agentmemory-architecture

What problem does it solve? Understanding how agentmemory stores, indexes, and retrieves memories end to end is difficult without a map of its internals. This Skill explains the iii engine primitives, the hybrid retrieval pipeline, the storage lifecycle, and the port layout so you can reason about the system, extend it, or answer how it works under the hood. ## Core Features & Use Cases - iii Engine Primitives: Explains that everything is a function, trigger, or worker state on the iii engine, with new capabilities added as functions plus triggers. - Hybrid Retrieval Model: Covers BM25 keyword search, on-device vector similarity, and graph expansion over linked concepts, with no API key required by default. - Storage, Lifecycle, and Ports: Details memory fields, sessions, commit links, the capture-compress-consolidate-forget lifecycle, and the port block anchored at REST 3111. - Use Case: When extending agentmemory with a new capability, use this Skill to confirm you should register a new mem::* function and an api::* HTTP trigger rather than bypassing the engine. ## Quick Start Ask the agent to explain how agentmemory stores and retrieves memories end to end using this architecture Skill.