supabase

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

• Third-party content exposure detected (medium risk: 0.30). exposure path: the required runtime workflow includes an agent evals harness that ingests scenario text from evals/<id>/PROMPT.md (authored in the external supabase/evals repo) as the agent’s task prompt, and the agent then runs Supabase tools accordingly.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill includes runtime commands that fetch and execute remote scripts (curl -fsSL https://supabase.link/setup.sh -o setup.sh; sh setup.sh) and clones an external git repo that contains eval PROMPTs and code (git@github.com:supabase/evals.git), both of which download remote content executed or used at runtime to control agent prompts or behavior.