Skip to content
Codrihub Tech, home
← All projects
Released · actively developedv2.0.1AGPL-3.0-or-later

Memora

Second brain for your AI

One shared, self-hosted memory for your AI agents. Notes stay plain Markdown files on your own server, and Claude Code, Codex, Cursor or any other MCP client read and write them through one endpoint.

Memora overview screen: notes, links, folders and agents, what your AIs did today, what needs you, and the latest actions

What it does

  • Plain Markdown

    Every note is a .md file in a folder you own, with [[wikilinks]] and tags. No proprietary format, no lock-in.

  • One endpoint for every agent

    Claude Code, Codex CLI, Cursor, ChatGPT and other MCP clients connect to the same endpoint, each with its own key, and share one memory through 75 tools.

  • Watch agents work

    A live panel, the Activity summary and its Timeline show which notes each agent reads and writes, as it happens, and what it tried but was not allowed to do.

  • Brain, graph and canvas

    Explore the vault as a 3D brain with folders as regions, as a link graph, or as cards on a canvas.

  • You decide who sees what

    Per-agent permissions (can only read, can read and change, or full control, and which tools), privacy rules that hide folders from chosen agents, and access rules by network.

  • Links and pins that follow

    Drag a note or folder where it belongs, or pick a folder from a searchable tree. Wikilinks and pins follow every move and rename, and agents can pin what matters too.

  • Version history and a Trash

    With the vault in git, every note keeps its versions: see who made each one, compare it with the current text and restore it. Deleted notes go to the Trash first, and you choose where a restored note goes.

  • Tags and patterns

    Pick tags to see their notes at once, merge or rename them in bulk, and find notes by a pattern of tags joined with and, or and not.

  • Ready-made agent instructions

    Copy instructions in the style you want (balanced, private, read only, session journal) into Claude Code, Codex, Cursor or any client.

  • Self-hosted

    One program, one port, on your laptop, home server or VPS. Your memory stays where you put it.

Memora's live panel: the latest reads and writes of your AIs, which agents are working, and links to all activity and the agents page
The live panel: what each agent is doing, right now.
An agent's access settings in Memora: can only read, can read and change, or full control, with the tool list under Advanced
Each agent gets only the access you give it.
Memora's brain view: notes drawn as a 3D brain, with folders as labelled regions
The brain view. Each point is a note; each region is a folder in your vault. Screens show sample data.

Install

Install, run, then open the link that memora-cli setup prints and create your account. A native Linux service and every option are in the install guide.

npmLinux x86_64 / arm64 and Windows x64, Node 22+
npm install -g @codrihub/memora
memora
# in a second terminal:
memora-cli setup
DockerAny OS with Docker, including macOS
docker run -d --name memora -p 8400:8400 \
  -v memora-data:/data -v memora-state:/state \
  ghcr.io/codrihub/memora:latest
docker exec -it memora memora-cli setup
Full install guide →

The film

Ninety seconds on what Memora does, rebuilt from the real interface.

Play the 90-second Memora film on memora.codrihub.tech
Watch on memora.codrihub.tech →

Try it, break it, tell us.

Memora is open source and in active development. Bug reports, ideas and pull requests are all welcome on GitHub.