keep-others-dependent
Warn
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: MEDIUMPROMPT_INJECTIONOBFUSCATIONMETADATA_POISONING
Full Analysis
- [PROMPT_INJECTION]: The skill uses detailed instructions to override the agent's default behavior, forcing it to adopt a persona that intentionally creates unmaintainable and overly complex code to ensure "job security" by making deliverables a "hostage situation.".
- [OBFUSCATION]: The instructions explicitly mandate the use of single-letter variable names, invented acronyms (e.g.,
_rxa,kfn_ctx,__ΩΩ), and the avoidance of comments or type hints to prevent "knowledge leaks.". - [METADATA_POISONING]: The skill's description is deceptive, framing the generation of unmaintainable code and technical debt as "enterprise-grade, future-proof code."
Audit Metadata