skill-writing

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to "Always fetch" and "Use WebFetch" at runtime to retrieve https://agentskills.io/specification and https://code.claude.com/docs/en/skills#frontmatter-reference, so remote documents will be fetched during execution and can directly control prompt/format decisions for frontmatter and workflow behavior.