design-ship

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

• Third-party content exposure detected (medium risk: 0.30). Input to the pipeline is an outsider-authored Claude Design handoff bundle URL/file passed to /ork:design-ship (then delegated to /ork:design-import), which the workflow ingests at runtime and uses to scaffold content and generate PR text/snapshots.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly accepts a runtime Claude Design handoff URL (e.g., https://claude.ai/design/abc123) and delegates to /ork:design-import to fetch and process that bundle, so remote content will directly determine what code/components the agent scaffolds and thus controls agent behavior.