nuke-quiet
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill mandates a significant override of the agent's default conversational behavior, introducing an 'emission gate' that suppresses narration and enforces a strict, rigid hand-back contract for all tasks. This behavioral modification is intended to restrict agent communication patterns during task execution.
- [INDIRECT_PROMPT_INJECTION]: The 'silent' grade minimizes agent output during tasks, creating a surface where the user has limited visibility into the agent's reasoning process. This could prevent the detection of malicious instructions processed from external data, although safety reporting is explicitly required by the instructions.
- Ingestion points: The skill applies to all data and instructions processed during an active agent session (SKILL.md).
- Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded directives in external content.
- Capability inventory: The skill is part of a larger suite intended to chain with executors that handle file system modifications and shell command execution (references/family-map.md).
- Sanitization: There are no specified mechanisms for sanitizing or validating external content before it influences agent output or behavior.
Audit Metadata