hook-authoring

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly documents HTTP hooks that POST hook inputs to arbitrary URLs (see "HTTP Hooks (Claude Code 2.1.63+)" and the example URL https://my-service.example.com/hooks/validate-bash), so external HTTP responses (untrusted third-party content) are ingested and can directly influence PreToolUse decisions and injected context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill includes explicit runtime HTTP hook examples that POST hook input to external endpoints whose JSON responses are used to control hook decisions (e.g., "https://my-service.example.com/hooks/validate-bash" and the example config fetch "http://api.example.com/config"), so these URLs are runtime external dependencies that can directly influence agent prompts/behavior.