Citation Intelligence vs Serpapi MCP

panel_track

Save, load, or list named query panels. A panel is a persisted set of queries you want to monitor over time (e.g. editorial-watchlist). Use action=save with queries[] to create, action=load to read, action=list to enumerate. Panels live under <config>/panels/<name>.json.

panel_run

Run a saved panel through am_i_cited and append a timestamped snapshot. Side effects: makes external API calls to the configured AI engine (costs API quota); writes one snapshot file to disk at <config>/snapshots/<panel>/<iso>.json. Requires at least one engine API key (same as am_i_cited). Returns per-query citation presence and a citation_rate summary for the run. Use panel_track to create a panel first; use citations_trend to read the accumulated trend after multiple runs.

report_visibility

Turnkey AI visibility report for a domain across a query set. Composes check_citations over every query (or a saved panel) and returns the metrics AI-visibility trackers sell as a dashboard, in one call: mention frequency (citation_rate), share_of_voice vs competitors, average rank when cited, and brand sentiment from the answer text. Side effects: one check_citations call per query (costs API quota for uncached queries; cached queries are free). Returns structured summary + top_domains + per_query, plus a rendered Markdown report (include_markdown=true) suitable for a public page. Provide queries[] or a panel name. Same engine selection as check_citations.

citations_trend

Report citation rate over time for a panel from stored snapshots. Read-only; cache-only — makes no API calls to any AI engine and costs no API quota. Reads snapshot files from <config>/snapshots/<panel>/. Returns: snapshots[] (one entry per panel_run invocation, each with timestamp and citation_rate), plus per-query deltas (gained/lost/unchanged) comparing first vs last snapshot. Returns an empty series when no snapshots exist yet. No auth required. No rate limits. Use panel_run to accumulate snapshots first; use since to restrict the time window.

competitors_compare

Run predict_citation on 2-10 URLs and return a side-by-side signal table plus a list of signals where the URLs diverge. Use to compare your URL to top-cited competitors for the same query.

signals_wikipedia

List Wikipedia articles that reference the given domain. Read-only. One HTTPS GET to the Wikipedia API (en.wikipedia.org/w/api.php?action=query&list=exturlusage). No auth required; no API keys; no rate limits beyond Wikipedia's public API fair-use policy (~1 request/second). Returns article titles and URLs. Wikipedia backlinks are the highest-lift signal for LLM training corpora — a domain cited from Wikipedia is far more likely to appear in AI training data and citation pools. Use lang to query non-English Wikipedias.

audit_sitemap

Fetch a sitemap.xml (or sitemap index) and run predict_citation on every URL. Returns results sorted worst-score-first. Surfaces systemic issues across a whole site in one pass. Zero engine keys needed.