pstack-make-bot-ui
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions explicitly direct the agent to treat external request bodies as untrusted data and perform input validation.
- [SAFE]: Sensitive information handling follows best practices by requiring credentials to be obtained through secure environment mechanisms and kept server-side.
- [SAFE]: Network exposure is restricted to localhost by default, with remote access only provided via Tailscale, a well-known and secure networking service, upon explicit user request.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external webhook data. While this presents an attack surface for indirect injection, the instructions include specific mitigations such as input validation and treating data as untrusted.
- Ingestion points: Processes webhook request bodies from external sources in SKILL.md.
- Boundary markers: Instructions explicitly label request bodies as untrusted data.
- Capability inventory: Capable of network operations (Tailscale, local proxy) and file system operations (building the UI/server).
- Sanitization: Mandates input validation and authorization checks in the implementation phase.
Audit Metadata