literature-review-generator
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to dynamically write and execute Python scripts using the Pandas library to process and filter literature metadata (e.g., generating
prisma_2_screened.csvfromprisma_1_identification.csv). - [EXTERNAL_DOWNLOADS]: The instructions mandate using the Python
requestslibrary to fetch metadata and open-access documents from external academic services including OpenAlex, arXiv, PubMed, bioRxiv, and EuropePMC. These are recognized well-known academic services for research data. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of ingesting and processing untrusted data from external academic sources.
- Ingestion points: Metadata (Titles, Abstracts) and full-text content retrieved via APIs from OpenAlex, ArXiv, and PubMed as described in the PRISMA workflow section of
SKILL.md. - Boundary markers: The skill does not define explicit boundary markers or instructions to ignore embedded commands within the fetched research data.
- Capability inventory: Includes local file system writes (CSV checkpoints), network access via Python
requests, and the ability to generate and execute Python code. - Sanitization: No specific sanitization or filtering logic is prescribed for the external text before it is processed by the agent or passed to secondary scripts.
Audit Metadata