Ralph Loop

SkillProductivity

Long-running iterative development loops. Run arbitrarily-long tasks without diluting model attention. Triggers: ralph, ralph loop, iterative loop, long-running task, development loop.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Ralph Loop skill

What this skill tells your AI

The instructions your AI receives, as published by neuron-mr-white/unipi in packages/ralph/SKILL.md and read by ahel’s review.

Long-running iterative development loops. Run complex tasks across multiple iterations with reflection checkpoints.

Commands

CommandDescription
/unipi:ralph start <name> [options]Start a new loop
/unipi:ralph stopPause current loop
/unipi:ralph resume <name>Resume a paused loop
/unipi:ralph statusShow all loops
/unipi:ralph cancel <name>Delete loop state
/unipi:ralph archive <name>Move loop to archive
/unipi:ralph clean [--all]Clean completed loops
/unipi:ralph list --archivedShow archived loops
/unipi:ralph nuke [--yes]Delete all ralph data

Arguments autocomplete: typing /unipi:ralph suggests subcommands with descriptions; resume/cancel/archive suggest loop names with live status; start suggests its flags.

Tools

ToolDescription
ralph_startStart a ralph loop (LLM-callable)
ralph_doneSignal iteration complete, request next

Options

  • --max-iterations N — Stop after N iterations (default: 50)
  • --items-per-iteration N — Process N items per iteration
  • --reflect-every N — Reflection checkpoint every N iterations

How It Works

  1. Start — Create a task file with goals and checklist
  2. Iterate — LLM works on task, updates checklist, calls ralph_done
  3. Reflect — Every N iterations, pause and assess progress
  4. Complete — When done, LLM outputs COMPLETE marker

Integration with Workflow

When @unipi/workflow is present:

  • /unipi:work suggests ralph for large tasks
  • Ralph loops emit events that workflow can track
  • Ralph state is available to info-screen (when present)

Task File Format

# Task

{Description}

## Goals
- Goal 1
- Goal 2

## Checklist
- [ ] Item 1
- [ ] Item 2
- [x] Completed item

## Notes
{Progress notes}

State

Loop state stored in .unipi/ralph/ directory:

  • .unipi/ralph/{name}.md — Task file
  • .unipi/ralph/{name}.state.json — Loop state
  • .unipi/ralph/archive/ — Completed loops

Tips

  • Use --items-per-iteration for large checklists
  • Use --reflect-every 10 for long-running tasks
  • Press ESC to pause, send message to resume
  • Task file is updated each iteration — check it for progress

Signals

GitHub stars
64
Forks
15
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
ralph-neuron-mr-white
Source
github.com/neuron-mr-white/unipi