citation-management
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by fetching and processing metadata from external academic sources such as Google Scholar, PubMed, and CrossRef.\n
- Ingestion points: External data ingested via
scripts/search_google_scholar.py,scripts/search_pubmed.py, andscripts/extract_metadata.py(SKILL.md, references/detailed-guide.md).\n - Capability inventory: The skill uses command execution to run Python scripts and performs file system operations to write output data (SKILL.md).\n
- Boundary markers: There are no explicit markers or instructions to isolate or ignore potentially adversarial content (e.g., malicious instructions hidden in paper titles or abstracts) within the fetched data.\n
- Sanitization: No sanitization or validation of the fetched external text content is mentioned before it is processed by the agent.\n- [COMMAND_EXECUTION]: The skill workflow involves executing multiple Python scripts from a local
scripts/directory to manage and validate BibTeX data (SKILL.md).
Audit Metadata