clear
Clear conversation history and reset token count. Start fresh.
/compact
Summarize the conversation to save tokens. Uses Claude to create a concise summary while preserving key information. Use when token count gets high.
/model [name]
Switch AI model. Options:
opus, sonnet, haiku (Anthropic) or glm, minimax, deepseek (OpenRouter). Without argument, shows current model and available options./provider [name]
Switch LLM provider. Options:
anthropic, openrouter. Without argument, shows current provider./cost
Show detailed cost breakdown: session total, input/output tokens, cache usage.
/tools
List all available tools (shortcut to CLI command).
/config
Show configuration info.
Session Commands
/save [name]
Save current conversation session with optional name.
/resume
Resume the last saved session.
/sessions
List all saved sessions.
/load [id]
Load a specific saved session by its ID.
/forget
Delete all saved sessions.
Process Commands
/processes
List running background processes with their IDs and uptime.
/stop [id]
Stop a background process by its ID.
/stopall
Stop all running background processes.
/exit
Exit the agent (also stops all background processes).
Keyboard Shortcuts
| Shortcut | Action |
|---|---|
Enter |
Send message (or queue if agent is working) |
Escape |
Interrupt current generation (sends queued message if any) |
Backspace |
Un-queue message (move back to input) when input is empty |
Ctrl+C |
Exit the agent (stops all processes) |
Queued Input
You can type while the agent is working. Your message will be queued and shown in the status bar:
📥 Queued: your next message...The queued message will automatically send when the current operation completes. Press Escape to interrupt and send immediately, or Backspace on empty input to un-queue it.
Model Selection
Anthropic Models (Claude)
| Model | Input | Output | Best For |
|---|---|---|---|
opus |
$5/MTok | $25/MTok | Complex reasoning, creative tasks |
sonnet |
$3/MTok | $15/MTok | Balanced performance (Anthropic default) |
haiku |
$1/MTok | $5/MTok | Simple tasks, high volume |
OpenRouter Models
| Model | Input | Output | Best For |
|---|---|---|---|
glm |
$0.40/MTok | $1.50/MTok | Coding, reasoning (OpenRouter default) |
minimax |
$0.15/MTok | $0.15/MTok | Fast, cheap, long context |
deepseek |
$0.27/MTok | $1.10/MTok | Reasoning, analysis |
You:
/provider openrouter
✓ Switched to OpenRouter
You:
/model glm
✓ Switched to GLM-4.7
Cost Tracking
The status bar shows real-time cost and token usage:
↵ Send • Esc interrupt • /help commands $0.0518 • ~1.9k tokens • ✓ ReadyUse /cost for detailed breakdown:
You:
/cost
💰 Session Cost: $0.0518
Token Usage (current context):
Input: 1.9k tokens
Output: 847 tokens
Cache Write: 0 tokens
Cache Read: 1.2k tokens
💡 Tip: Use /model haiku for cheaper operations