canary

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via the browse daemon’s text/page content capture during Phase 2/4/5 ($B text and $B snapshot), where the monitored URL’s page body is authored by an external party and is ingested as readable text at runtime.

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

• Potentially malicious external URL detected (high risk: 0.80). The setup step downloads and then executes a remote installer script from https://bun.sh/install (curl -fsSL "https://bun.sh/install" -o "$tmpfile" followed by bash "$tmpfile"), which is a runtime fetch that executes remote code.