remb-cli

Node.js CLI for Remb — persistent memory layer for AI coding sessions.

Install

npm install -g remb-cli

Prefer the Go binary? It's faster and has zero runtime dependencies:

curl -fsSL https://www.useremb.com/install.sh | sh

Quick Start

# 1. Authenticate (opens browser for OAuth)
remb login

# 2. Initialize a project
cd my-project
remb init

# 3. Save context
remb save -f auth -c "Implemented PKCE OAuth flow with refresh tokens"

# 4. Retrieve context
remb get -f auth

# 5. Scan a directory
remb scan --path src/auth

# 6. Start MCP server (for AI tool integration)
remb serve

Commands

| Command | Description | |---------|-------------| | remb login | Authenticate via browser OAuth or --key | | remb logout | Remove stored credentials | | remb whoami | Show authentication status | | remb init | Initialize project tracking | | remb save | Save a context entry | | remb get | Retrieve context entries | | remb scan | Auto-scan a directory | | remb link | Link features together | | remb serve | Start the MCP server |

Configuration

Project config (.remb.yml)

Created by remb init in your project root:

project: my-project
api_url: https://www.useremb.com

Credentials (~/.config/remb/credentials)

Stored securely with chmod 600:

api_key=remb_your_key_here

Environment variables

| Variable | Description | |----------|-------------| | REMB_API_KEY | Override credential file |

MCP Server

The remb serve command starts a Model Context Protocol server over stdio, exposing context and memory tools to AI assistants.

Add to your MCP client config (e.g., Claude Desktop, Cursor):

{
  "mcpServers": {
    "remb": {
      "command": "remb",
      "args": ["serve", "--project", "my-project"]
    }
  }
}

Or use the remote SSE server — no local install required:

{
  "mcpServers": {
    "remb": {
      "url": "https://mcp.useremb.com/sse"
    }
  }
}

Links

• Website
• GitHub
• VS Code Extension
• Go CLI (remb-go)