supabase

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow instructs fetching public Supabase web content (e.g., https://supabase.com/changelog.md and linked docs pages as markdown), which is outsider-authored free text that can be ingested into the agent’s LLM context via the “fetch docs pages as markdown” / “scan changelog” path.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to fetch https://supabase.com/changelog.md at runtime and act on its contents (scan for breaking-change tags and follow linked pages), which means remote content directly controls the agent's behavior (the product security page https://supabase.com/docs/guides/security/product-security.md is also referenced for on-demand runtime fetches).