reference-manager-interop
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external bibliographic data files (RIS, BibTeX, CSL-JSON). While these files could contain malicious instructions, the conversion script treats the data as passive text and does not execute or evaluate it.
- Ingestion points:
args.input.read_text()inscripts/convert_refs.py. - Boundary markers: None.
- Capability inventory: File system read and write capabilities using the
pathlibmodule. - Sanitization: Basic string normalization is performed for citation key generation, but no specific sanitization against prompt injection is implemented.
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected. All conversion operations are local, and references to external style repositories target well-known, established community resources.
Audit Metadata