MCP Server Overview

The Rulebound MCP (Model Context Protocol) server exposes rule enforcement as tools that AI coding agents can call directly. This enables pre-write enforcement — validating code against project rules before it is written to disk.

What is MCP?

The Model Context Protocol is an open standard for connecting AI agents to external tools and data sources. Rulebound's MCP server provides five tools that AI agents call during their workflow.

Architecture

AI Coding Agent (Claude, Copilot, etc.)
    |
    |  MCP Protocol (stdio)
    |
Rulebound MCP Server
    |
    +-- Rule Loader
    |       Reads .rulebound/rules/*.md
    |
    +-- Validator
    |       Semantic pattern matching
    |
    +-- AST Analyzer
            tree-sitter structural analysis

Available Tools

Tool	Purpose	When to Call
`find_rules`	Find relevant rules for a task	Before starting implementation
`validate_plan`	Validate an implementation plan	Before writing any code
`check_code`	Check a code snippet	After writing code
`list_rules`	List all project rules	When exploring project standards
`validate_before_write`	Pre-write code validation	Before writing any code file

Quick Start

# Install
pnpm add @rulebound/mcp

# Run the MCP server
npx rulebound-mcp

Agent Configuration

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "rulebound": {
      "command": "npx",
      "args": ["rulebound-mcp"],
      "cwd": "/path/to/your/project"
    }
  }
}

Claude Code

Add to .claude/settings.json:

{
  "mcpServers": {
    "rulebound": {
      "command": "npx",
      "args": ["rulebound-mcp"]
    }
  }
}

Auto-Detection

The MCP server automatically:

Detects project stack from project files (package.json -> TypeScript, pom.xml -> Java, etc.)
Finds rules directory by scanning .rulebound/rules/, rules/, and examples/rules/
Filters rules by stack so agents only see relevant rules

Next Steps

Tools Reference — Detailed tool documentation
Configuration — Agent integration patterns