What happens when a task reaches its maximum retry limit?

If a task hits the iteration limit without meeting completion criteria, it is moved to the '/vault/Failed/' directory and logged for manual review.

Can I manually stop an autonomous task loop?

Yes, the skill supports an emergency stop capability via a specific trigger file (.ralph_stop) or through manual admin overrides.

Does this skill prevent infinite loops?

Yes, it includes built-in safety controls including configurable maximum iteration limits per task and a global iteration cap to prevent runaway processes.

What is the primary purpose of the Ralph Wiggum Loop?

It ensures Claude Code persists and continues working until a task meets specific completion criteria, preventing the model from exiting before the work is actually finished.

How does the skill detect if a task is completed?

The system monitors the file system, specifically looking for task files to be moved to the '/vault/Done/' folder or for the presence of 'status: completed' markers.

Ralph Wiggum Autonomous Loop

Name: Ralph Wiggum Autonomous Loop
Author: NaveedTechLab

byNaveedTechLab

0•

Productivity & Workflow

Implements an autonomous task persistence system that ensures Claude completes complex workflows through state tracking and continuation prompts.

The Ralph Wiggum Loop is a robust automation skill designed for the Personal AI Employee framework to ensure long-running tasks reach full completion without human intervention. By utilizing a SQLite-based state tracker and a specialized vault folder structure, this skill intercepts premature exits and re-injects continuation prompts until specific success criteria are met. It provides a bridge between simple command execution and true autonomous agency, offering fine-grained control over iteration limits, safety boundaries, and detailed audit trails for every task Claude performs.

Key Features

01Integrated emergency stop and detailed audit logging

02Automatic continuation prompt injection for incomplete workflows

03Folder-based completion detection via vault integration

04Configurable safety limits for global and per-task iterations

050 GitHub stars

06SQLite-based task state tracking and persistence

Use Cases

01Automating multi-step data processing tasks that require persistent agent activity

02Managing complex file-system workflows within a structured vault environment

03Ensuring reliability for autonomous operations that span across multiple Claude sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add naveedtechlab/personal-ai-employee-hackathon-0- ralph-wiggum-loop

For use in Claude.ai and ChatGPT

Key Features

01Integrated emergency stop and detailed audit logging

02Automatic continuation prompt injection for incomplete workflows

03Folder-based completion detection via vault integration

04Configurable safety limits for global and per-task iterations

050 GitHub stars

06SQLite-based task state tracking and persistence

Use Cases

01Automating multi-step data processing tasks that require persistent agent activity

02Managing complex file-system workflows within a structured vault environment

03Ensuring reliability for autonomous operations that span across multiple Claude sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add naveedtechlab/personal-ai-employee-hackathon-0- ralph-wiggum-loop

For use in Claude.ai and ChatGPT