lit-synthesizer
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is exposed to indirect prompt injection via external data ingestion.
- Ingestion points: External research data is fetched from PubMed (via Entrez API) and bioRxiv/medRxiv APIs.
- Boundary markers: No specific delimiters or instructions are defined to separate ingested text from agent instructions.
- Capability inventory: The agent performs text summarization and generates structured literature reviews based on untrusted external data.
- Sanitization: No sanitization or validation of the external content is performed before LLM processing.
Audit Metadata