literature-review-builder
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted research data from external sources.
- Ingestion points: Untrusted data enters the agent context through search snippets, abstracts, and full-text documents during the bibliography verification, screening, and evidence extraction phases described in SKILL.md.
- Boundary markers: The instructions do not mandate the use of specific delimiters or warnings to ignore instructions embedded in the literature, which could allow adversarial text in a paper to influence the agent's behavior.
- Capability inventory: The skill is scoped to research and synthesis; it does not grant the agent capabilities to execute arbitrary code or perform unauthorized network exfiltration based on paper content.
- Sanitization: There are no instructions provided for sanitizing or filtering data ingested from scientific papers before it is analyzed by the agent.
Audit Metadata