qa

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

• Third-party content exposure detected (medium risk: 0.30). The workflow reads Linear issues, comments, test plans, and design attachments created by users/outsiders, ingesting outsider-authored free text into the agent context.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the agent to download and execute an external script via curl in SKILL.md line 73 (curl -L "<test_plan_attachment_url>" -o /tmp/qa-test-plan.md), but this URL is a dynamic runtime variable supplied by an issue tracker attachment rather than an authoritative or fixed operational dependency.