Claude Code

Connect Claude Code to Trurlic so every implementation task is constrained by your architectural decisions.

Prerequisites

• Trurlic installed and .trurlic/ initialized in your project
• Claude Code installed (npm install -g @anthropic-ai/claude-code)

Setup

The fastest way:

trurlic install --ide claude-code

This runs claude mcp add trurlic -s user -- trurlic serve to register the Trurlic MCP server with Claude Code at the user scope.

Project-scoped configuration

For team sharing via git, use claude mcp add with project scope directly. This writes to .mcp.json at the project root:

claude mcp add trurlic --scope project -- trurlic serve

Every developer who clones the repo gets the same Trurlic MCP configuration automatically.

Manual setup

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "trurlic": {
      "command": "trurlic",
      "args": ["serve"]
    }
  }
}

Adjust the path if trurlic is not on PATH. Use which trurlic to find it.

Agent instructions

Add to your project’s CLAUDE.md:

## Trurlic

This project uses Trurlic for architectural decisions.

Before implementing any task:

1. Call `advance` with the component name.
2. Follow the returned action exactly.
3. Call `advance` again after completing each action.
4. Repeat until `ready: true`.
5. Implement, constrained by every decision in the context brief.

See CLAUDE.md for the full recommended instructions, including the two-mode workflow (design mode vs. implementation mode).

End-to-end verification

1. Initialize Trurlic: trurlic init && trurlic add component auth -d "Authentication"
2. Install config: trurlic install --ide claude-code
3. Start a new Claude Code session.
4. Ask Claude Code to “design the auth component” — it should call advance, then get_step_prompt, and guide you through the Socratic flow.
5. After design, ask it to “implement auth” — it should call get_context and follow the recorded decisions.
6. Verify: trurlic status should show the recorded decisions.