add-skill

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

• Third-party content exposure detected (high risk: 0.75). This workflow reads .local/ recursively and also performs “web-search the topic” in step 2, so at runtime it can ingest outsider-authored free text (local files not authored by the operating user and/or fetched public web pages) into the agent’s LLM context.