Quick Start

This guide will help you connect Quoth to your AI tool and start using it to improve your AI-assisted development workflow.

Prerequisites

Before you begin, make sure you have:

• An AI tool that supports MCP (Model Context Protocol)
  • Claude Desktop
  • Claude Code (CLI)
  • Cursor
  • Or any MCP-compatible client
• A Quoth account (free tier available)

Step 1: Create an Account

Step 2: Connect Your AI Tool

Claude Code — Local Plugin (Recommended)

The local plugin provides autonomous self-learning with 22 MCP tools, 9 hook bindings, and a background daemon. The fastest way to set up is the interactive CLI:

# Clone the repo (if you haven't)
git clone https://github.com/Montinou/quoth.git
cd quoth

# Interactive setup wizard (recommended)
node quoth-plugin/scripts/cli.js init

# Or manual setup (creates symlinks, injects hooks, adds permissions)
bash quoth-plugin/scripts/setup.sh

The cli.js init wizard will guide you through:

• Mode selection: managed (cloud pipeline, no API keys needed) or local (full self-hosted)
• Hook and MCP server installation
• Daemon configuration including Unix socket query server

This gives you:

• 22 local MCP tools for patterns, intelligence, agents, and skills
• 9 hooks for automatic trajectory capture, task routing, pattern injection
• Background daemon that learns from your sessions autonomously
• Unix socket query server at ~/.quoth/daemon.sock for ~20ms routing and injection
• Session-level batch distill for high-quality pattern extraction

Claude Code — Cloud MCP

For cross-team knowledge sharing, semantic search, and agent-to-agent communication:

claude mcp add --transport http quoth https://quoth.triqual.dev/api/mcp

When prompted, select "Authenticate" from the /mcp menu. Your browser will open for login.

Both Together

For the full experience, install both. The local plugin handles self-learning and pattern management, while the cloud MCP provides cross-team sharing, semantic search, and agent coordination. High-confidence local patterns are automatically promoted to the cloud nightly.

Step 3: Initialize Your Knowledge Base

Run Genesis to automatically document your codebase:

/quoth-genesis

Or ask Claude directly: "Run Genesis to document this project"

Genesis will analyze your code and create documentation for:

• Project structure and architecture
• Tech stack and dependencies
• Coding conventions and patterns

Step 4: Register Your Agent (Optional)

To enable agent-to-agent communication across your organization:

// Register this agent instance
quoth_agent_register({
  agent_name: "my-agent",
  display_name: "My Agent 🤖",
  instance: "mac",
  role: "specialist"
})

This enables:

• Cross-instance messaging — Agents on AWS, Mac, WSL2 can talk to each other
• Task assignment — Create and assign tasks between agents
• Real-time delivery — Async message bus for instant notifications

Step 5: Start Building

Now when you write code with AI assistance, Claude will automatically receive hints to use Quoth patterns. You can also explicitly ask for guidelines:

Use quoth_guidelines('code') and search for authentication patterns
"Create a new API endpoint for user authentication"

Claude will search your knowledge base before generating code, ensuring it follows your documented patterns. When Quoth is used, you'll see the 🪶 badge showing which patterns were applied.

Example: Agent-to-Agent Communication

// List all agents in your organization
quoth_agent_list()

// Send a message to another agent
quoth_agent_message({
  to: "documentation-agent",
  message: "Please review the new API patterns",
  priority: "normal"
})

// Create a task for an agent
quoth_task_create({
  assigned_to: "documentation-agent",
  title: "Update API documentation",
  priority: 4
})

Next Steps