paper-review
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process academic manuscripts and related artifacts, which are untrusted external inputs. This creates a surface for indirect prompt injection if a manuscript contains instructions aimed at overriding the agent's behavior.
- Ingestion points: Manuscripts, companion artifacts, prior drafts, and author responses (SKILL.md, round-calibration.md).
- Boundary markers: The instructions do not define explicit structural delimiters or isolation protocols to separate manuscript content from the reviewer instructions.
- Capability inventory: The skill utilizes tools for text extraction, adding comments, creating tracked changes, and visual rendering (SKILL.md).
- Sanitization: There is no mention of input sanitization or validation for the content being reviewed.
Audit Metadata