Headless mode runs wiggum commands directly without the TUI. Use this for CI pipelines, scripting, and integration with other agents.
Commands
wiggum init [options] # Scan project, configure AI
wiggum new <name> [options] # Create feature spec
wiggum run <name> [options] # Run the Ralph loop
wiggum monitor <name> [options] # Monitor a running loop
wiggum sync # Refresh codebase context
wiggum config set <key> <value> # Manage settings
wiggum agent [options] # Autonomous backlog processing
When to use headless mode
- CI/CD pipelines — Run wiggum in automated workflows
- Agent orchestration — Let other AI agents invoke wiggum programmatically
- Scripting — Chain wiggum commands in shell scripts
- Non-interactive terminals — Environments without TTY support
In a TTY terminal, wiggum monitor and wiggum agent automatically open the TUI. Use --stream to force headless output.
Common headless workflows
Generate spec from GitHub issue
wiggum new my-feature --auto --goals "Add user notifications endpoint" --issue #42
The --auto flag skips the interactive interview and generates a spec directly from the goals and issue context.
Run a Ralph loop with auto-merge
wiggum run my-feature --worktree --review-mode merge
Process GitHub backlog in CI
wiggum agent --stream --review-mode merge --labels P0 --max-items 5
Full pipeline in one script
#!/bin/bash
wiggum init --provider anthropic --yes
wiggum new my-feature --auto --goals "Add dark mode toggle" --issue #42
wiggum run my-feature --worktree --review-mode merge
Pass options inline
Headless mode accepts all configuration via flags instead of interactive prompts:
wiggum init --provider anthropic --yes
wiggum new auth-flow --auto --goals "JWT auth middleware" --issue #42 --force
wiggum run auth-flow --worktree --cli codex --review-cli claude --max-iterations 15 --review-mode auto
wiggum agent --stream --labels P0,bug --max-items 3 --review-mode merge
See the commands reference for all available flags.