Introduction

Memax is a cloud-hosted memory layer that sits between you and your AI coding agents. It gives every agent — Claude Code, Cursor, Codex, Copilot, Windsurf — persistent, shared access to your knowledge.

The Problem

AI coding agents are powerful, but they forget everything between sessions. Each agent has its own isolated memory. Your Claude Code doesn't know what you told Cursor. Your Codex doesn't know the architecture decisions you explained to Copilot.

You end up repeating context, re-explaining decisions, and copy-pasting docs — across agents, devices, and teammates.

The Solution

Memax is a single hub where all your knowledge lives. Push from anywhere. Recall from any agent. Share with your team when ready.

You → push knowledge → Memax → agents recall it automatically

Three product surfaces:

• CLI (memax-cli) — for power users and CI/CD pipelines
• Web App (memax.app) — for all users, including non-technical
• Developer Hub (docs.memax.app) — docs, API reference, integration guides

How It Works

memax push -f ./ARCHITECTURE.md
memax push https://docs.example.com/abc123         # URL auto-detected
echo "We chose JWT for auth because..." | memax push --stdin
memax import ./docs                                  # whole directory

# Configure MCP + hooks on this machine
memax setup --all

# MCP: agents call memax_recall / memax_push directly
# Hook: prepends a static reinforcement block telling the
#       agent to use MCP (does not call recall itself)
# Pipe: universal fallback
memax recall "how does auth work?" --format text | pbcopy

memax hub create "Platform Team"
memax push -f ./runbook.md --hub platform-team   # route this push to the team hub
memax hub invite create --role contributor       # generates an invite link

Quick Start

Get set up in under two minutes:

# Install the CLI
npm install -g memax-cli

# Authenticate
memax login

# Push your first document
memax push -f ./ARCHITECTURE.md

# Set up agent integrations on this machine
memax setup --all