RulesForge
STANDARDS COMPLIANT · 100% LOCAL
ALL-IN-ONE MULTI-AI PROMPT STANDARD · V2.0

One project. Perfect AI rulebooks for every platform.

Configure your project goals, commands, and constraints once. RuleForge formats the instructions dynamically according to the official specifications for Claude Code, OpenAI/AAIF AGENTS.md, Cursor, GitHub Copilot, Windsurf, and ChatGPT.

01 / PROJECT & AGENT CONFIGURATION

Define your assistant

0 characters
Target AI platforms & specifications 1 platform selected
One per line

⚡ Scoped Rules (.mdc / Path-Specific) (0 defined)

Generates modular scoped rules: .cursor/rules/<name>.mdc with glob activation, .agents/rules/*.md, .github/instructions/*.instructions.md, etc.

🧩 Agent Skills (SKILL.md) (0 defined)

Agent Skills generate compliant .agents/skills/<name>/SKILL.md packages with structured frontmatter and procedures.

🤖 Custom Subagents (0 defined)

Generates specialized subagents with tool permissions and models for Claude Code (.claude/agents/*.md), Gemini CLI (.gemini/agents/*.md), Copilot (.github/agents/*.agent.md), OpenCode, Antigravity, and Kiro.

⌨️ Commands & Prompts (0 defined)

Generates reusable command prompt files (e.g. .github/prompts/<name>.prompt.md for Copilot, .claude/commands/<name>.md for Claude Code, .gemini/commands/*.md, .opencode/commands/*.md).

🔌 Model Context Protocol (MCP) (0 defined)

Connects coding agents to external tools. Standard format for Claude Code (.mcp.json), Cursor (.cursor/mcp.json), Cline (cline_mcp_settings.json), VS Code (.vscode/mcp.json), Antigravity, OpenCode, and Kiro.

🪝 Lifecycle Hooks (0 defined)

Automate tool verification actions (e.g. running linters pre-commit, tests post-tool execution) with timeout and blocking policies.

🛡️ Permissions, Sandbox & Settings

Configured

Configures default model preferences and security boundaries (Claude .claude/settings.json, Cursor .cursor/settings.json, Gemini .gemini/settings.json, Antigravity, OpenCode).

Standard Note: Instructions guide LLM behavior effectively, but do not replace technical access controls, CI test pipelines, and human code review for mission-critical operations.
02 / LIVE PREVIEW CLAUDE.md Anthropic Specification 


        

          

            Ready · Auto-generated
            
          

          

            

              
              
              
            

            0 words · 0 lines
          

        

      


    


    
    

    
    

      

        
        

          TECHNICAL REFERENCE & BEST PRACTICES
          
The Complete Guide to Multi-AI Context Engineering

          

            Modern coding assistants (Claude Code, OpenAI Codex, Cursor Composer, GitHub Copilot, Windsurf, and Antigravity) are only as reliable as the boundary rules and architectural context committed into your repository. RulesForge standardizes this workflow, creating persistent, version-controlled rulebooks that eliminate prompt repetition and hallucinated commands.
          

        


        
        

          

            
⚡ Command-First Verification

            

              Coding agents are strictly instructed to execute your actual test, build, and lint commands before marking any feature as complete, preventing phantom test passes.
            

          

          

            
🛡️ Defensive Guardrails

            

              Imperative constraints prevent accidental secrets disclosure (e.g. committing .env), unauthorized dependency bloat, and destructive file overrides.
            

          

          

            
📐 Multi-Specification Sync

            

              Generate router-style CLAUDE.md for Anthropic, frontmatter-scoped .mdc for Cursor, and AAIF-compliant AGENTS.md in a single click.
            

          

        


        
        

          
AI Rulebook Specifications by Platform

          
Overview of standardized instruction formats, repository locations, and parsing conventions.

