pycalphad
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external thermodynamic database files (.tdb) and configuration files (.json). \n
- Ingestion points: The script
scripts/equilibrate.pyreads user-provided files fromdatabase_pathandsettings_pathvia command-line arguments.\n - Boundary markers: While the script parses data into specific thermodynamic models, it does not explicitly sanitize text values extracted from these files (such as phase names or labels) before they are written to reports that the agent subsequently reads.\n
- Capability inventory: The script's capabilities are restricted to the local filesystem, including directory creation and writing structured data files (
report.jsonandphase-equilibria.csv). It lacks network access or system-level privileges.\n - Sanitization: The implementation employs strict JSON schema validation and chemical consistency checks against the thermodynamic database elements, which mitigates risks from malformed input data formats.
Audit Metadata