agui-dotnet-human-in-the-loop
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of educational content and C# code snippets for the 'ag-ui-protocol' vendor SDK. It provides guidance on implementing security best practices, such as gating sensitive tool execution behind human approval.
- [PROMPT_INJECTION]: The skill demonstrates how to ingest free-form human input via
InterruptRequestContent. While this introduces an indirect prompt injection surface where user-supplied data is added to the agent context, the implementation is presented as a mechanism for human oversight and interaction. - Ingestion points: Client-side
AskHumancall in code snippets. - Boundary markers: None provided in the simplified code examples.
- Capability inventory: Illustrative
DeleteFiletool and chat message history management. - Sanitization: None demonstrated in the snippets; the logic relies on the host application's implementation.
- [COMMAND_EXECUTION]: The documentation includes placeholder function definitions for tools like
DeleteFile. These are illustrative examples for developers and do not result in unauthorized command execution during the skill's normal use or analysis.
Audit Metadata