          
          

            
              

                Platform / Tool
                Target File Path
                Format & Structure
                Recommended Context Strategy
              

            
            
              

                Anthropic Claude Code
                CLAUDE.md
                Markdown Router
                Concise router index under 200 lines pointing to detailed docs
              

              

                OpenAI Codex / AAIF
                AGENTS.md
                Standard Open Spec
                Declarative capabilities, commands, and security boundaries
              

              

                Cursor Composer
                .cursor/rules/*.mdc
                YAML Frontmatter
                Scoped rule matching based on file glob patterns
              

              

                GitHub Copilot
                .github/copilot-instructions.md
                Markdown Guidelines
                Project-wide coding conventions, test patterns, and styles
              

              

                Windsurf Cascade
                .windsurfrules
                Workflow Rules
                Step-by-step confirmation loops and execution gates
              

              

                ChatGPT & General LLMs
                chatgpt-instructions.md
                Custom Persona
                High-level persona, response tone, and architectural perspective
              

            
          

        


        
        

          
How to Integrate Into Your Repository

          

            

              STEP 1
              
Define Architecture

              
Input your project tech stack, build & test commands, and critical boundaries into the generator form above.

            

            

              STEP 2
              
Select Targets

              
Check the platforms used by your development team (Claude Code, Cursor, Copilot, etc.) to generate synchronized specs.

            

            

              STEP 3
              
Download Bundle

              
Use "Copy", "Download File", or the "Download ZIP" button to get the pre-structured repository directory tree.

            

            

              STEP 4
              
Commit to Git

              
Commit the rulebooks to your root branch. Any developer or CI agent opening the codebase inherits the instructions instantly.

            

          

        


        
        
Frequently Asked Questions (FAQ)

        
        

          Where should I save the exported files in my codebase?
          

            Each AI platform monitors a specific repository path:
            
    
              
  • Claude Code: Save CLAUDE.md directly at the root of your project.
    • 
              
  • OpenAI Codex / AAIF: Save AGENTS.md in the project root directory.
    • 
              
  • Cursor: Place files under .cursor/rules/rules.mdc (or multiple modular .mdc files).
    • 
              
  • GitHub Copilot: Place at .github/copilot-instructions.md.
    • 
              
  • Windsurf: Save as .windsurfrules in your repository root.
    • 
            

            Clicking "📦 Download ZIP" automatically builds this exact folder structure for you.
          

        


        

          Is my project idea, prompt, or code sent to an external server?
          

            No, absolutely not. RulesForge is built on a 100% local-first, client-side architecture. All text generation, YAML frontmatter formatting, and ZIP compression happen in your local browser runtime via vanilla JavaScript. Nothing is sent to, stored on, or analyzed by remote servers. Review our full Privacy Policy for technical details.
          

        


        

          Can I use generated rulebooks for commercial software and enterprise projects?
          

            Yes. You retain 100% intellectual property ownership of all instruction files produced. You are completely free to use, modify, distribute, and license them across private enterprise codebases or public open-source repositories without royalties, attribution, or restriction.
          

        


        

          How does RulesForge prevent AI hallucinations and unauthorized edits?
          

            RulesForge generates imperative boundary directives. Instead of vague suggestions, the output injects strict non-negotiable clauses: mandatory test execution gates, prohibition of modifying secret files (such as .env or credentials), explicit architectural constraints, and requirements to explain defects before applying patches.
          

        


        

          Why does Anthropic recommend keeping CLAUDE.md under 200 lines?
          

            Anthropic’s official guidelines advise keeping CLAUDE.md concise (serving as a "router" file) because excessively long system prompts degrade context attention and increase latency on every turn. RulesForge formats CLAUDE.md compactly with references to secondary documentation rather than dumping raw codebases into system context.
          

        


        

          Is RulesForge completely free to use?
          

            Yes, RulesForge is 100% free with no paywalls, subscriptions, or login requirements. We maintain the service and server infrastructure through non-intrusive contextual advertising provided by Google AdSense.