style-clone

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes runtime calls that ingest user-supplied writing samples or whole-project chapter text into POST /api/style-clone/analyze (including “manual-paste” and concatenated completed chapters), so outsider-authored free text can be processed without selecting a specific item first.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill calls POST /api/style-clone/analyze (and POST /api/projects/:id/style-clone) at runtime and uses the returned "systemPrompt" to control agent prompts/instructions, so the runtime API responses directly determine agent behavior.