Search the report archive by free text, motifs, phenomenon type, source type, stance, evidence grade, or date range. Returns brief records with grades and safety flags; use timeline_get_report for the full record.

Return one full report: long summary, labeled claims, source metadata, anchors, motifs, related reports, safety flags, and recommended phrasing for agents.

Browse Mandela Effect catalog items: remembered vs on-record variants, why communities link them to timeline lore, and the mainstream memory-science explanation for each.

The motif constellation: recurring symbols with definitions, families, frequencies, and weighted co-occurrence edges — ready for graph rendering or motif-based filtering.

Curated narrative arcs, each written three ways (belief / neutral / skeptical) with motif chips, timeline anchors, member reports, and research prompts.

Timeline events on one axis, three labeled layers: official-science (always cited), community-wave (interpretation), symbolic-anchor (reading). Layers never blend.

Compare belief-framed, neutral-archive, and skeptical framings for one topic, with the shared facts (officially sourced) and disputed points separated out.

Generate research directions for a report, motif, or thread: source-search queries, related motifs, comparison questions, story-expansion prompts, and a verification checklist. Never invents sources.

Corpus statistics: totals by phenomenon, motif, source type, evidence grade, stance, year, and language, plus catalog/thread/event counts and flagged items.

The rules of the corpus: what it is, safe language, citation policy, recommended first calls, install snippet, and raw data endpoints. Call this first.