Store and extend complete RDF/OWL documents through MCP. import accepts Turtle/N-Triples/N-Quads/TriG without conversion; update executes INSERT DATA for additive edits. Replacing/removing axioms is unsupported; use a new versioned LBB graph for a revised document. Named RDF graphs are unsupported; Turtle/N-Triples use the default graph, and dataset formats must contain only default-graph quads. These write graph facts, distinct from lbb_configure's native schema metadata. A first RDF write selects RDF-native storage, which refuses later property-graph commits; choose the write workflow before bootstrap. Retries deduplicate by content unless idempotency_key is supplied. A completed write schedules publication; inspect action=publication and verify using lbb_query entailment=owl.

Read graph context and exact graph facts. Actions: guide, graphs, publication, ontology, ontology_conformance, schema, ontology_search, metadata, entity, ontology_suggestions. graphs works before bootstrap; ontology_suggestions lists the ontology change suggestions that wait for review (or were decided). publication reports whether writes are queryable. ontology and schema return complete entries with page_size, section and cursor; follow next until absent. schema reads active native ontology/SHACL metadata without running validation. Query asserted RDF/OWL axioms separately with lbb_query. ontology_conformance serves the durable report referenced by the pinned published root. entity returns one node's attributes and current relationships from the RDF read. For a node's past values, run SPARQL with as_of_commit_seq through lbb_query. Use lbb_query with SPARQL property paths for precise path selection.

Analytical and expert reads. Modes: question (a question in plain words), structured (SPARQL-subset JSON body), sparql (SPARQL text), search (instances by meaning over every searchable class; filter narrows by class and relationship; every hit checked against the graph), analyze. Use mode=question when you have a question in plain words and no SPARQL query: the server selects the kind of query (route), writes the SPARQL query from a description of the graph, checks it, runs it, and returns the route, the rationale, the query and its rows. Continue or correct that query with mode=sparql. Each question uses model tokens and counts toward a daily limit of the stack. SPARQL is the query language; search finds what the words describe. SPARQL text can also search by meaning inside the query with ?x https://littlebigbrain.com/search#similarTo "words", so the other patterns filter and join the hits in one query; the result's search field reports the plan. To plan a search: lbb_embeddings action=list, then SPARQL for a class's relationships on a sample, then mode=search with explain=true to check the resolved filter before the real search (lbb_inspect action=guide has the queries). Relations are https://littlebigbrain.com/r/NAME and types https://littlebigbrain.com/class/NAME (both lowercased); entities are content-addressed, so anchor a named one by its rdfs:label rather than building its IRI. Structured and text queries pin one published watermark for the request.

Read model-training inputs or compare retrieval configurations over one pinned published snapshot, or read what the managed models did. shadow_eval takes the API ShadowEvalRequest body; dataset actions return bounded training examples at an optional signal split. activity reads one month of the stack's model use (all graphs): calls, items, estimated tokens and cost per feature (index, search, fit, judge, training) and model, by day and by graph, the months with activity, and the model each feature uses now.

Search setup, read only: embeddings declared on classes (an embedding on a class covers its subclasses; one model per graph). list shows each embedding's status (serving and building version, backfill progress, lag) and the graph's model; get shows one; preview shows every candidate fact of a class with its coverage and examples, and the exact text of sample instances, and stores nothing. Declare, refresh, or change the model with lbb_embeddings_manage; search with lbb_query mode=search. list is the first step of planning a search (lbb_inspect action=guide has the recipe).

Declare or refresh an embedding, or change the graph's model. Run lbb_embeddings action=preview first and show the user the text it will embed. declare sets the class, the fields (one-hop property paths such as label, description, calls/label; omit for an automatic choice); a new embedding takes the graph's model, and another model is refused. On an existing embedding it changes only what you name, and a changed recipe builds a new version while the old one serves. model moves every embedding of the graph to model: each builds a new version, and the graph switches when all are ready (a search never mixes two models). refresh runs one step of the embed job now (it also runs by itself after each published commit).

Delete an embedding: its name and its vectors (index-gc removes the stored runs). Its class stops being searchable until it is declared and built again. Ask the user first; confirm must repeat the name.

Managed evals: the thumbs up / thumbs down of the graph, per result. A query run with request (lbb_query) records a trace with one item per result (hit or row); trace reads it back with the item ids. label marks one result (item + valid) or several (items) relevant or not; the labels become the golden's ground truth. golden freezes a query (every result it returns now is relevant). run replays every golden at the current commit: pass when every relevant result is back and no wrong one is; new results open a review trace. judge lets the platform's judge model (the hosted frontier model) label unlabeled results. summary, traces, goldens, results, and settings read state. Model checks: a judge model checks a sample of the model calls LBB makes for the graph (rerank, route, rewrite, fit, propose, label). checks_summary reads a month per job and model (checks, score, right, partly, wrong, reviews) and the judge's agreement with people; checks lists the month's checks, newest first, with the judge's verdict, score and reason and the ground truth (truth). review_check records a person's review of one check (call_id): agree=true keeps the judge's verdict; agree=false with verdict (and an optional score, reference and note) corrects it. The review becomes the call's ground truth, so show the check to the user and review only what the user confirmed.

Write graph facts, retract them, or label ranked results. mode=facts writes triplets/embeddings/properties; mode=retract removes a wrongly-added fact (by edge or by entity) without a full reset; mode=search_feedback stores query/result relevance labels (Feedback grades: 3=ideal/good, 1=partial, 0=bad; include query, search_id when available, target, rank, score). Explicit idempotency_key wins; when omitted, MCP derives a stable content hash so content-identical retries dedupe. Facts mode defaults edge_idempotency to append; pass skip_unchanged for re-runnable backfills.

Manage native schema metadata. Actions: define_ontology (friendly spec with super_types), evolve_ontology (ordered edits including add_super_types), list_starters (the base ontologies crm, documents and work, each with its status on the graph: absent, partial or applied, what applying adds, and conflicts), apply_starter (add what the graph lacks of a starter in one ontology version; a relation the graph has is widened; refused with starter_conflict when the graph holds a term differently; dry_run previews), publish_schema (SHACL activation), suggest_ontology_change (file a change for a person to review instead of applying it; prefer it when the graph's owner reviews ontology changes, and list the result with lbb_inspect action=ontology_suggestions), get_rewrite_profile and set_rewrite_profile (the graph's notes and up to 20 worked question-to-SPARQL examples that lbb_query mode=question reads for every question; pass the version you read as expected_version). define, evolve, publish and set_rewrite_profile support dry_run previews. Definition/import here extracts native metadata; it does NOT store the complete RDF/OWL document as queryable graph facts. Use lbb_rdf import for full OWL and lbb_rdf update for additive INSERT DATA revisions; RDF deletions are unsupported. Publish_schema accepts unchanged ontology plus shapes; use define/evolve for native ontology changes. Publication enqueues durable conformance; a preview does not validate the whole graph.