skill-writing

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

• Third-party content exposure detected (low risk: 0.10). The workflow fetches documentation from agentskills.io and code.claude.com via web fetch and scripts, which constitutes active searching/fetching of specific outsider-authored or public vendor-authored content.

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

• Potentially malicious external URL detected (high risk: 0.70). The script fetches the runtime specification from agentskills.io via curl, which governs the agent's behavior and formatting rules, constituting an operational dependency from an unofficial or arbitrary external source.