Open-source AI memory framework

Build AI wheels that roll forward forever

Never repeat yourself to ChatGPT, Claude, Copilot… ever again.

Wheelwright is a hub-and-spoke memory layer that persists project context across any LLM. Drop a WAI-Spoke folder in your project and every assistant starts with the same shared memory—code or not.

Bootstrap a project Clone the framework

What is Wheelwright?

Wheelwright is an open-source system for persistent AI context. A single WAI-Spoke folder carries your architecture, decisions, and working state so any model can pick up where the last one left off.

Hover or tap a pain point to reveal the WAI benefit.

Memory resets every session Hub + Spoke memory keeps context

Switching LLMs = blank slate LLM-agnostic continuity

Context is expensive to rebuild 50–80% fewer tokens

Knowledge trapped in one repo Works beyond code

Docs drift from reality Single source of truth

Teams lose momentum Open-source, portable context

How it works + why it matters

How it works

Hub + spoke memory

The hub consolidates the shared project memory, while spokes store specialized context for tasks like coding or review.

LLM-agnostic by design

Use ChatGPT, Claude, Grok, or your own models. They all read the same wheel and keep it rolling.

Works beyond code

Use it for research, writing, product strategy, design systems, onboarding, or any project where continuity matters.

The pain it solves

Tokens evaporate every session

You pay to re-explain architecture, tech stack, decisions… over and over.

Switching LLMs = blank slate

Claude in the morning, ChatGPT at night—neither remembers what you already built.

Wheelwright fixes it in one file

A single WAI-Spoke/ folder that every AI loads automatically.

Proof in the numbers

Measured across real feature work: fewer tokens, faster delivery, and zero rework.

Metric	Without WAI	With WAI	Saved
Tokens / feature	≈ 15 k	≈ 5 k	67 %
Time / feature	3–4 h	1 h	2–3 h
Re-work cycles	2–3	0	100 %

Start from a spoke or a hub

Start with a single project, or set up a hub if you plan to run multiple spokes.

Start with a spoke

Clone the framework: git clone https://github.com/wheelwright-ai/framework
Make the CLI executable (and optionally add it to your PATH).
Initialize a spoke in your project with WAI-CLI init.

Start with a hub

Clone the framework: git clone https://github.com/wheelwright-ai/framework
Create your hub: WAI hub create.
Add spokes from any project with WAI-CLI init.

Open bootstrap guide →

Want to hear updates on new features?

Host your WAI files online, sync across machines, share with your team—be first to get access.

Alpha invites rolling out weekly; local version always free.