alterlab-citation-mgmt
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill interacts exclusively with well-known and trusted academic data providers, including CrossRef (api.crossref.org), NCBI PubMed (ncbi.nlm.nih.gov), and arXiv (arxiv.org). All network operations are aligned with the skill's stated purpose of citation retrieval.
- [PROMPT_INJECTION]: The skill processes external, untrusted metadata from academic databases, which introduces a potential surface for indirect prompt injection.
- Ingestion points: Paper titles, authors, and abstracts are fetched from external APIs in scripts/extract_metadata.py, scripts/search_google_scholar.py, and scripts/search_pubmed.py.
- Boundary markers: Not present in the generated BibTeX or JSON outputs.
- Capability inventory: The skill has Bash access (allowed-tools) and file-writing capabilities (scripts/format_bibtex.py, scripts/validate_citations.py).
- Sanitization: Metadata is formatted for BibTeX compatibility (e.g., escaping capitalization in titles), but there is no explicit sanitization to prevent the ingestion of malicious instructions embedded in paper metadata from influencing the agent's logic.
Audit Metadata