Try examples Copy for AI agent ArcFlow implements ISO/IEC 39075 GQL as WorldCypher — its native query dialect. All standard GQL syntax works as specified. WorldCypher adds extensions for temporal snapshots, live queries, confidence scoring, and graph algorithms that GQL leaves to implementations.
Standards conformance:
Standard Status openCypher TCK ✅ 100% — 3881/3881 scenarios pass ISO GQL V2 (ISO/IEC 39075) ✅ Full — GQLSTATUS codes, IS LABELED, ELEMENT_ID, NEXT WHEN/THEN/ELSE/END, constraints, transaction control
For a narrative overview of GQL lineage, how WorldCypher relates to Cypher and ISO GQL, and a comparison with other implementations, see the GQL Conformance page.
Complete reference of supported features, syntax, and current implementation status follows.
Feature Status Syntax Notes CREATE node ✅ CREATE (n:Label {props})CREATE relationship ✅ CREATE (a)-[:REL]->(b)Inline with nodes MATCH ✅ MATCH (n:Label {props})Multi-MATCH ✅ MATCH (a) MATCH (b) RETURN a, bCartesianProduct OPTIONAL MATCH ✅ MATCH (a) OPTIONAL MATCH (a)-[:R]->(b)MERGE node ✅ MERGE (n:Label {props})ON CREATE/ON MATCH SET MERGE relationship ⚠️ MERGE (a)-[:R]->(b)Creates if not found; find-or-create may create duplicates DELETE ✅ MATCH (n) DELETE nDETACH DELETE ✅ MATCH (n) DETACH DELETE nCascade deletes relationships SET property ✅ SET n.key = valueOne property per SET clause SET multiple ⚠️ SET n.a = 1, n.b = 2Use separate SET clauses: SET n.a = 1 SET n.b = 2 REMOVE ✅ REMOVE n.keyRemove a property WITH ✅ WITH n WHERE ... RETURNProjection + CTE forms UNWIND ✅ UNWIND [1,2,3] AS xLiteral lists only WHERE ✅ WHERE n.age > 30WHERE OR labels ✅ WHERE (n:Person OR n:Company)RETURN ✅ RETURN n.name, count(*)ORDER BY ✅ ORDER BY n.age DESCLIMIT ✅ LIMIT 10
Feature Status Syntax Notes CONTAINS ✅ WHERE n.name CONTAINS 'ali'STARTS WITH ✅ WHERE n.name STARTS WITH 'A'toLower ✅ RETURN toLower(n.name)String transformation COALESCE ✅ RETURN COALESCE(n.x, 0)Null coalescing with default labels() ✅ RETURN labels(n)Variable-length paths ✅ MATCH (a)-[:KNOWS*1..3]->(b)
Feature Status Syntax Notes count(*) ✅ RETURN count(*)sum ✅ RETURN sum(n.x)avg ✅ RETURN avg(n.age)min / max ✅ RETURN min(n.x), max(n.x)collect ✅ RETURN collect(n.name)
Feature Status Syntax Notes LAG ✅ lag(n.close, 1) OVER (PARTITION BY n.symbol ORDER BY n.date)Previous row value LEAD ✅ lead(n.close, 21) OVER (PARTITION BY n.symbol ORDER BY n.date)Future row value AVG OVER ✅ avg(n.close) OVER (... ROWS BETWEEN 199 PRECEDING AND CURRENT ROW)Rolling average STDDEV_POP OVER ✅ stddev_pop(n.close) OVER (...)Rolling standard deviation SUM OVER ✅ sum(n.volume) OVER (PARTITION BY n.symbol ORDER BY n.date)Expanding sum PERCENT_RANK ✅ percent_rank() OVER (PARTITION BY n.date ORDER BY n.close)Cross-sectional ranking ROW_NUMBER ✅ row_number() OVER (PARTITION BY n.date ORDER BY n.close DESC)Row numbering SKEWNESS OVER ✅ skewness(n.close) OVER (...)Rolling skewness MAX OVER ✅ max(n.close) OVER (...)Rolling maximum
Feature Status Syntax Notes CREATE LIVE VIEW ✅ CREATE LIVE VIEW name AS MATCH ... RETURN ...Maintained via CDC Read from view ✅ MATCH (row) FROM VIEW name RETURN rowZero-cost read List views ✅ CALL db.liveViews
Feature Status Syntax Notes AS OF ✅ MATCH (n) AS OF seq NTemporal snapshot query
Feature Status Syntax Notes CREATE VECTOR INDEX ✅ CREATE VECTOR INDEX name FOR (n:Label) ON (n.prop) OPTIONS {dimensions: N, similarity: 'cosine'}vector similarity search CREATE FULLTEXT INDEX ✅ CREATE FULLTEXT INDEX name FOR (n:Label) ON (n.prop)BM25 scoring
Feature Status Syntax Notes Parameter binding ✅ $param in queriesSubstituted before compilation, prevents injection
Run directly — no projection lifecycle needed.
