Reliquary Pre-alpha

Shared context and credentials for AI teams

Your agents propose. You approve.

Stop emailing API keys. Stop re-explaining the project.

One shared vault of context and credentials for your team and every AI tool you use. Claude, ChatGPT, Cursor and Claude Code read the same approved context. Secrets stay out of the chat.

EU-hosted. Bring your own model. Invite-only while in beta.

Pre-alpha: things change and may break; data is backed up daily. See the roadmap.

Open brief.md Canon

Proposed by Claude Code for Ana · needs 1 approval

An agent proposes a change to canon. It lands only when a person approves.

The problem

How it works

  1. Connect any AI tool with one URL

    Paste Reliquary's MCP URL into Claude, ChatGPT or Claude Code and sign in. Other MCP clients use a scoped, expiring token you can revoke.

  2. Agents read and propose

    Open folders take notes directly. Canon folders change only when a person approves, with a quorum if you want one. Every change is in an append-only log.

  3. Run with secrets, never show them

    reliquary run --env development -- npm start puts variables into one process. No AI tool ever receives a value.

What makes it different

People sign off on what becomes fact

Agents from any vendor read the same approved context over MCP and propose changes. A person, or a quorum of people, approves before anything becomes canon.

The limits live in the database

An agent acts as its person, minus a ceiling: it can't approve, change the rules, add people or reveal a secret. Postgres row-level security enforces it, and hostile tests check it on every push.

Context and credentials in one vault

Memory tools don't hold secrets and secret managers don't hold context. A project needs both, with one invite and one access log.

EU-hosted, bring your own model

Data sits in Frankfurt. Reliquary runs no model and resells no inference, so it stays cheap and your AI provider stays your choice.

Who it's for

Agencies and consultancies

One vault per client: context the whole team's agents read, approvals on anything that becomes fact, credentials used without being pasted, and one log to show the client.

Solo builders

Several AI tools and machines, one source of truth. Stop copying CLAUDE.md between tools and keeping .env files where any coding agent can read them.

Small product teams

Some people use Claude, some ChatGPT, some Cursor. Give them the same reviewed context, and keep keys out of chat messages.

What Reliquary is not: an AI model or chatbot, search over your Drive and Slack, or a memory that writes itself.

Works with Claude, Claude Code, ChatGPT, Cursor and any MCP client. For client work, invites, credential requests and client guests are coming.

What your agent can't do

Enforced in the database, not the prompt, and tested on every push. Data is hosted in the EU; variable values are encrypted with a key the database never holds. We say plainly what we can't promise: an agent can read what reaches its process, and we, as the operator, could technically decrypt values.

Read how security works

Pricing

Early access: free while in beta. Prices below are what we expect to charge after the beta. Nothing is billed today, and you'll hear from us well before anything is.

Every plan: unlimited agents and AI tools, approvals and quorum, the agent ceiling enforced in the database, the activity log, and export.

Questions

Is Reliquary an AI model or a chatbot?

No. Reliquary runs no model. It holds context and credentials, and the AI tools you already use connect to it over MCP.

Do I need to host anything?

No. Reliquary is hosted in the EU, and nothing depends on your machine staying on.

Which AI provider sees my data?

Only the ones you connect, and only what their agent reads through your account. They process it under your own agreement with them. See sub-processors.

Can my agents see my secrets?

Not through Reliquary: no MCP tool returns a variable's value. reliquary run puts values into one process, and an agent that can run commands in that process could read them there. The security page says what that means.

How is it different from Claude Projects or ChatGPT memory?

It works across vendors, people approve changes before they become fact, and it holds credentials as well as context.

Can I leave?

Yes. An owner can export a whole vault as plain markdown files at any time, and delete it for good.

Is it for regulated data?

Not yet. Reliquary has no SOC 2 report or SSO today. Don't store health or payment card data in it.

Try it on your next project

Reliquary is invite-only while in beta. Tell us about your team and the AI tools you use, and we'll set up your first vault with you.