design-review

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

• Third-party content exposure detected (high risk: 0.75). The required workflow can ingest outsider-authored free text via runtime browsing of a user-supplied Target URL (public web content / arbitrary site pages) using $B goto <url> and subsequent page reads/snapshots, which can include attacker-controlled text that may be incorporated into the LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's setup step can run a remote installer via "curl -fsSL https://bun.sh/install | bash" (SKILL.md:249), which fetches and executes remote code at runtime.