convex-domains
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates several local CLI tools such as flarectl, aws, gcloud, doctl, and vercel to create DNS records. It includes safeguards such as requiring user confirmation and checking authentication status before execution. These tools belong to well-known service providers and are used within their intended scope.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, including domain names and deployment targets, which are used to construct shell commands.
- Ingestion points: User-provided domain names and Convex deployment URLs processed in the workflow.
- Boundary markers: The instructions mandate that the agent must show exact commands and obtain explicit user confirmation before any execution.
- Capability inventory: Shell command execution via local CLI tools and environment variable modification (SITE_URL, RP_ID, ORIGIN).
- Sanitization: Relies on human-in-the-loop verification of the generated command strings.
Audit Metadata