ralph-loop
Interactive self-referential AI loops (the Ralph Wiggum technique) for iterative development until completion.
🛡️ AgentReady threat assessment
MAESTRO 7-layer threat model + OWASP AIVSS risk score for ralph-loop, derived from its capabilities.
AIVSS 8.6 · High
View MAESTRO 7-layer threat model →Overview
Anthropic's official ralph-loop plugin. It implements the Ralph Wiggum technique — Claude repeatedly works on the same task, seeing its previous work each iteration, until the task is complete. Surface is a command that re-invokes the model in a loop over accumulating context.
Key features
- Self-referential iteration loop
- Model sees its prior work each pass
- Runs until task completion
Use cases
- Grind a stubborn task to completion iteratively
- Long-running refactors that need repeated passes