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 list_sources({host, kind: "urls", days, limit: 100}) and list_sources({host, kind: "domains", days, limit: 100}). Match returned URLs against my supplied page inventory, documenting normalization rules. Inspect representative get_prompt_runs({host, prompt_id, days}) after list_prompts({host, limit: 100}). Separate observed cited pages, inventory pages absent from returned rows, and unmatched sources. Explain row limits and sampling. Recommend pages for further investigation; absence is not proof that a page was ignored or could not be retrieved. Return evidence links, coverage limits, and explicit unknowns; never infer unsupported scores.What you need
- A visible host, supported window, and an inventory of your page URLs.
- Supply page text independently for any content-level analysis beyond recorded citation rows.
Reconcile citations with your inventory
Match recorded source URLs to your own page list using explicit normalization rules. Preserve ambiguous matches and distinguish your pages from third-party sources, creating an auditable inventory of observed citations without implying complete retrieval coverage across every answer or engine.
Prioritize pages for closer inspection
Use the reconciled list to choose pages for a focused content or technical review. Pages absent from limited citation results become investigation candidates, with the report explaining sampling limits and the extra evidence needed before drawing conclusions about retrieval or content quality.