ai-native-interface
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The content consists entirely of markdown documentation and logic templates defining an interaction architecture. No scripts, executables, or unauthorized network calls are present in the skill files. The skill defines high-level principles for UX orchestration and does not implement technical functions.- [INDIRECT_PROMPT_INJECTION]: Section §9 (Homework Opener) describes a design where the AI ingests untrusted user input, such as company URLs, to analyze and form a response. This design pattern creates a surface for indirect prompt injection if the external website contains malicious instructions aimed at the AI agent.
- Ingestion points: User-provided company URLs or names entered at the start of the conversation (§9).
- Boundary markers: The architecture does not define specific delimiters or instructions to ignore embedded content in user inputs within this paradigm layer.
- Capability inventory: The interaction model implies the use of tools for network fetching or web scraping to analyze user-provided URLs as part of its 'doing homework' posture.
- Sanitization: No explicit sanitization or validation logic is defined within this architectural document; it delegates implementation-specific security to other layers like security-app-llm (§11).
Audit Metadata