market-recon

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the orchestrator to fetch and skim public Reddit posts from specific subreddits (see "Query Strategy" in SKILL.md) and write them to .godag/context/{task_id}-market.md for a subagent to read and act on, exposing the agent to untrusted user-generated content from Reddit.