cloudflare-workers-security

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is a defensive resource providing code templates and documentation for securing serverless functions. It correctly handles sensitive operations like secret management and authentication with best-practice implementations.
  • [INDIRECT_PROMPT_INJECTION]: The skill templates define a surface for processing untrusted HTTP request data. This is safely managed through the mandatory use of Zod for schema validation and custom sanitization logic.
  • Ingestion points: The skill handles untrusted data via request.json(), request.headers.get(), and URL parameters in templates/secure-worker.ts, templates/auth-middleware.ts, and references/input-validation.md.
  • Boundary markers: It implements robust boundary markers through the extensive use of Zod schemas to define strict data shapes and ignore unexpected input.
  • Capability inventory: The provided templates have capabilities to interact with KV storage, D1 databases, and R2 buckets, and can perform outbound HTTP requests via fetch.
  • Sanitization: The skill includes specific sanitization logic for HTML escaping (escapeHtml), path sanitization (sanitizePath), and generic content cleaning (sanitizeUserContent) to mitigate injection and traversal risks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 12:00 AM
Security Audit — agent-trust-hub — cloudflare-workers-security