The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): No attempts to override the agent's core instructions or bypass safety filters were detected. The skill is instructional and focused on coding standards.\n- [DATA_EXFILTRATION] (SAFE): No hardcoded secrets, sensitive path access, or unauthorized network operations were found. The code examples use placeholder data.\n- [REMOTE_CODE_EXECUTION] (SAFE): No remote scripts or execution of untrusted external code were found. The skill mentions common, reputable Ruby gems (e.g., Sidekiq, Bullet, RSpec) as standard dependencies.\n- [COMMAND_EXECUTION] (SAFE): No evidence of unauthorized shell command execution or privilege escalation attempts.\n- [INDIRECT_PROMPT_INJECTION] (SAFE): While the skill is designed to process and review user-provided code (a potential injection surface), it lacks the exploitable capabilities (such as network or filesystem write access) required to fulfill a malicious payload.

ruby-conventions