academic-paper-reviewer
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process academic manuscripts provided by users, which constitutes an attack surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context during Phase 0 when the user provides a paper via text paste or file attachment for review.
- Boundary markers: The instructions for the agents do not currently implement explicit boundary markers or delimiters to isolate the paper's content from the system instructions.
- Capability inventory: The agents are limited to generating structured review reports and Socratic dialogue based on templates. There are no capabilities for network operations, file writing, or system modifications.
- Sanitization: There is no evidence of input sanitization or filtering applied to the paper content before it is processed by the agent team.
Audit Metadata