DiluxOne

Open source · native MCP

Stop explaining the same things to your AI every day.

Diluxite is not another note editor: it is the memory your AI reads, writes and searches by meaning. Store knowledge once and Claude, Copilot, Codex and any MCP client recall it across sessions and across tools.

  • A loose .md file cannot do this: your AI can neither remember it nor search it semantically.
  • Hybrid search: “the Microsoft cloud” finds your note about Azure.
  • Ten native MCP tools, no adapters and no glue code.
Start FREERequest private preview

All of DiluxOne is in private preview. Registering is free, asks for no card and commits you to nothing.

Knowledge graph connected to AI clients

What you get

Knowledge that accumulates instead of evaporating

Your AI remembers your context

Decisions, notes and projects stay available without re-explaining everything in every new session.

Search by meaning

Hybrid: keyword and semantic combined, fused with RRF. You can force keyword-only or semantic-only whenever you need to.

Wikilinks, backlinks and graph

Markdown notes that link to each other, with a graph view and tags to navigate what you have built.

One memory, all your agents

Native MCP server with ten tools: search, read, write, append, list, backlinks and more.

Live collaboration

Real-time collaborative editing with cursors, presence and conflict-free merges. Multi-user by workspace.

Corporate-grade security

OIDC SSO, passkeys, TOTP 2FA, audit log with retention and per-workspace isolation.

How it connects

The same brain for all your tools

Diluxite exposes a native MCP server with per-user tokens. Connect your agents once and they share a single persistent memory — what Claude wrote yesterday, Copilot reads today.

  • Core edition, open sourceAGPL-3.0, self-hosted, with a smart installer and a management menu: update, reconfigure, backup, restore and uninstall.
  • Cloud editionMulti-user and hosted, built on exactly the same engine. Not a different product wearing the same name.
  • Six UI languagesSpanish, English, Portuguese, Italian, Catalan and Chinese.
  • Trash with soft deleteRestore or purge notes, with a per-workspace trash view.
  • Solid stackNode + TypeScript, Fastify, PostgreSQL with pgvector, React. Over 850 tests green.
MCP clients sharing one memory

In action

A second brain that actually looks like one

Real captures over a workspace with 1,492 notes. This is what knowledge looks like once it stops being scattered.

Your space Your memory, measured 1,492 notes, 30 folders, 60 tags, 625 wikilinks. Not a folder of loose files: a body of knowledge with structure, and that panel tells you how much it has grown.
Editor Markdown on the left, the result on the right A real ADR with tags, status, context, decision and alternatives. The [[wikilinks]] connect to other notes and generate backlinks on their own: next time, your AI gets there without being told.
Graph You can see where the knowledge concentrates The 1,502-node graph shows the real clusters: here, every postmortem hanging off a Knowledge Hub. When something repeats three times in the graph, it stopped being an incident and became a pattern.
Search It finds what you meant Search “azure” and 14 matches across 14 notes show up, with the snippet and tags highlighted. And semantic search goes further: it finds the note even if you typed “the Microsoft cloud”.
MCP connection Connect your AI in two clicks Generate a named token with an expiry, paste it into Claude or Copilot, and done: your agent can now read from and write to your memory. Tokens are revoked one at a time or all at once.
Organization And when the brain belongs to the whole team Members by email with roles, workspaces with isolation, org tokens, embedding configuration and an audit log. What it takes for a second brain to be corporate rather than an experiment.

Pricing

From personal use to your organization’s brain

Start on your own for a hundred dollars a month, or give your whole company one shared memory.

FREE · self-hosted

Core edition, open source

USD 0 /mo

  • AGPL-3.0 license, single-user
  • Smart installer with management menu
  • Native MCP server and hybrid search
  • Official Docker image
  • Community support
Start FREE

Installation and configuration are on you, at your own risk.

Personal

Your second brain, hosted

USD 100 /mo

  • Up to 1,000 documents
  • All your agents against one shared memory
  • Backups and upgrades included
  • No servers to maintain
Request private preview
Private preview

Organization

Your company’s second brain

USD 200 /mo · from 10 users

  • Includes 10 users; USD 20/mo per additional user
  • Unlimited documents
  • SSO and corporate security standards
  • Workspaces with isolation and sharing
  • Audit log with configurable retention
  • Direct support from the team
Request private preview
Private preview

Request your access

Tell us whether it is for you or for your organization and we will enable your space. You can also start self-hosted today.

  • Free, and no credit card.
  • We email you as soon as a seat opens.
  • Meanwhile you can keep using the open source edition.

By submitting you agree to be contacted about the private preview. No cost, no credit card, no commitment.

Frequently asked questions

How is this different from a folder of Markdown files?
Your AI can use it on its own. A loose .md cannot be searched by meaning and is not remembered across sessions; Diluxite exposes the memory over MCP, with hybrid search, links and permissions.
Which AI clients can I connect?
Any that speaks MCP: Claude, Copilot, Codex and the rest of the ecosystem. Each user gets a token and they all share the same memory.
How are documents counted?
The Personal plan includes up to 1,000 documents, plenty for intensive individual use. The Organization plan has no document limit.
Do my notes get locked in?
No. They are Markdown, the Core edition is open source, and you can export or move to self-hosted whenever you want.