alterlab-rma-literature-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external academic databases and project files, creating a potential surface for indirect prompt injection attacks.
- Ingestion points: Web search results from academic databases (e.g., Google Scholar, JSTOR, Scopus) and existing project-related markdown files.
- Boundary markers: The instructions do not define specific delimiters or instructions to prevent the agent from following prompts potentially hidden within academic articles or research notes.
- Capability inventory: Web search functionality and the ability to read and write markdown files (
.md) to the local project directory. - Sanitization: The skill lacks explicit sanitization steps for processing untrusted text retrieved from the web or project files before it is integrated into the final synthesis.
Audit Metadata