customware-email

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: Credential Isolation. The skill mandates that "POSTMARK_SERVER_TOKEN" and "POSTMARK_INBOUND_WEBHOOK_SECRET" remain strictly server-side and are never exposed to the client or browser bundles.
  • [SAFE]: Secure Logging. A specific policy is defined to exclude sensitive personally identifiable information (PII) and email content from operational logs, preventing data exposure.
  • [SAFE]: Input Validation. Inbound webhooks are protected by HTTP Basic Authentication and structural validation using Zod to ensure data integrity.
  • [PROMPT_INJECTION]: The skill handles untrusted data from inbound emails, which constitutes an indirect prompt injection surface.
  • Ingestion points: Inbound webhook defined in "references/inbound.md".
  • Boundary markers: Structural validation via Zod is implemented for the JSON payload.
  • Capability inventory: File system persistence (SQLite/Drizzle) and network operations (Postmark API).
  • Sanitization: The skill emphasizes schema validation for the email envelope but does not specify sanitization of the email body text.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 11:36 AM
Security Audit — agent-trust-hub — customware-email