The Agent Skills Directory

[SAFE]: The skill is entirely instructional and does not include any executable code, network operations, or external dependencies. It follows a rigorous pedagogical approach to refactoring, explicitly requiring 'characterization tests' before any changes are made. It also includes specific security considerations, instructing the agent to ensure that tests cover both 'allowed and blocked behavior' when refactoring code involving sign-in, permissions, or secrets. No malicious patterns such as prompt injection, data exfiltration, or obfuscation were detected.

sensei-prune