Skill Operations

Describe a relationship rule in natural language. ArcFlow compiles it once — using your configured LLM endpoint — into a graph pattern stored in the skill. Every subsequent PROCESS NODE and REPROCESS EDGES executes the compiled pattern directly: no model calls, no token cost, sub-millisecond per node. The LLM is a compiler frontend. ArcFlow is the runtime.

Operations#

Introspection#

-- List all registered skills
CALL arcflow.skills()
  YIELD name, tier, allowed_on, threshold, active, version
 
-- Provenance chain walk — trace a node's derivation back through skills
CALL db.provenance(42)
  YIELD nodeId, label, name, confidence, depth
 
-- Bundle export / import
CALL arcflow.skills.export('my-pack', '1.0.0') YIELD json
CALL arcflow.skills.import(json)               YIELD name, version, skill_count

Lifecycle#

DROP SKILL summarize
DROP REACTIVE SKILL match_similar

See Also#

• Using Skills — step-by-step guide with real examples
• Programs — bundle skills, triggers, and executor endpoints into an installable manifest
• Triggers — CREATE TRIGGER to run a skill automatically when a graph event fires
• Observations & Evidence — the observation model that skills write to
• Confidence & Provenance — how skill edges carry confidence scores
• Agent Governance — verification state machine and receipts
• Use Case: Knowledge Management — skills in an entity extraction pipeline