opentiny-next-app-integration
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the use of the official
@opentiny/tiny-robot-clito scaffold chat components within the project directory. This is a standard vendor-provided integration workflow and is executed within the project boundary. - [EXTERNAL_DOWNLOADS]: The integration process requires the installation of scoped dependencies from the OpenTiny ecosystem, such as
@opentiny/genui-sdk-core,@opentiny/next-sdk, and@opentiny/tiny-robot-kit. All packages originate from the official vendor repository. - [DYNAMIC_EXECUTION]: The provided templates use standard Vue dynamic imports (
defineAsyncComponent) to load GenUI materials and renderers. This is a legitimate technique for optimizing client-side performance and does not involve executing untrusted remote code. - [INDIRECT_PROMPT_INJECTION]: The skill includes a proactive security layer for
PageToolactions. It defines aPageToolPolicythat validates every model-requested action against an explicit allowlist and stable element identifiers, preventing the model from interacting with sensitive or unauthorized UI components. - [DATA_EXFILTRATION]: The skill manages LLM API keys for various providers and its own GenUI service. It implements specific routing logic in
resolveChatAuthorizationKeyto ensure credentials are only sent to their designated endpoints and provides clear documentation warning against hardcoding secrets in client-side build artifacts.
Audit Metadata