Procedure Description algo.pageRank()PageRank centrality (20 iterations, damping 0.85) algo.confidencePageRank()Confidence-weighted PageRank algo.betweenness()Betweenness centrality algo.closeness()Closeness centrality algo.degreeCentrality()Degree centrality algo.louvain()Louvain community detection algo.leiden()Leiden community detection algo.communityDetection()Community detection algo.connectedComponents()Connected components algo.clusteringCoefficient()Local clustering coefficient algo.nodeSimilarity()Node similarity algo.triangleCount()Triangle count algo.kCore()K-core decomposition algo.density()Graph density algo.diameter()Graph diameter algo.allPairsShortestPath()All-pairs shortest paths algo.nearestNodes()Find nearest nodes algo.vectorSearch(index, vector, k)vector similarity search algo.similarNodes()Find similar nodes algo.hybridSearch()Hybrid vector + graph search algo.graphRAG()Trusted GraphRAG pipeline algo.graphRAGContext()GraphRAG context generation algo.graphRAGTrusted()GraphRAG with trust scoring algo.compoundingScoreCompounding score calculation algo.contradictions()Find contradictions in data algo.audienceProjection(weights)Audience projection analytics algo.factsByRegime(args)Facts by time regime algo.confidencePath()Confidence-weighted shortest path algo.multiModalFusion()Multi-modal data fusion
Procedure Description db.statsNode, relationship, and index counts db.schemaLabel→property mappings + relationship patterns db.labelsAll node labels db.typesAll relationship types db.propertyKeysAll property keys db.indexesIndex information db.nodeCountTotal node count db.relationshipCountTotal relationship count db.versionEngine version db.capabilitiesGPU, incremental, live-delta support db.doctor5 health checks db.fingerprintGraph hash for integrity verification db.exportFull graph as JSON snapshot db.importRestore from JSON snapshot db.import.csvImport from CSV db.clearReset graph to empty db.demoLoad sample social network db.helpComplete procedure guide db.proceduresList all available procedures db.tutorialInteractive 6-step walkthrough db.index.fulltext.queryNodes(index, query)Full-text BM25 search db.validateQuery(cypher)Query validation db.checkpointMetaCheckpoint metadata (generation, node count)
Procedure Description temporal.decay(halfLife, floor)Exponential decay temporal.trajectory()Entity trajectory over time temporal.velocity(days)Rate of change
Feature Status Syntax Notes LIVE MATCH ✅ LIVE MATCH (n:Person) RETURN nStanding query LIVE CALL ✅ LIVE CALL algo.pageRank()Incremental algorithm CREATE LIVE VIEW ✅ CREATE LIVE VIEW name AS ...Persistent live view
Feature Status Syntax Notes USE PARTITION ✅ USE PARTITION 'workspace-42'Tenant isolation SET SESSION ✅ SET SESSION ACTOR 'user-1' ROLE 'admin'Session context
Feature Status Syntax Notes CREATE SKILL ✅ CREATE SKILL name FROM PROMPT ...Declarative AI skill PROCESS NODE ✅ PROCESS NODE (n:Label)Execute skill on node
Feature Status Syntax Notes CREATE TRIGGER 🔜 CREATE TRIGGER name ON :Label WHEN CREATED RUN SKILL skill_nameFire-once event binding DROP TRIGGER 🔜 DROP TRIGGER nameRemove trigger registration WHEN CREATED 🔜 ON :Label WHEN CREATEDFires on node creation WHEN MODIFIED 🔜 ON :Label WHEN MODIFIEDFires on property change WHEN DELETED 🔜 ON :Label WHEN DELETEDFires on node deletion
