Audit-Ready Memory

Audit-Ready Memory is a local-first memory layer for AI agents with transparent logs, deterministic replay, and explicit deletion workflows.

The goal is to make agent memory inspectable, reproducible, and accountable in practice, rather than hidden inside opaque model state or external services.

This work is funded by SIDN Fund.

What the project focuses on

• Time-stamped memory records instead of hidden internal state
• Transparent audit trails for memory writes and reads
• Deterministic replay for reproducibility and debugging
• Explicit deletion workflows and lifecycle controls
• Local-first operation suitable for public-interest and privacy-sensitive settings

Documentation status

This documentation set is being established alongside the repository. It currently covers the project direction, documentation workflow, and release/deployment setup. Technical reference material will expand as implementation lands in the repository.