bagisto-data-transfer

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill facilitates the ingestion of external data files (CSV, XLSX, XML), which creates a surface for indirect prompt injection where malicious instructions could be embedded within the data rows.
  • Ingestion points: External data enters the context through the Helpers\Sources\ classes in SKILL.md which iterate over various file formats.
  • Boundary markers: The skill mandates the implementation of a validateRow() method and a $validColumnNames property (documented in importers.md) to enforce schema constraints.
  • Capability inventory: The skill uses database repositories for writing data and dispatches queued jobs (ImportBatch, LinkBatch, IndexBatch) as described in pipeline.md.
  • Sanitization: The instructions in importers.md recommend using Laravel's Validator to verify data types and constraints.
  • [EXTERNAL_DOWNLOADS]: The pipeline includes a specific state for downloading product images from remote sources, including external URLs (IMAGE_SOURCE_URL). This results in the system performing network requests to retrieve assets based on the contents of the import file.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:20 PM
Security Audit — agent-trust-hub — bagisto-data-transfer