feature-usage-feed

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches raw LLM traces and user messages via posthog:query-llm-traces-list and posthog:query-llm-trace (Step 2), which ingest untrusted, user-generated content (including first user messages and sampled $current_url) that the agent must read and classify to drive eval creation, runs, and Slack postings—allowing third-party text to materially influence tool actions.