partnerships-architect
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for strategic business partnership management. It provides structured frameworks and workflows for evaluating, designing, and modeling partnerships.
- [COMMAND_EXECUTION]: The skill includes several Python scripts (
scripts/partner_evaluation_scorer.py,scripts/partner_program_designer.py,scripts/partnership_roi_modeler.py) used for analytical tasks. Analysis of these scripts confirms they use only the Python standard library and do not perform any network operations, system-level modifications, or execution of arbitrary shell commands. They strictly process input data to generate reports. - [INDIRECT_PROMPT_INJECTION]: The scripts ingest user-provided data via YAML files (e.g., partner specifications, organization details). While this constitutes an attack surface for indirect prompt injection, the scripts' functionality is restricted to mathematical calculations and structured text output, with no capabilities to execute code or access the internet, which effectively neutralizes the risk.
- Ingestion points: Partner, organization, and partnership specification files processed by the scripts in the
scripts/directory. - Boundary markers: None explicitly implemented for the YAML data ingestion.
- Capability inventory: No subprocess calls, network access, or file-writing capabilities detected across the scripts.
- Sanitization: Input data is parsed through a manual, limited-scope YAML parser that restricts data types to basic scalars (strings, integers, floats, and booleans).
Audit Metadata