bids

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill includes scripts/update_schema.py, which fetches the BIDS schema and Extension Proposals (BEPs) list from official community repositories hosted on ReadTheDocs and GitHub (bids-standard organization). These are official authorities for the BIDS standard.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process Brain Imaging Data Structure (BIDS) datasets, which typically contain numerous metadata files in JSON and TSV formats (such as dataset_description.json, participants.tsv, and modality-specific sidecars). These files serve as ingestion points for external, potentially untrusted data into the agent's context.
  • Ingestion points: Dataset indexing via BIDSLayout (PyBIDS), BIDS validation, and metadata editing workflows described in SKILL.md.
  • Boundary markers: The skill does not provide explicit delimiters or instructions to ignore embedded prompts when processing these files.
  • Capability inventory: The skill facilitates file system read/write operations and the execution of neuroimaging CLI tools (heudiconv, dcm2bids, bids-validator).
  • Sanitization: No specific sanitization or validation logic for the content of processed metadata files is defined in the skill instructions.
  • [COMMAND_EXECUTION]: The skill provides instructions and examples for executing various command-line utilities used in neuroscience research, including heudiconv, dcm2bids, and bids-validator.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 08:08 PM
Security Audit — agent-trust-hub — bids