Skip to main content
Launch wiggum with no arguments to open the interactive terminal UI:
wiggum
The TUI provides a persistent session with command autocomplete, live feedback, and a visual interface for every phase of feature development.

Available commands

CommandAliasDescription
/init/iScan project and configure AI provider
/new <name>/nCreate a feature spec via AI interview
/run <name>/rStart the autonomous Ralph loop
/monitor <name>/mMonitor a running feature loop
/issueBrowse GitHub issues in a columnar picker
/sync/sRe-scan project and update context
/config [args]Manage API keys and settings
/help/hShow available commands
/exit/qExit wiggum
Press Esc to cancel any operation or background a running loop.

Screens

Main shell

The default screen is a command shell with / prefix commands. Features:
  • Slash command autocomplete — Type / and see suggestions
  • Spec name autocomplete — Tab-complete spec names for /run
  • Background run tracking — See active loops in the header
  • Context age indicator — Shows when your codebase context was last updated

Init screen

Launched by /init. Walks through six phases with progress tracking:
  1. Scanning (tech stack detection)
  2. Provider selection
  3. API key input
  4. Model selection
  5. AI analysis (agentic codebase enrichment)
  6. File generation

Interview screen

Launched by /new <name>. The AI interview has five phases:
1

Context

Add reference URLs, files, or GitHub issues. Use the built-in issue picker to browse and select issues visually.
2

Goals

Describe what you want to build in plain language.
3

Interview

The AI asks up to 10 clarifying questions. Each question comes with structured answer options you can select, or type a custom answer. The AI uses codebase tools (file reading, pattern search, directory exploration) to ask project-specific questions.Say “done” at any point to skip remaining questions.
4

Generation

The AI synthesizes all context and produces a detailed markdown spec.
5

Complete

Shows a spec completion summary with overview, then saves to .ralph/specs/.

Run screen

Launched by /run <name>. Shows live progress while the Ralph loop executes:
  • Phase progress bars — Each phase (planning, implementation, E2E testing, verification, PR review) shows status with duration
  • Iteration counter — Current iteration out of max (e.g. 3/10)
  • Activity feed — Real-time stream of coding agent actions
  • Token tracking — Input/output token counts and cache metrics
  • Task checklist — Implementation plan tasks with completion status
  • Git branch info — Current branch and commit count
Background mode: Press Esc to background the loop and return to the shell. The loop keeps running. Re-attach with /monitor <name>. Completion summary: When done, shows phase durations, final iteration count, file changes, commit history, and PR link.

Agent screen

Launched by wiggum agent (not available as a TUI command — run from the CLI). Shows the autonomous orchestrator in action:
  • Backlog ranking and issue selection with reasoning
  • Real-time loop progress for the current issue
  • Completion outcomes for each processed issue

Issue picker

Launched by /issue or automatically during /new when using --issue. Shows a columnar table of GitHub issues with title, labels, and status. Select an issue to pre-load its context into the spec interview.

Keyboard shortcuts

KeyAction
EscCancel current operation / background running loop
TabAutocomplete command or spec name
EnterSubmit command or answer
Ctrl+CExit wiggum