Skip to content
← Back to site
Soleri | Docs
ENUKIT

Getting Started

Prerequisites

Section titled “Prerequisites”
  • Node.js 18+ — check with node -v
  • Claude Code or Cursor — your AI-powered editor
  • npm — ships with Node.js

Create Your Agent

Section titled “Create Your Agent”

One command to scaffold a fully functional agent:

Terminal window
npm create soleri my-agent

The interactive wizard asks for:

PromptWhat it means
Agent nameYour agent’s identity (e.g., “sentinel”, “architect”)
RoleOne-line description of what it does
DomainsKnowledge areas — frontend, backend, security, or custom
Persona voiceHow the agent communicates — professional, casual, technical

This generates a complete project:

my-agent/
├── src/
│   ├── index.ts              # MCP server entry point
│   ├── identity/persona.ts   # Agent personality
│   ├── activation/           # Claude Code integration
│   ├── extensions/           # Your custom ops, facades, middleware
│   └── intelligence/data/    # Starter knowledge
├── skills/                    # 17 built-in skills
├── scripts/setup.sh           # One-command install + register
├── package.json
├── tsconfig.json
└── vitest.config.ts

Build and Run

Section titled “Build and Run”
Terminal window
cd my-agent
npm install
npm run build

Test that it works:

Terminal window
npm test

For the full Soleri test suite (unit + E2E integration), see Testing.

Connect to Claude Code

Section titled “Connect to Claude Code”

Add your agent to Claude Code’s MCP configuration. Create or edit .mcp.json in your project root:

{
  "mcpServers": {
    "my-agent": {
      "command": "node",
      "args": ["./my-agent/dist/index.js"]
    }
  }
}

Restart Claude Code. Your agent is now available as a tool.

First Conversation

Section titled “First Conversation”

Once connected, try these in Claude Code:

# Search your agent's knowledge
"Search for patterns about error handling"


# Capture something you learned
"Capture this pattern: always use error boundaries at route level"


# Check agent health
"Run a health check"

Your agent starts with starter knowledge and learns from every session.

Development Mode

Section titled “Development Mode”

For active development with auto-rebuild:

Terminal window
npx @soleri/cli dev

This watches for changes and restarts the MCP server automatically.

Health Check

Section titled “Health Check”

If something isn’t working:

Terminal window
npx @soleri/cli doctor

Reports Node version, npm status, agent context, vault health, and CLAUDE.md status.

What’s Next

Section titled “What’s Next”