token-optimizer

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs offloading and summarising "100 articles/pages" and bulk-processing arbitrary texts via the Groq integration (see references/groq-offloading.md and SKILL.md's "Groq Offloading" section), which causes the agent to ingest untrusted/public third‑party content and then feed those results into its decision/workflow.