Skip to content

foryourhealth111-pixel/Vibe-Skills / bundled/skills/ralph-loop

ralph-loop

Verified

Codex-compatible Ralph loop runner with dual engines (compat local state loop + optional open-ralph-wiggum backend).

ralph-loop

This is a Codex-oriented ralph-loop command with two execution engines.

Engine model

  • Keeps the same command name: ralph-loop.
  • Keeps the same default state file format: .claude/ralph-loop.local.md.
  • Default compat engine keeps local-state semantics and manual --next.
  • Optional open engine delegates to external open-ralph-wiggum CLI for auto-iteration.

Script

  • Script path: scripts/ralph-loop.ps1

Usage

$codexHome = if ($env:CODEX_HOME) { $env:CODEX_HOME } else { Join-Path $HOME '.codex' }
$script = Join-Path $codexHome 'skills/ralph-loop/scripts/ralph-loop.ps1'

# Start a local compat loop
powershell -ExecutionPolicy Bypass -File $script Build a todo API --max-iterations 20 --completion-promise DONE

# Move to the next iteration manually
powershell -ExecutionPolicy Bypass -File $script --next

# Show current loop state
powershell -ExecutionPolicy Bypass -File $script --status

# Force restart with a new prompt
powershell -ExecutionPolicy Bypass -File $script New prompt --max-iterations 10 --force

# Use open-ralph-wiggum backend (auto loop, defaults to --agent codex and --no-commit)
powershell -ExecutionPolicy Bypass -File $script --engine open Build a todo API --max-iterations 20 --completion-promise DONE

Vibe compatibility

  • Safe in /vibe routed sessions as a direct execution tool.
  • Does not force multi-agent orchestration.
  • Keeps command names stable for unified memory and invocation.
  • open engine remains mutually exclusive with active XL team orchestration.

Notes

  • Compat mode: if max_iterations is reached, the local state file is removed automatically.
  • Compat mode: completion promises are tracked in local state.
  • Open mode: --next, --force, --state-file, --stop are not available (managed by external ralph CLI semantics).
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall