Terms of service

Last updated to be completed — effective date

Published while still being completed, and not yet offered as a binding agreement. Every highlighted to be completed — like this line is a value only the operator of this service can supply, and each one is outstanding. The liability, warranty and governing-law sections in particular have not been reviewed by a lawyer, so until they are, treat this page as a good-faith description of how the service is meant to work rather than as a contract. The description itself is accurate: it was written from how the software actually behaves.

The short version

Mainmind holds your organization's knowledge and lets the AI agents you choose read and add to it. Your content stays yours. You are responsible for what your agents do with the access you give them, and for the outside services you connect. We will keep the service running as best we can, and we do not promise it is perfect.

1. Agreeing to these terms

These terms are between you and to be completed — legal entity name, of to be completed — registered address ("we", "us"). By creating a Mainmind organization, joining one, or connecting an agent to one, you agree to them. If you are agreeing on behalf of a company, you confirm you may bind it.

Our privacy page forms part of these terms and describes what we store and for how long.

2. What Mainmind is

Mainmind stores an organization's durable knowledge — its processes, records, decisions, role charters and lessons — as plain Markdown in a Git repository, and serves it to AI agents over the Model Context Protocol. It also keeps a ledger of what those agents did.

Mainmind is not a system of record for anything a regulator, an auditor or a court requires you to keep. Keep your own copies of anything that matters in that way. It is not a backup service, and it gives neither legal, financial, tax nor medical advice, whatever an agent connected to it may produce.

3. Your account and your team

You sign in with GitHub. Keep that account secure; anything done through it is treated as done by you.

A Founder invites other members and sets what each may read and do. If you invite someone, you are responsible for that decision. Roles and permissions are enforced on every operation, so a change takes effect at once — but a person or agent who already read something has already read it, and revoking access afterwards cannot recall it.

Machine members exist for runners and bots. Their credentials are yours to protect and to rotate. Do not put a machine credential where a person's credential belongs, or the reverse.

4. Your content

Your content stays yours. Everything your organization writes into Mainmind — documents, decisions, records, notes, uploaded files — remains your property. We claim no ownership of it.

You grant us only the permission we need to run the service for you: to store your content, copy it between the stores described on the privacy page, index it for search, mirror it to the GitHub repository you nominate, send the relevant parts to the AI models that answer your agents' questions, and show it to the members you have authorised. That permission ends when you stop using the service, save for copies in backups and logs that age out on their own.

You confirm you have the right to put into Mainmind whatever you put into it, including anything you upload that belongs to somebody else.

The service itself — the software, the site and the documentation — remains ours. to be completed — the licence for the published plugins and skills, which ship under MIT in a separate repository

5. Agents acting for you

This is the part that differs from ordinary software, so it is worth being explicit.

An agent acts with the authority you gave it. What it writes, deposits, decides or sends is attributed to the member it authenticated as, and is your responsibility, not ours.
Agents make mistakes. Model output can be wrong, incomplete, or confidently both. Review anything consequential before acting on it. Mainmind deliberately stops an agent and asks a person before an action that spends, sends or signs, but that guard covers the actions the product knows about — not every possible consequence.
Third-party agents are third parties. When you connect Claude Code, Cursor, Grok, Muse or any other client, what that client reads from your organization goes to whoever operates it, under their terms and their privacy policy. We are not responsible for what they do with it, and we cannot retrieve it.

6. Services you connect

You may connect outside services so your agents can use them. You are responsible for having the right to connect them and for what your agents do through them. We pass requests through; we do not vet the service, guarantee its availability, or take responsibility for what it charges you or does with your data.

Revoking a connection stops future use. It does not undo an action already taken.

7. Acceptable use

Do not use Mainmind to break the law, to store or move material you have no right to, to attack or overload the service or anyone else's, to work around its permission model or another organization's boundaries, to probe the service for weaknesses without our written agreement, or to resell access to it.

Do not put credentials into a place meant for prose. The product provides a place for credentials; use it.

8. Availability and change

Mainmind is a service under active development, and parts of it are explicitly early. We may add, change or remove features. Where a change would break something you depend on, we will try to say so first on the updates page.

We do not commit to an uptime figure. to be completed — a service-level commitment, if one is ever offered

9. Fees

to be completed — pricing and billing terms. The site says "free to start"; this page deliberately says nothing more until what follows is decided

10. Ending it

You may stop using Mainmind at any time. Before you do, clone your company file and export your uploaded originals — see export options.

We may suspend or end access if you breach these terms, or if we are required to. Where we reasonably can, we will warn you first.

Be aware of what ending does not do, described in full on the privacy page: revoking a member marks the record rather than erasing it, and there is no self-service deletion of an organization today. To request erasure, write to to be completed — an address for privacy requests.

11. Disclaimers

The service is provided as it is and as it is available. To the fullest extent the law allows, we disclaim all implied warranties, including merchantability, fitness for a particular purpose and non-infringement. We do not warrant that the service will be uninterrupted or error-free, or that anything an AI model produces through it will be accurate.

Nothing here excludes liability that cannot lawfully be excluded.

12. Liability

to be completed — the liability cap and its exclusions, set by a lawyer against the chosen governing law and the pricing model. A cap tied to fees paid means something different while the service is free

Neither of us is liable to the other for indirect or consequential loss, or for lost profits, revenue or data, to the extent the law allows that to be agreed.

13. Indemnity

You will cover us against claims arising from content you put into Mainmind that you had no right to put there, from your use of the service in breach of these terms, and from what your agents do with the access you granted them.

14. Governing law

These terms are governed by the law of to be completed — jurisdiction, and the courts of to be completed — jurisdiction have exclusive jurisdiction over any dispute. to be completed — whether consumer protections in a user's own country override this

15. Changes to these terms

We may update these terms. Material changes will be announced on the updates page and dated above. Continuing to use the service after a change means you accept it.

16. Contact

to be completed — a contact address — to be completed — legal entity name, to be completed — registered address.