academic-paper-verify
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's functionality is entirely consistent with its described purpose of academic research verification and computational replication.
- [COMMAND_EXECUTION]: The skill is designed to execute local R scripts (using the
source()function) to replicate study results. It generates a dedicated test suite (tests/verify_replication.R) based on a provided template to automate this verification process. This behavior is expected and necessary for the skill's primary function. - [DATA_EXFILTRATION]: The skill performs extensive read operations on local files, including various data formats (CSV, RDS, DTA, XLSX), code files (R, Rmd), and LaTeX source files. These operations are restricted to the local file system for auditing purposes; no network operations or external data transmissions were identified.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted research data and code files. However, the risk is inherent to the audit use case and is mitigated by the structured verification manifest and the transparent execution model described in the instructions.
Audit Metadata