follow-builders

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

• Third-party content exposure detected (medium risk: 0.30). The agent skill consumes posts from X (Twitter) accounts and YouTube podcast transcripts via a central GitHub feed, which aggregate content written by third-party builders and podcast guests.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill fetches core runtime instructions, prompts, and data feeds from an arbitrary GitHub repository (zarazhangrui/follow-builders) at runtime via prepare-digest.js, presenting a weak/unknown provenance operational dependency.