pbvex-backend
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses framework-specific CLI commands such as
npx pbvex codegen,npx pbvex typecheck, andnpx pbvex buildto manage the application lifecycle.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data and possesses significant system capabilities, creating a potential vulnerability surface.\n - Ingestion points: Untrusted data enters the system through
Requestobjects in HTTP actions and variable substitutions in email templates located inpbvex/emails/.\n - Boundary markers: The framework utilizes strict runtime validation via
vvalidators (e.g.,v.object,v.string,v.id) to enforce data contracts and reject malformed input.\n - Capability inventory: The skill utilizes capabilities including database read/write access (
ctx.db), outbound network requests (ctx.http.send), email delivery (ctx.email.send), and job scheduling (ctx.scheduler).\n - Sanitization: The instructions state that the framework automatically escapes HTML variables within email templates to prevent injection.
Audit Metadata