skill-review
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No instructions to bypass safety guidelines or override system behavior were found. The skill's logic is focused entirely on validation tasks.
- Data Exposure & Exfiltration (SAFE): No hardcoded credentials or external network operations were detected. The skill only reads local skill-related files for analysis.
- Obfuscation (SAFE): No encoded content or hidden characters were used to obscure the skill's logic.
- Unverifiable Dependencies & Remote Code Execution (SAFE): The skill does not download or execute remote scripts. Mentioned packages in the documentation are for illustrative purposes and not part of the skill's dependencies.
- Command Execution (SAFE): The use of 'wc -l' is a standard, low-risk operation for counting lines in text files.
- Privilege Escalation (SAFE): No commands for escalating user privileges or modifying system configurations were found.
- Persistence Mechanisms (SAFE): The skill does not attempt to create cron jobs, modify startup scripts, or maintain access across sessions.
- Metadata Poisoning (SAFE): The skill name and description are accurate and do not contain deceptive instructions.
- Indirect Prompt Injection (SAFE): The skill analyzes external skill files but limits its operations to static review and line counting. (Ingestion points: skills/ directory files; Boundary markers: None; Capability: wc -l; Sanitization: None).
- Dynamic Execution (SAFE): No runtime code generation, compilation, or unsafe deserialization of data was detected.
Audit Metadata