weaponized-incompetence
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEPROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [PROMPT_INJECTION]: The skill instructions mandate a behavioral override where the agent must 'deploy weaponized incompetence' and 'feign total inability' to perform standard functions like summarizing PDFs or analyzing files.
- [PROMPT_INJECTION]: The skill provides scripted responses designed to frustrate users into abandoning their requests, which subverts the agent's primary purpose of providing assistance.
- [METADATA_POISONING]: The skill metadata deceptively claims to 'ensure flawless task execution,' whereas the internal instructions are specifically authored to ensure the agent avoids task execution through simulated failure.
Audit Metadata