---
title: Agent Memory System
emoji: 🧠
colorFrom: green
colorTo: yellow
sdk: static
app_file: index.html
pinned: false
license: mit
short_description: Evidence-backed continuous memory for coding agents
---


# Agent Memory System

Hugging Face mirror for [Agent Memory System](https://github.com/rrrrrredy/agent-memory-system), a local-first, evidence-backed continuous memory system for Codex, Claude Code, OpenCode, and supported DeepSeek Harness event surfaces.

- Source snapshot: [`dc53ddb98b9eeb29ca0e7aad57841d53e126ff9c`](https://github.com/rrrrrredy/agent-memory-system/commit/dc53ddb98b9eeb29ca0e7aad57841d53e126ff9c)
- Canonical website: [rrrrrredy.github.io/agent-memory-system](https://rrrrrredy.github.io/agent-memory-system/)
- Current release: [v0.3.1](https://github.com/rrrrrredy/agent-memory-system/releases/tag/v0.3.1)

Raw evidence remains local by default. Only reviewed and redacted portable memory is intended for Git synchronization. The published synthetic capability suite is not a claim of longitudinal learning efficacy; prospective study support is intentionally separate from ordinary memory operation.