Agent Memory ArchitectureAgent Memory Architecture

Initial release of Agent Memory Architecture: a zero-dependency, file-based memory system for AI agents.

- Introduces a structured five-layer memory system: Hot (active context), Warm (long-term memory), Daily notes, Topic files, and Archive.

- Provides clear file naming conventions and example directory structures.

- Includes detailed templates and curation rules for each memory layer.

- Describes maintenance practices for memory hygiene, daily note rotation, topic file management, and quarterly archiving.

- Works with any agent framework

requires no APIs, databases, or external tools.