The Agent Skills Directory

[SAFE]: The skill implements a robust safety framework for autonomous agent operations, including circuit-breaking logic to halt loops that exhibit stagnation patterns or repetitive errors.
[SAFE]: It defines explicit file protection rules for sensitive configuration directories (such as .ralph/ and .claude/) and critical project files, effectively preventing the execution of destructive commands like rm or git clean on these paths.
[SAFE]: Rate limiting and timeout detection mechanisms are included to manage API usage and prevent resource exhaustion or infinite wait states.
[SAFE]: The instructions are strictly focused on operational safety and reliability, with no evidence of obfuscation, data exfiltration, or malicious prompt injection patterns.

circuit-breaker