dev-ralph

An iteration-based persistence loop repeatedly runs agent-driven debugging tasks with persisted state until tests pass or a circuit breaker halts the process. It carries forward previous failure context each iteration to avoid repeating mistakes.

A circuit breaker stops the agentic loop after 5 iterations or after encountering 3 identical failure fingerprints. This prevents infinite fix attempts and safely halts the workflow when progress stalls.

You need a predefined completion signal such as tests passing, a green build, or verifier approval. Explicit completion criteria ensure the persistence loop terminates for the right reason rather than stopping arbitrarily.

Yes, state persistence tracks what was tried and why it failed across each iteration. This allows targeted fixes for flaky CI scenarios and hands off the full state to an escalation agent if completion never arrives.

The circuit breaker halts the loop after 5 iterations or 3 identical failures, and the accumulated state from all attempts is preserved. This state can then be handed off to an escalation agent for further analysis.