ollama-launch

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs pulling models from the public Ollama registry (e.g., "ollama pull qwen2:7b") and the skill runs local Ollama for RLM distillation and embeddings (Learning Loop Seal Phase / RLM Factory), so the agent fetches and consumes third‑party model outputs that can materially influence decisions and tool use.