bids

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The script scripts/update_schema.py is configured to download the BIDS schema and Extension Proposals list from official project infrastructure, including bids-specification.readthedocs.io and raw.githubusercontent.com/bids-standard/bids-website. These represent the primary, trusted sources for the BIDS community.- [COMMAND_EXECUTION]: The documentation provides examples of shell commands for converting DICOM files, validating datasets, and running containerized BIDS-Apps (e.g., heudiconv, dcm2bids, and docker). These commands are standard for neuroimaging workflows and do not exhibit malicious patterns.- [PROMPT_INJECTION]: The skill processes external neuroimaging datasets, presenting an attack surface for indirect prompt injection. Ingestion points: Metadata is ingested from files like dataset_description.json, participants.tsv, and JSON sidecars when querying datasets via PyBIDS. Boundary markers: No explicit boundary markers or isolation instructions are used to distinguish dataset content from agent instructions. Capability inventory: The skill enables subprocess execution of conversion, validation, and container tools. Sanitization: No specialized sanitization or validation of metadata content is implemented within the provided maintenance script.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 11:29 AM
Security Audit — agent-trust-hub — bids