til

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

• Third-party content exposure detected (high risk: 0.90). The skill calls the public OpenTIL API (e.g. GET /site and GET /entries at https://opentil.ai) and explicitly loads the site's "custom_instructions" from the GET /site response as additional content-guideline rules used during generation (Execution Flow step 1.5), meaning untrusted/user-generated site data can directly influence the agent's outputs and actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill runs npx @opentil/cli at runtime (and explicitly fetches https://registry.npmjs.org/@opentil/cli/latest), which causes remote npm-hosted code to be fetched and executed for updates/image uploads, so external content is executed during runtime.