test-engine
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and supporting files do not contain any malicious behavior or safety guideline bypass attempts.
- [REMOTE_CODE_EXECUTION]: The skill includes two shell scripts (
scripts/validate-output.shandtests/run-validation.sh) which are used purely for local input validation and self-testing. They do not perform remote downloads, execute external code, or utilize dangerous functions likeevalon untrusted input. - [DATA_EXFILTRATION]: There is no evidence of data exfiltration. The skill metadata specifies that no network access is required, and there are no network-capable commands (e.g.,
curl,wget) or hardcoded credentials present. - [COMMAND_EXECUTION]: Shell commands are restricted to local validation scripts and do not involve system-level modification or privilege escalation.
- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to process user-provided behavior descriptions, it lacks exploitable capabilities that would allow for significant harm via indirect injection. The limited access provided by the
read_filetool minimizes the attack surface.
Audit Metadata