Terminal CLI Commands and Shortcuts
Reference for Andromity terminal commands, slash commands, headless andromity run options, background processes, and keyboard shortcuts.
Andromity can be driven interactively via the chat bar (in VS Code or the TUI) or headlessly via scripted CLI commands.
Interactive Slash Commands
Type these commands into the terminal app's prompt bar. The VS Code chat has its own set, which shares many of these (/model, /mode, /trust, /untrust, /undo, /compact, /new, /clear, /sessions, /cron, /skills, /settings); type / in the chat to see them all.
| Command | Shortcut | Functionality |
|---|---|---|
/model | Ctrl + L | Open the model switcher modal to change provider or LLM |
/profile [name] | Ctrl + U | Switch active profile: builder, coder, reviewer, planner |
/mode [safe|trust|full|yolo] | — | Set permission mode for the current session |
/trust | — | Grant full trust to the current folder |
/untrust | — | Remove trust from the current folder |
/sessions | Ctrl + O | Open interactive session browser to switch tasks |
/new | — | Start a clean session in the current workspace |
/rename <name> | — | Assign a memorable title to the active session |
/undo | — | One-click rollback: reverts the last turn and all file changes |
/dry-run | — | Toggle dry-run mode — simulates tool executions without writing to disk |
/debug | — | Toggle real-time inline debugging of tool calls and parameters |
/mcp | — | Display the status of configured MCP servers and their tools |
/skills | — | Browse, install and remove skills |
/subagent [set <role> <model> [provider]] | — | List subagent roles, or give a role its own model |
/cron | — | Open the background task scheduler modal |
/compact | — | Summarize and prune old turns to conserve context window tokens |
/clear | — | Clear current conversation display buffer |
/keys | — | Show which providers have an API key |
/keys set <provider> <key> | — | Save an API key to config.toml (stored in plain text) |
/settings | Ctrl + E | Open configuration and preferences modal |
/logs | — | Print absolute path to agent debug log files |
Headless CLI Commands (andromity run)
You can run Andromity non-interactively in shell scripts, pre-commit hooks, and CI/CD pipelines:
andromity run "Refactor database queries in src/db.py to use prepared statements"
andromity run "Fix linting errors reported by ruff" --yes
andromity run "Upgrade dependencies in pyproject.toml" --dry-run
andromity run "Audit src/security/ for OWASP vulnerabilities" \
--provider anthropic --model claude-sonnet-4-6 \
--profile reviewerBackground Process System (shell_bg)
Andromity features a non-blocking process manager for long-running processes such as development servers (npm run dev), test suites, or Docker containers:
- Non-Blocking Launch: The agent launches commands in the background without hanging the prompt loop.
- Interactive UI Controls: Running background processes display a live status badge in VS Code with a [⏹ Stop] button.
- Reactive Auto-Wake: When a background command exits or crashes, Andromity automatically wakes up with the exit code, duration, and output logs to analyze failures and fix bugs.
- Tree Teardown (
shell_kill): Terminating a background task safely kills the entire child process group tree across Windows (taskkill /F /T) and Unix (SIGTERM->SIGKILL), preventing orphan port locks.
Configuration Reference
Reference for Andromity's config.toml (every section and key), the files Andromity keeps on disk, and the environment variables it reads.
Schedule AI Coding Tasks with Cron
Run Andromity on a schedule with /cron to execute tests, fix failures, and refactor code while you are away.