Browse / General-purpose / memforks
memforks
memforks-dev
Git for AI agent memory
Open memforks-dev/memforks → Does not match the four rules.
Works with any agent
ships config for cursor, agents.
In its own words
git for AI agent memory
A versioned commit graph; branches, forks, merges, time-travel for everything your agents learn.
Contents
- Overview
- The Problem
- What it is
- Who it's for
- Getting Started
- Quick start
- How the agent uses it
- Configuration
- memfork init --quick
- memfork install
- Built with MemForks
- Adapters
- Vercel AI SDK
- LangGraph
- Reference apps
- memforks-chat
- memforks-research
- Visualizer
- Architecture & Internals
- How it uses Walrus, Sui & SEAL
- Gas sponsorship
- Artifact storage
- Repository structure
- How it compares
- Project
- Status
- Vision
- What's next
- Development
- Documentation
- Links
- License
Overview
The Problem
Remembering is the easy part now. Plenty of good tools can persist what an agent learns across sessions. What none of them give you is version control over that memory, and without it, persistence quietly starts working against you.
Persistent memory layers like MemWal solve the storage half: durable, encrypted, semantically-recalled memory on Walrus. But persistence alone leaves memory as a flat, linear append-log, and that breaks down the moment agents do real work:
Quoting 6% of the README. Read the rest on GitHub ↗
Where it came from
What this copy built: +apps +contracts +packages +research +services +spikes — top-level directories this copy added on top of the original.
And a tool: +.cursor — configuration the original does not ship.
Read from the repository's git tree on 2026-08-17.