Skills
Navigate
skills/fcakyon/claude-codex-settings/durable-objects/Gen Agent Trust Hub

durable-objects

Pass

Audited by Gen Agent Trust Hub on Oct 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for processing untrusted external data from HTTP requests, which is a standard requirement for web workers.\n
  • Ingestion points: request.json() and url.searchParams are used to receive external data in references/workers.md.\n
  • Boundary markers: The skill provides specific examples of schema-based validation using the zod library to define boundaries for untrusted input.\n
  • Capability inventory: The skill demonstrates the use of sql.exec for persistent storage operations in SKILL.md and references/workers.md.\n
  • Sanitization: Robust input sanitization is modeled through zod validation blocks in the handleSendMessage pattern.\n- [EXTERNAL_DOWNLOADS]: The skill references official Cloudflare documentation and resources for development guidance.\n
  • Evidence: Links to developers.cloudflare.com for API references, best practices, and testing guides are included throughout the documentation.\n- [COMMAND_EXECUTION]: The skill includes standard CLI commands for managing Cloudflare Worker deployments and resources.\n
  • Evidence: references/workers.md lists development commands such as wrangler dev, wrangler deploy, and wrangler d1 execute for local development and database management.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 3, 2026, 11:42 PM