lynx-openui
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface by processing natural language user requests to generate UI code, but it provides strong mitigation instructions.
- Ingestion points: Ingests untrusted natural language UI requests and host-supplied component signatures as described in SKILL.md (Workflow step 1).
- Boundary markers: The "Output Contract" in SKILL.md strictly limits the output to OpenUI Lang, prohibiting the inclusion of prose, markdown, or other executable formats.
- Capability inventory: The DSL supports triggering tool mutations and navigating to URLs via the
@OpenUrlaction (references/runtime.md). - Sanitization: The skill mandates a final verification step to validate syntax and references. It also explicitly instructs the agent to use only trustworthy or caller-provided URLs for navigation.
Audit Metadata