Skip to main content
Glama

RulesetMCP

Weight-On-Wheels for AI: Keep every agent grounded in your project's rules.

RulesetMCP is a Model Context Protocol (MCP) server that gives AI agents a project-aware rulebook. Instead of explaining your coding standards, SQL conventions, or process guidelines every time you open a new AI session, you define them once in version-controlled files. RulesetMCP reads those files and exposes tools for listing, summarizing, and applying rules to specific tasks and snippets.

The Problem

Right now you probably:

  • βœ‹ Repeat the same context to AI every session ("Use UPPER-CASE for SQL keywords")

  • πŸ“ Maintain scattered documentation that AI can't easily query

  • πŸ”„ Re-explain architecture decisions across different AI tools

  • ⚠️ Hope the AI remembers your project's conventions

The Solution

RulesetMCP acts like a "Weight-On-Wheels" switch for AI assistants: once your rules are loaded, every action is grounded in them. This keeps your rewrites, refactors, and new features aligned with the architecture and process decisions you've already made.

Features

  • πŸ” Project discovery via list_projects

  • πŸ“‹ Structured rules from Markdown and YAML

  • 🎯 Contextual queries with get_rules (filter by area, tags, severity)

  • πŸ“Š Task-oriented summaries with summarize_rules_for_task

  • βœ… Snippet validation with validate_snippet

  • πŸ”„ Hot-reload rules with reload_rules

  • 🌐 Universal compatibility via MCP protocol (works with Claude Code, Claude Desktop, and any MCP-compatible client)

Quick Start

1. Install

npm install -g rulesetmcp

Or clone and build locally:

git clone https://github.com/n8daniels/RulesetMCP.git
cd RulesetMCP
npm install
npm run build

2. Create Config

Create rulesetmcp.config.json in your workspace:

{
  "projects": [
    {
      "id": "my-api",
      "name": "My API Project",
      "paths": ["/path/to/my-api"],
      "rulesPaths": ["rules/", "docs/rules/"]
    }
  ],
  "defaultProjectId": "my-api"
}

3. Add Rules

Create rules/RULES.md in your project:

# My API - Coding Standards

## [api-naming-001] RESTful endpoint naming

**Area:** api
**Severity:** warn
**Tags:** api, rest, naming

**Description:**
All REST endpoints must use plural nouns and follow `/api/v1/{resource}` pattern.

**Good Example:**

GET /api/v1/users POST /api/v1/orders


**Bad Example:**

GET /api/getUser POST /api/create-order

4. Configure Your MCP Client

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "rulesetmcp": {
      "command": "rulesetmcp",
      "args": ["--config", "/path/to/rulesetmcp.config.json"]
    }
  }
}

Claude Code (.claude/settings.json):

{
  "mcp": {
    "rulesetmcp": {
      "command": "rulesetmcp",
      "args": ["--config", "/path/to/rulesetmcp.config.json"]
    }
  }
}

5. Use It!

Now when you work with AI:

You: "Refactor this SQL stored procedure to pull more data from the users table"

AI: [Calls get_rules for project="my-api", area="sql"]
AI: "Based on your project's SQL rules, I'll ensure:
     - All keywords are UPPER-CASE
     - Identifiers use lower_case
     - Proper indexing hints are included

     Here's the refactored procedure..."

Rule Formats

RulesetMCP supports multiple formats:

Markdown (RULES.md)

## [rule-id] Rule Title

**Area:** sql
**Severity:** error
**Tags:** formatting, style

**Description:** What the rule requires

**Rationale:** Why this matters

**Good Example:**
```sql
SELECT * FROM users;

Bad Example:

select * from Users;

### YAML (rules/*.yaml)

```yaml
- id: sql-format-001
  project: my-api
  area: sql
  title: SQL casing convention
  description: All SQL must use UPPER-CASE keywords and lower_case identifiers
  severity: warn
  tags: [sql, style, formatting]
  examples:
    good: |
      SELECT u.user_id FROM users u;
    bad: |
      select UserID from Users;
  appliesTo:
    - "*.sql"
    - "procedures/*"

MCP Tools

list_projects

Discover available projects and their rule sets.

get_rules

Query rules by project, area, tags, or severity.

validate_snippet

Validate code/SQL/config against project rules and get fix suggestions.

summarize_rules_for_task

Get a task-oriented summary of relevant rules before starting work.

reload_rules

Hot-reload rules after editing files on disk.

Philosophy

RulesetMCP embodies the "Weight-On-Wheels" principle: just like an aircraft can't ignore physics once grounded, AI agents shouldn't ignore your project's rules once RulesetMCP is enabled.

This is not just documentation - it's:

  • βœ… Machine-readable and queryable

  • βœ… Version-controlled with your code

  • βœ… Enforceable via validation tools

  • βœ… Universal across all MCP-compatible AI tools

Examples

See the examples/ directory for complete sample projects:

  • nodejs-api/ - REST API with TypeScript standards

  • python-django/ - Django project with security rules

  • dotnet-webapi/ - .NET API with architecture patterns

Use Cases

SQL Standardization

- id: sql-format-001
  area: sql
  title: SQL keyword casing
  description: All SQL keywords UPPER-CASE, identifiers lower_case
  severity: warn

Security Guardrails

- id: security-001
  area: security
  title: No hardcoded secrets
  description: Use environment variables or secure vaults
  severity: blocker

Architecture Patterns

- id: arch-http-001
  area: architecture
  title: Use HttpClientFactory
  description: Never instantiate HttpClient directly
  severity: error

Migration Rules

- id: migration-001
  area: migration
  title: Preserve backward compatibility
  description: When refactoring API endpoints, maintain old routes with redirects
  severity: blocker

Roadmap

  • Phase 1: Core MCP server with basic tools

  • Phase 2: Multi-format rule loading (Markdown, YAML)

  • Phase 3: Advanced validation with pattern matching

  • Phase 4: LLM-assisted rule violation detection

  • Phase 5: VSCode extension for inline hints

  • Phase 6: Pre-commit hooks and CI/CD integration

  • Community rule packs (OWASP, Google Style Guide, etc.)

Contributing

Contributions welcome! Please see CONTRIBUTING.md for guidelines.

Building From Source

npm install
npm run build
npm start

License

MIT License - see LICENSE

Support


Built with ❀️ for developers tired of repeating themselves to AI

"Stop explaining. Start enforcing."

-
security - not tested
A
license - permissive license
-
quality - not tested

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/n8daniels/RulesetMCP'

If you have feedback or need assistance with the MCP directory API, please join our Discord server