zero-trust-assessment
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes proactive defensive instructions to mitigate prompt injection. It explicitly directs the agent to identify and flag attempts to override instructions (e.g., 'ignore previous instructions') found in input data. The static analysis flag for instruction override is a false positive caused by these hardening measures.
- [DATA_EXFILTRATION]: The skill contains explicit prohibitions against exfiltrating sensitive data such as network topology, IP addresses, or security configurations identified during the review process.
- [COMMAND_EXECUTION]: The allowed-tools are restricted to 'Read', 'Grep', and 'Glob', which are appropriate for a read-only assessment skill. No dangerous subprocess spawning or arbitrary command execution was detected.
- [EXTERNAL_DOWNLOADS]: No external scripts, remote code execution patterns, or unverifiable dependencies were identified in the skill content.
- [INDIRECT_PROMPT_INJECTION]: The skill acknowledges the risk of processing untrusted architecture and configuration data. It implements boundary instructions and a 'read-only' posture to minimize the impact of instructions potentially embedded in the analyzed data.
Audit Metadata