api-caching-strategy
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to process user-supplied content regarding API strategies and has high-privilege tools available.
- Ingestion points: Processes user requests, questions about best practices, and potentially external API specifications (REST, GraphQL, OpenAPI) referenced in the
SKILL.mdtriggers. - Boundary markers: The instructions do not specify any boundary markers (like XML tags) or explicit instructions for the agent to ignore potentially malicious commands embedded within user-supplied API data.
- Capability inventory: The skill allows access to
Read,Write,Edit, andBash(curl:*), which provides the agent the ability to modify project files or perform network requests based on instructions extracted from the input data. - Sanitization: There is no evidence of sanitization, filtering, or validation logic defined to prevent external data from influencing the agent's behavior during the generation of configurations or execution of bash commands.
Audit Metadata