Skip to main content

​
Prerequisites

  • Node.js 20+
  • An MCP-compatible client (Claude Desktop, Claude Code, Cursor, VS Code, etc.)

​
Installation

If you have @cohaku/cli installed globally: 
cohaku init
This will:
  1. Initialize the database
  2. Auto-detect installed AI tools (Claude Code, Cursor, VS Code, Windsurf, etc.)
  3. Register the MCP server with each detected tool
See Integrations for the full list of 12 supported tools.

​
Manual Setup

For manual configuration, see the individual integration guides:

Claude Code

Claude Desktop

Cursor

VS Code

Windsurf

OpenCode

​
Verify Connection

Start your MCP client and check that the Cohaku tools appear. You should see 24 tools including add_memory, search_memories, get_context, session_start, generate_claude_md, etc.

​
First Usage

Once connected, your AI agent can immediately start using Cohaku: 
Agent: "Remember that this project uses React 19 with TypeScript"
→ calls add_memory({ content: "Project uses React 19 with TypeScript", layer: "rule", type: "rule" })

Agent: "What do you know about this project?"
→ calls search_memories({ query: "project" })
→ returns the stored memory with relevance score

​
CLI Commands

If installed globally, the cohaku CLI provides commands for all memory operations:
CommandDescription
cohaku initInitialize database and register MCP with detected tools
cohaku memory addStore a new memory
cohaku memory searchSearch memories
cohaku memory contextLoad prioritized context
cohaku graph entity addAdd a knowledge graph entity
cohaku graphSearch the knowledge graph
cohaku generate <tool>Export memories to editor config
cohaku statusShow memory statistics
See the CLI reference for the full command list.

​
Configuration

The default database path is ~/.config/cohaku/memory.db. This can be configured via environment variables:
VariableDefaultDescription
COHAKU_DB_PATH~/.config/cohaku/memory.dbSQLite database file path

​
Next Steps

  • Learn about Memory Layers to understand how memories are organized
  • Browse the MCP Tools reference for all available tools
  • Explore the CLI reference for terminal-based memory management
  • See all Integrations for supported AI coding tools