found-vs-deliverable
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a strict validation gate using a local Python script (
score.py) that only uses the Python standard library for data processing. - [SAFE]: Tool access is strictly scoped: write access is limited to the
./gtm/directory via YAML frontmatter configuration, and bash execution is restricted specifically to the scoring script, preventing general shell access. - [SAFE]: No network operations, external downloads, or remote code execution patterns were detected. The script explicitly avoids calling external APIs.
- [INDIRECT_PROMPT_INJECTION]: The skill processes email data and JSON files that may originate from untrusted sources. Ingestion occurs via the
gtm/verify.jsonfile. The skill uses JSON as a boundary format, and thescore.pyscript performs sanitization by validating fields against a fixed schema and canonical values. The capability inventory is restricted (no network or dynamic execution), and the script does not interpret data as instructions.
Audit Metadata