Memorav1.3.0Latest
Star on GitHub
DocsGet started

DocsStart

Get started

Install Memora, claim it with a one-time code, choose the vault, connect your first AI agent and give it the recommended instructions.

v1.3.0Latest
On this page

Memora runs on your own server: one process, one port, your notes as plain Markdown files. This guide takes you from nothing to an AI agent that remembers, in about ten minutes.

1. Install Memora#

One-command installer (Linux with systemd)#

On a Linux server (x86_64 or aarch64), run:

sh
curl -fsSL https://raw.githubusercontent.com/Codrihub/memora/main/scripts/install.sh | sudo sh

The installer downloads the release for your CPU, verifies it against the release SHA256SUMS, creates a memora system user and installs a hardened memora.service unit. Data lives in /var/lib/memora. If you prefer to read the script first, download it, open it, then run sudo sh install.sh (add --behind-proxy when a local reverse proxy fronts Memora).

Docker Compose#

bash
git clone https://github.com/Codrihub/memora.git && cd memora
cp .env.example .env
mkdir -p data state && sudo chown -R 1000:1000 data state
docker compose up -d

Copying .env.example is optional: every value has a default. The image is ghcr.io/codrihub/memora for linux/amd64 and linux/arm64. Every option, path and flag is in Install.

2. Get a one-time setup code#

A fresh instance cannot be claimed by whoever finds the port. Open the operator console on the server:

sh
sudo memora-cli

With Docker, open the console inside the container instead:

sh
docker compose exec -it gateway memora-cli

Type /web at the console prompt. It opens a short setup window and prints a one-time code. If the window expires unused, the instance locks until you run memora-cli setup or memora-cli unlock on the server.

3. Create the owner account#

Open http://<host>:8400/login in your browser, enter the code and create the owner account. The server listens on port 8400; put a reverse proxy with TLS in front of it before you expose it to the internet (see Deployment).

4. Choose the vault#

Pick the folder that holds your notes: /var/lib/memora/vault with the native installer, /data in Docker. Every note is a .md file with optional frontmatter and [[wikilinks]], and every change is committed to git, so you can always see, diff and restore a revision.

5. Connect an agent#

Create an API key for each agent, in the web UI under Agents & API keys or with /keys in the console. A key is shown once. Then type /mcp in the console for a ready-made config, or add Memora yourself. Claude Code:

bash
claude mcp add --transport http memora https://memory.example.com/mcp \
  --header "Authorization: Bearer memora_live_REPLACE_WITH_YOUR_KEY"

Clients with an mcpServers JSON config:

json
{
  "mcpServers": {
    "memora": {
      "type": "http",
      "url": "http://localhost:8400/mcp",
      "headers": {
        "Authorization": "Bearer memora_live_REPLACE_WITH_YOUR_KEY"
      }
    }
  }
}

Clients that support MCP authorization (such as Claude or ChatGPT) can connect with only the URL: you approve them in the browser and they show up as keys you can restrict or revoke. Details for every client are in MCP & agents.

6. Paste the agent instructions#

Connecting gives an agent the tools; instructions tell it when to use them. Copy the recommended prompt from Agent instructions into your client's rules or custom instructions, and your agent will search Memora before it answers and save what is worth keeping.

Help inside Memora#

  • In the console, type /docs for the same guides in your terminal.
  • In the web app, open the Documentation page for these guides and the agent instructions with Copy buttons.
  • Watch your agent work in Activity: every read, write and search is logged as it happens.

Next steps#