pinchtab-tiktok-media-extraction

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

• Third-party content exposure detected (high risk: 0.85). High: the required workflow uses PinchTab to navigate to a runtime-supplied tiktok_url (outsider-controlled public web content) and then extracts readable page text/hydration JSON from TikTok’s #__UNIVERSAL_DATA_FOR_REHYDRATION__ via /tabs/$TAB_ID/evaluate, which can contain attacker-controlled free text that is parsed/ingested into the agent context.

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

• Potentially malicious external URL detected (high risk: 0.80). This skill makes runtime calls to the PinchTab server (PINCHTAB_SERVER_URL endpoints such as /tabs/.../evaluate, /navigate, /cookies, /profiles) which execute supplied JavaScript in the remote browser and return hydration JSON that the skill parses and uses to control its behavior and downloads, so it is a required runtime dependency that executes remote code.