disaster-recovery-testing
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No evidence of instructions attempting to bypass safety filters, jailbreak the model, or override system prompts. The skill uses structured prompts to maintain analytical boundaries.
- [DATA_EXPOSURE]: The skill explicitly forbids reading credentials or accessing production systems. There are no hardcoded secrets or sensitive file path references (e.g., .ssh, .env) in the provided files.
- [REMOTE_CODE_EXECUTION]: No patterns involving external script downloads (curl, wget) or execution of remote code (eval, exec) were found.
- [COMMAND_EXECUTION]: There are no shell command execution patterns or dynamic context injection strings (
!command) detected in the skill definition or reference files. - [INDIRECT_PROMPT_INJECTION]: The skill processes external information regarding disaster scenarios. It includes a robust 'Input Audit' framework (classifying data as known, missing, stale, or assumptions) and strictly requires separating facts from inferences, which effectively mitigates the impact of potential instructions embedded in the input data.
- [OBFUSCATION]: No obfuscated strings, hidden Unicode characters, or Base64-encoded commands were detected.
Audit Metadata