dtg-base

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFEDATA_EXFILTRATIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXPOSURE]: The skill documents utility methods zip_dir and zip_dirs which allow the agent to archive arbitrary filesystem directories into binary data. This capability provides a mechanism to read and package multiple files from the host environment.
  • [DATA_EXFILTRATION]: The documentation includes a "Common Pattern" (Pattern 4) illustrating the use of the requests.post method within a decorator to send data to external URLs. Combined with the zipping utilities, this establishes a clear functional path for exfiltrating system or record data.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates processing data from Odoo records and search terms, creating an injection surface.
  • Ingestion points: Ingests untrusted data through Vietnamese text fields and record search terms mentioned in odoo-18-dtg-base-guide.md.
  • Boundary markers: There are no instructions for using delimiters or boundary markers to prevent the agent from executing instructions embedded in processed records.
  • Capability inventory: Includes file zipping (zip_dir), directory size probing (_get_file_size), and network exfiltration (requests.post).
  • Sanitization: The guide lacks instructions for sanitizing or escaping external content before it is processed by the described utilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 09:24 PM