design-agentic-coding-ecosystem

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

• Third-party content exposure detected (high risk: 0.75). The required workflow includes a live web-search step (SKILL step 5) that fetches outsider-authored public web pages at runtime to extract tool/pricing info, which can be ingested as readable text into the agent’s LLM context (even if the subagent is instructed to return only distilled results).