cut-the-curve
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided source material and inputs to generate implementations or artifacts. It also interacts with external SandBase APIs (
sandbase_run). This combination creates a surface for indirect prompt injection where malicious instructions embedded in processed data could attempt to influence the agent's behavior or manipulate API arguments. - Ingestion points: User source material, inputs, and existing conventions (SKILL.md).
- Boundary markers: None identified; the instructions do not specify the use of delimiters or warnings for the agent to ignore instructions embedded within the source material.
- Capability inventory: Network operations via vendor-provided tools
sandbase_runandsandbase_run_get(SKILL.md, references/sandbase-api-map.md). - Sanitization: The instructions do not define specific sanitization or validation steps for content extracted from external inputs before processing.
Audit Metadata