test-parallel-sdk
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected in the skill instructions.
- [DATA_EXPOSURE]: The skill counts files in the
.claude/skills/andscripts/directories. This activity is limited to metadata (file counts) rather than reading sensitive file content or credentials. - [INDIRECT_PROMPT_INJECTION]: The skill uses
$ARGUMENTSto dynamically name an output file. While this processes user-controlled input, it is used in a standard instructional manner for parameterization without evidence of bypass or override attempts.
Audit Metadata