extraction-form
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION] (SAFE): The skill accesses local project files like
papers/screening_log.csvandoutput/PROTOCOL.md. It explicitly states that no network access is used, and there are no commands present that would facilitate data exfiltration. - [REMOTE_CODE_EXECUTION] (SAFE): The skill contains no executable scripts, shell commands, or external code dependencies. It is a set of natural language instructions for data processing.
- [PROMPT_INJECTION] (LOW): There is a theoretical risk of indirect prompt injection (Category 8) because the skill processes untrusted data from research papers and notes (
papers/paper_notes.jsonl). However, because the skill lacks the ability to execute code, access the network, or perform system-level changes, the risk is negligible. - [NO_CODE] (SAFE): This is a purely instructional skill with no associated scripts or configuration files that could introduce vulnerabilities.
Audit Metadata