What problem does it solve? In multi-agent systems, different agents often encounter the same real-world entity (customer, company, product) from different sources and create duplicate or conflicting records. This Skill operates a shared identity graph so every agent deterministically resolves the same record to the same canonical entity_id, preventing duplicate charges, conflicting actions, and cascading errors. ## Core Features & Use Cases - Deterministic Entity Resolution: Normalizes fields (emails, E.164 phones, nickname expansion), blocks candidates, scores field-level matches, and returns a canonical entity_id with confidence scores. - Evidence-Based Merge Proposals: Proposes merges and splits with per-field evidence and reason codes so other agents or humans can review before execution. - Conflict Detection & Audit Trail: Flags conflicting proposals between agents, tracks every mutation with optimistic locking, event history, and rollback support. - Use Case: A billing agent and a support agent both encounter "Bill Smith" and "William Smith" at the same email. The operator resolves both to one canonical entity with 0.94 confidence, preventing a duplicate customer record and double charge. ## Quick Start Ask the agent to resolve an incoming customer record against the identity graph and return the canonical entity_id with confidence and match evidence.
