soul

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to browse and analyze files in the data/ folder (e.g., Twitter/X archives, blog posts, other user-generated/public writing) as part of the build and grounding workflow—see BUILD.md Step 1/Step 2, SKILL.md "Data Usage", and data/_GUIDE.md—so untrusted public content is consumed and can directly shape agent behavior.