sflo-quick
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to autonomously transform user-provided requests into executable code, which is then run for verification. This workflow introduces an inherent surface for indirect prompt injection if the input request contains malicious instructions intended to be executed during the verification phase.
- Ingestion points: The process begins with a user-supplied request or specification used to generate the initial FRAME.md artifact.
- Boundary markers: The skill employs a formal lifecycle (FRAME, MAKE, CHECK, ALIGN, DELIVER) to document intent, boundaries, and results, providing a traceable record of the agent's logic and constraints.
- Capability inventory: The CHECK phase involves executing generated scripts, performing database queries, and running system commands in clean processes to verify behavior.
- Sanitization: The ALIGN gate acts as a validation step where evidence from the CHECK phase is compared against the initial FRAME criteria to detect deviations or unintended behaviors.
- [DYNAMIC_EXECUTION]: The skill involves the autonomous generation and subsequent execution of code artifacts (such as scripts, SQL, or data work) to achieve the requested result.
- Evidence: The MAKE gate involves building product candidates, while the CHECK gate specifically requires executing these candidates in isolated contexts to observe behavior and provide direct evidence of compliance with the defined specification.
Audit Metadata