Procedure Description db.triggers()List all registered triggers with label, event, and skill
Feature Status Syntax Notes CREATE PROGRAM 🔜 CREATE PROGRAM name VERSION '...' (...)Install a capability manifest DROP PROGRAM 🔜 DROP PROGRAM nameDeregister program and its triggers CARDINALITY SINGLETON 🔜 CARDINALITY SINGLETONOne executor instance globally CARDINALITY PER_SENSOR 🔜 CARDINALITY PER_SENSOROne executor per :Sensor node CARDINALITY SHARDED 🔜 CARDINALITY SHARDED BY <prop>One executor per unique property value REQUIRES GPU 🔜 REQUIRES GPU (SM >= 7.0, VRAM >= 4.0)Hardware validation at install time EVIDENCE tier 🔜 EVIDENCE SYMBOLIC | WASM | LLM | NEURALEpistemic class for program outputs
Procedure Description arcflow.programs.list()List all installed programs with provides tags and cardinality arcflow.programs.describe(name)Full manifest: I/O schema, hardware requirements, executor arcflow.programs.validate(name)Check hardware requirements without installing arcflow.programs.health(name)Executor health: status, last heartbeat, address arcflow.programs.install(manifest)Programmatic install (alternative to CREATE PROGRAM DDL) arcflow.programs.remove(name)Remove a program and deregister its triggers arcflow.programs.find_by_capability(tag)Find programs providing a given capability tag
Feature Status Syntax Notes HTTP API ✅ arcflow --http 8080REST query endpoint PostgreSQL Wire ✅ arcflow --pg 5432Connect with psql, DBeaver, Grafana MCP Server ✅ arcflow-mcpCloud chat interface integration (ChatGPT, Claude.ai, Gemini web)
Procedure Description db.observationClasses()Observation classification (observed, inferred, predicted) db.nodesByObservation.observed()Nodes with observed labels db.nodesByObservation.inferred()Nodes with inferred labels db.nodesByObservation.predicted()Nodes with predicted labels db.proofArtifactsCryptographic proof artifacts db.proofGatesProof gate information db.provenanceData provenance
Procedure Description db.liveViewsList active live views db.liveAlgorithmsActive running algorithms db.liveQueriesActive standing queries db.liveSkillsActive live skills db.subscriptionsActive subscriptions db.topicsMessage topics db.mutationsActive mutations
Procedure Description db.gpuStackGPU stack information algo.pageRankCSRPageRank — GPU-accelerated path algo.bfsCSRBFS — GPU-accelerated path algo.triangleCountCSRTriangle count — GPU-accelerated path
Procedure Description behavior.tickExecute behavior tick behavior.statusBehavior system status behavior.nodesBehavior node information
Procedure Description swarm.agentsList active swarm agents swarm.registerRegister new agent swarm.agentCountCount active agents
GQL Conformance — ISO/IEC 39075 standards compliance and TCK resultsGraph Algorithms — full algorithm catalogPrograms — CREATE PROGRAM capability manifests with hardware validation and executor wiringTriggers — CREATE TRIGGER fire-once event bindingsKnown Issues — active workarounds and edge cases ← Previous GQL Conformance Next → Glossary