Command
Use list_projects({}) to confirm my supplied project host is visible. Use my chosen days value of 7, 30, or 90 for every time-windowed call. Read get_brand_insights({host, days}), then search_chats({host, query: myConcern, days, limit: 20}). Use list_prompts({host, limit: 100}) and get_prompt_runs({host, prompt_id, days}) to inspect relevant samples. Cross-reference list_sources({host, kind: "urls", days, limit: 100}) without fetching arbitrary pages. Separate directly observed negative wording from suspected source influence. Produce a correction brief with quotes, citation URLs, missing context, and factual checks; do not invent a damage score or assert causality. Return evidence links, coverage limits, and explicit unknowns; never infer unsupported scores.What you need
- A visible host, supported window, and the brand concern or phrase to investigate.
- Verified brand facts; supply page text separately if source content must be checked.
Locate the wording behind a concern
Start with perception summaries, then inspect matching answer samples to understand what was actually said. A cited URL is evidence of a reference, not proof that its page caused the wording or changed a brand metric in the observed window.
Prepare a factual response brief
Group concerns by claim and attach the available answer context, cited sources, and verification needs. Recommend corrections only where supplied facts support them, giving brand and communications teams a bounded review document rather than an unsupported reputation verdict.