paper-reviewer
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill provides structured guidelines and a Python-based checking tool for managing manuscript reviews and responses. The included script
scripts/check_coverage.pyuses only standard Python libraries and performs local text analysis without any network access, file writing, or dynamic execution capabilities.- [INDIRECT_PROMPT_INJECTION]: The skill processes manuscript text and referee reports from external sources (entered via modes described inSKILL.mdor as files for thecheck_coverage.pyscript). It incorporates robust boundary markers by mandating the use of verbatim blockquotes for referee comments and utilizing HTML machine tags for metadata. Thecheck_coverage.pyscript includes a normalization function to sanitize text for comparison, and the skill possesses no capabilities for unauthorized file system modifications or network communication, minimizing the risk of potential malicious instructions in the processed data.- [EXTERNAL_DOWNLOADS]: The documentation contains references to numerous official scientific reporting standards and journal policy pages from established organizations such as Nature Portfolio, CONSORT, and ARRIVE. These URLs are provided for informational purposes and do not trigger any automated downloads or executions.
Audit Metadata