qa

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill's smoke workflow instructs users to provide or load credentials, enable Chrome remote debugging, and save/load auth state files—actions that could weaken security boundaries or expose sensitive credentials though they appear intended for QA/test purposes.

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

• Third-party content exposure detected (medium risk: 0.30). Source outsider-authored free text can be supplied via /qa smoke <source> where source is treated as-is in references/workflows/smoke.md Step 1 (Free-text), then used to derive browser-agent smoke scenarios whose pages/snapshots are ingested at runtime.