hn

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

• Third-party content exposure detected (medium risk: 0.30). Hacker News (via the Algolia API) front page/search results and then hn read ID consume story and comment plaintext authored by third parties, so an outsider can submit/poison text through HN items that the workflow reads at runtime.

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

• Potentially malicious external URL detected (high risk: 1.00). The SKILL includes an installation command that fetches and builds remote code ("go install github.com/voska/hn-cli/cmd/hn@latest"), which downloads and executes third-party Go code during setup and thus can execute remote code.