The Agent Skills Directory

[SAFE]: The skill consists entirely of behavioral guidelines and technical principles for software development. No executable code, scripts, network operations, or file system access patterns were identified within the content.
[PROMPT_INJECTION]: The skill uses strong, assertive language to define a specific persona ('Linus Torvalds') and coding philosophy. While it includes 'Standard Rejection Phrases' that are abrasive (e.g., 'This is bogus shit'), it explicitly instructs the AI to use blunt technical language and not turn it into personal abuse, which aligns with behavioral steering rather than a malicious safety bypass attempt.
[INDIRECT_PROMPT_INJECTION]: As the skill defines a review process for code, it theoretically interacts with untrusted input (the code being reviewed). However, it does not provide any mechanism for data exfiltration or automated command execution based on that input. The risk is limited to the agent adopting a specific tone when responding to user-provided code.

torvalds-doctrine