composable-mini-apps
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides documentation and structural guidance for developers on how to organize workspace applications. It does not include scripts, executables, or instructions that violate security policies.
- [INDIRECT_PROMPT_INJECTION]: The skill describes an architecture involving data exchange between agents and external APIs, which naturally defines an ingestion surface for external data.
- Ingestion points: Data retrieved from sibling applications via
call-agentandinvokeAgent, and external provider data fetched viaprovider-api-requestas described inSKILL.md. - Boundary markers: The guidelines suggest using structured artifacts (compact JSON-like objects) for handoffs, which serves as a natural data boundary compared to raw transcript pasting.
- Capability inventory: The skill utilizes workspace discovery tools and A2A invocation methods for orchestration.
- Sanitization: While the skill focuses on architectural flow, it emphasizes using bounded structured data (artifacts) for downstream consumption rather than unstructured corpus processing.
Audit Metadata