OPEN CORE

Open core, and honest about it.

We call it fair-code, not open source — because it isn’t. The core is free to use, read, modify, and self-host. What we license is the hosted graph and meeting service. That’s how we stay alive.

Everything in the core today stays in the core, free, forever. We will never move a feature you already have behind the licence. New commercial features may be licensed — nothing you already use will be taken away.

Our three goals

As permissive as possible.

Free to use, read, modify, and self-host. Almost everything you want to do is allowed.

Able to sustain the people building it.

One boundary keeps the maintainers funded so the project keeps shipping.

As clear as possible about what’s allowed.

One sentence, real examples, no lawyer required to understand the line.

The licence, plainly

The core ships under our fair-code licence — source-available, in the SUL family (the same shape n8n uses). The full text lives in LICENSE in the repo; read it before you build on Mnema.

Use Mnema for your own work or your company’s internal work, freely. What you can’t do is resell Mnema itself as a hosted service to your customers.

You can
  • Use Mnema for your own work
  • Use Mnema for your company's internal work
  • Deliver consulting to a client using your own Mnema
  • Read, modify, and self-host the core
You can’t
  • Resell Mnema itself as a hosted service to your customers
  • White-label Mnema and sell it as your own product

What’s free vs licensed

Self-host community is the full core, free: the author, flows, MCP, and hybrid search. Run it on your own machine, keep your keys.

Licensed is the hosted graph and meeting service — the parts we operate for you. That’s early access today (contact us), not self-serve, because we’re still building it.

Cloud tiers and prices are on the pricing page.

Why fair-code

Freely-licensed work can be captured wholesale: a cloud provider takes it, sells it as a service, and returns nothing to the people who built it. Fair-code exists so the value flows back to the maintainers — the boundary is small and the code stays readable.

And there’s a reason particular to Mnema: this is a brain. It holds your decisions, your meetings, your agents’ work. You should be able to run it yourself and read exactly what it does. A second brain you can’t inspect isn’t yours.

This licence is not OSI-approved. If OSI-approved licensing is a hard requirement for you, Mnema isn’t the right fit, and we’d rather tell you here than waste your time.

Project status

v0.1.0
Latest release · 2026-07-07
6
GitHub stars
5
Open issues
7
Contributors

Live from the GitHub API at build time.

Contributing

Contributions are welcome and encouraged. There’s no CLA — you own what you contribute. Start with a good first issue, read CONTRIBUTING.md for the dev setup, open a PR, and a maintainer reviews it in the open.

The model works: n8n runs on fair-code with 129k+ GitHub stars, and only about 18% of its contributions come from the core team — fair-code didn’t suppress the community, it inverted it.

Security

Found a vulnerability? Report it privately per SECURITY.md. We acknowledge reports quickly and keep you updated through the fix — please don’t file security issues in the public tracker.

Governance

BOPPL maintains Mnema. Decisions are made in the open on GitHub — the issue tracker and changelog are the record. We don’t claim a foundation we don’t have; it’s a company maintaining a source-available project, honestly.

Self-host

The core self-hosts in about four minutes. Follow the self-host guide — clone, set your keys, run. Your data, your machine.