refactor-agents-md

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The required workflow fetches documentation from platform.claude.com, code.claude.com, developers.openai.com, and learn.chatgpt.com via WebFetch.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (medium risk: 0.30). The skill instructs fetching operational guidance and best practices from external URLs (platform.claude.com, code.claude.com, developers.openai.com, learn.chatgpt.com) that govern the agent's behavior during execution, constituting a common/reputable unofficial dependency.