context-management

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

• Third-party content exposure detected (high risk: 0.90). SKILL.md's "Miner" recipe and the "Research / Logs" guidance explicitly instruct the agent to read/search the web and external documents and then "EXECUTE the Next Step" from the injected summary, so untrusted public web content can be ingested and materially influence subsequent actions.