analyzing-research-papers
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process content from various untrusted external sources, creating a potential vector for malicious instructions to be embedded in the data the agent is analyzing.
- Ingestion points: The agent is instructed to access and read content from local file paths, DOI resolution services (doi.org), and remote URLs (arXiv, bioRxiv, and journal websites) as defined in the 'Paper Access Methods' section of SKILL.md.
- Boundary markers: The skill does not provide specific instructions to the agent to treat external paper content as untrusted or include markers to prevent the agent from following instructions found within those documents.
- Capability inventory: The instructions assume the agent has permissions to access the local file system and perform network requests to retrieve document content.
- Sanitization: There are no requirements defined for the agent to sanitize or validate the content extracted from papers before summarizing them.
Audit Metadata