ralph-wiggum
Interactive self-referential AI loops that rerun the same task until completion.
🛡️ AgentReady threat assessment
MAESTRO 7-layer threat model + OWASP AIVSS risk score for ralph-wiggum, derived from its capabilities.
AIVSS 8.0 · High
View MAESTRO 7-layer threat model →Overview
ralph-wiggum is an official Anthropic plugin implementing self-referential iterative loops: Claude works the same task repeatedly, seeing its prior work each pass, until the task is done. It drives an autonomous retry loop over a single objective.
Key features
- Self-referential task loops
- Sees prior work each iteration
- Runs until completion
Use cases
- Iterate on a hard task autonomously
- Let the agent refine work over passes