Skip to Content
remimem

Quickstart

Index a project and connect your coding agent in about two minutes.

You need Node.js 22.13 or newer. Check with node --version.

Set up the project

In the root of the repository you want your agent to understand, run:

npx -y remimem init

This does four things:

  1. Creates .remimem/config.json (empty: every setting has a default).
  2. Adds .remimem/index.db* to .gitignore, so the index stays local.
  3. Builds the index: it parses your code, embeds it, and builds the call graph. The first run downloads a 23 MB embedding model.
  4. Prints the configuration for Claude Code, Cursor, and VS Code.

The output starts like this:

Created .remimem/config.json (empty: every setting has a default).
Added .remimem/index.db* to .gitignore.
Privacy: everything runs on this machine. No code leaves it and there is no telemetry.
Indexed 570 of 570 files (0 unchanged, 0 removed) in 38.04s. Embedded 5300 chunks ...

Connect your agent

claude mcp add remimem -- npx -y remimem serve

Other clients and older VS Code versions are covered in Connect a client.

Check the tools are there

In Claude Code, enter /mcp: remimem should be listed as connected with 9 tools. In Cursor, look under Settings, MCP. In VS Code, open the Command Palette and run Chat: Open Customizations, then MCP Servers.

Ask a question

Ask your agent to start a task, for example:

Use remimem's get_context for "add rate limiting to the login endpoint",
then make the change.

The agent gets the relevant code in a few thousand tokens instead of reading whole files. When it learns something worth keeping, it can call remember, and the next session sees it.

Try it without an agent

Every tool has a command-line twin. To see what an agent would get:

npx -y remimem context "check a password at login" --budget 1000
npx -y remimem search "password hashing"
npx -y remimem stats

What next

On this page