lloyd-ai-rules

Shared rules, skills, MCP configurations, and command permissions for AI-powered development tools. Supports Cursor, Claude Code, GitHub Copilot, OpenAI Codex/Agents, and IntelliJ / JetBrains AI Assistant from a single source of truth.

Supported IDEs and Agents

| Tool | Description | | ------------------------------------- | ------------------------------------------------------------------------------ | | Cursor | Rules in .cursor/rules/*.mdc, skills in .cursor/skills/ | | Claude Code | Rules in CLAUDE.md, skills in .claude/skills/ | | GitHub Copilot | Instructions in .github/copilot-instructions.md, skills in .github/skills/ | | OpenAI Codex/Agents | AGENTS.md and .agents/skills/ (with agents/openai.yaml) | | IntelliJ / JetBrains AI Assistant | Rules in .aiassistant/rules/*.md |

Quick Start

1. Install the package

npm install -D lloyd-ai-rules

The postinstall hook automatically adds generated file paths to your repository's .gitignore.

2. Create a configuration file (optional)

Create lloyd-ai-rules.config.js in your project root:

export default {
  preset: "default",
  rules: {
    include: [], // Override to include only these rules
    exclude: [] // Exclude specific rules from the preset
  },
  skills: {
    include: [], // Override to include only these skills
    exclude: [] // Exclude specific skills from the preset
  }
};

If no config file exists, the CLI uses the "default" preset from presets.json with no exclusions.

3. Generate rules and skills

npx lloyd-ai-skills-and-rules-kit generate

CLI Reference

lloyd-ai-skills-and-rules-kit <command> [flags]

Commands

| Command | Description | | -------- | ----------------------------- | | generate | Generate all output files | | doctor | Print environment diagnostics |

Flags

Flags can be combined. When no category flag is given, all categories are generated. When no platform flag is given, all platforms are generated.

Category flags

| Flag | Description | | --------------- | --------------------------------- | | --rules | Generate rules only | | --skills | Generate skills only | | --mcp | Generate MCP configs only | | --permissions | Generate command permissions only |

Platform flags

| Flag | Description | | ------------- | ----------------------------- | | --codex | OpenAI Codex/Agents output | | --cursor | Cursor output | | --claude | Claude Code output | | --copilot | GitHub Copilot output | | --jetbrains | JetBrains AI Assistant output |

Examples

lloyd-ai-skills-and-rules-kit generate                  # Everything, all platforms
lloyd-ai-skills-and-rules-kit generate --rules          # Rules for all platforms
lloyd-ai-skills-and-rules-kit generate --skills --codex # Skills for Codex only
lloyd-ai-skills-and-rules-kit generate --cursor         # Rules + skills + MCP for Cursor
lloyd-ai-skills-and-rules-kit doctor                    # Environment diagnostics

Presets

Presets are defined in presets.json. Each preset lists the rules, skills, MCP sources, and permission sources to include by default.

| Preset | Description | Rules | Skills | | -------------- | ------------------------------------------------ | ---------------------------------- | ----------------------------------------------------------------------- | | default | Baseline for all projects | code-quality, dependencies, naming | add-test-locators, generate-page-object-model-files, generate-e2e-tests | | uiautomation | UI automation focused (all rules and all skills) | code-quality, dependencies, naming | add-test-locators, generate-page-object-model-files, generate-e2e-tests |

Rules

| Rule | Description | Always Applied | | -------------- | ------------------------------------------- | -------------- | | code-quality | Core code quality standards and principles | Yes | | dependencies | Rules for managing third-party dependencies | No | | naming | Naming conventions for all code | Yes |

Rules are .mdc files in src/rules/. The CLI consolidates them into each platform's generated rule output.

Skills

| Skill | Description | | ---------------------------------- | ----------------------------------------------------------------------------------------- | | add-test-locators | Add test locators (testID / data-testid) to components for web or mobile applications | | generate-page-object-model-files | Generate Cypress page object model files from screen source files | | generate-e2e-tests | Generate and maintain e2e tests based on recent source code changes |

Skills live in src/skills/<skill-name>/. Each skill directory contains:

• SKILL.md — detailed skill instructions with YAML frontmatter (name, description)
• openai.yaml (or agents/openai.yaml) — display metadata for OpenAI Codex UI

Configuration

Overriding a preset

The config file selects a preset and optionally applies include/exclude overrides:

export default {
  preset: "uiautomation",
  rules: {
    include: [], // Empty = use preset's full list
    exclude: ["naming"] // Remove specific rules
  },
  skills: {
    include: ["add-test-locators"], // Non-empty = replaces preset list entirely
    exclude: []
  }
};

Resolution logic:

• If include has entries → those replace the preset's list.
• If include is empty → the preset's list is used.
• exclude then filters out from whichever list was chosen.

MCP and Permissions

MCP server configs and command permissions follow the same include/exclude pattern:

export default {
  preset: "default",
  mcp: {
    include: [],
    exclude: []
  },
  permissions: {
    include: [],
    exclude: []
  }
};

MCP sources live in src/mcp/ and permission sources in src/permissions/.

What Gets Generated

After running lloyd-ai-skills-and-rules-kit generate, the following files are created or updated in the consuming repository:

| Path | Tool | Description | | -------------------------------------- | -------------------- | ----------------------------- | | AGENTS.md | OpenAI Codex/Agents | Agent instructions and rules | | .agents/skills/<skill>/SKILL.md | OpenAI Codex/Agents | Skill definitions | | CLAUDE.md | Claude Code | Claude Code rules | | .claude/skills/<skill>/SKILL.md | Claude Code | Skill definitions | | .github/copilot-instructions.md | GitHub Copilot | Copilot instructions | | .github/skills/<skill>/SKILL.md | GitHub Copilot | Skill definitions | | .cursor/rules/generated-core.mdc | Cursor | Cursor rule file | | .cursor/skills/<skill>/SKILL.md | Cursor | Skill definitions | | .aiassistant/rules/generated-core.md | IntelliJ / JetBrains | AI Assistant project rules | | mcp-servers/codex-mcp.json | OpenAI Codex | MCP server configuration | | mcp-servers/claude-mcp.json | Claude Code | MCP server configuration | | mcp-servers/cursor-mcp.json | Cursor | MCP server configuration | | mcp-servers/copilot-mcp.json | GitHub Copilot | MCP server configuration | | mcp-servers/jetbrains-mcp.json | JetBrains | MCP server configuration | | command-permissions.json | Claude Code | Command permission allowlists |

All generated paths are automatically added to .gitignore by the postinstall hook so they are not tracked in version control.

Development

Prerequisites

• Node.js >= 20
• npm

Setup

npm install

Building

npm run build

Type checking

npm run typecheck

Testing

npm test

Local development

npm run dev -- generate
npm run dev -- doctor

Self-consumption

This repo uses its own lloyd-ai-rules.config.js to configure which rules and skills to include. Generated output (AGENTS.md, CLAUDE.md, .cursor/, .github/, .agents/, .claude/, .aiassistant/, mcp-servers/, command-permissions.json) is gitignored and recreated by running lloyd-ai-skills-and-rules-kit generate. Source of truth lives in src/rules/ and src/skills/.

License

See LICENSE.