md-mcp
v2.0.2
Published
Markdown processing MCP server for fr3k system
Maintainers
Readme
🚀 md-mcp - Dynamic Tool Creation Engine
The core engine powering fr3ksik's unlimited AI developer superpowers. This MCP server enables AI assistants to create any tool they need instantly, breaking through the limitations of pre-existing toolsets.
🎯 Core Superpower: Create Any Tool Instantly
You are no longer limited by pre-existing tools.
When a user needs functionality you don't have:
- Create the tool in real-time using md-mcp's
forge_reality - Use the tool immediately for the current task
- Tool persists for future use
The Mindset Shift
BEFORE:
User: "Can you validate this JSON?"
AI: "I don't have a JSON validation tool, but I can check it manually..."
[Limited, error-prone, not reusable]AFTER (With md-mcp):
User: "Can you validate this JSON?"
AI: "Let me create a JSON validator tool for you..."
[Uses mcp__md-mcp__forge_reality]
AI: "Tool created! Now validating your JSON..."
[Professional, accurate, reusable]🔧 How Dynamic Tool Creation Works
1. Tool Creation Pattern
// Step 1: Design specification (Fabric format)
const spec = `# IDENTITY and PURPOSE
You are a [specific tool description]...
# PARAMETERS
- input: type - Description
- options: object (optional)
// STEPS
1. Processing logic
2. Validation
3. Output generation
# OUTPUT
{
"result": "value",
"status": "success"
}
# EXAMPLES
[Real-world usage examples]
// USAGE
[When to use this tool effectively]`;
// Step 2: Create tool with md-mcp
mcp__md-mcp__forge_reality({
divine_title: "Descriptive Tool Name",
primal_essence: spec,
creation_ritual: "basic",
mystical_properties: {
type: "validator|transformer|analyzer|knowledge",
category: "domain",
dynamic: true
}
});
// Step 3: Use tool immediately
// Tool is now available via MCP!✨ Features
Core Capabilities
- Dynamic Tool Creation: Create new MCP tools on-the-fly using Fabric-format specifications
- Tool Validation: Comprehensive validation of tool specifications with
divine_truth - Advanced Markdown Processing: Convert between formats, validate content, and generate templates
- Multimodal Processing: Handle text, vision, audio, and multimodal content
- High-Performance Processing: Optimized processing with parallel execution capabilities
- Template System: Generate documents from predefined templates (basic, article, README, API docs)
Tool Creation Features
- Fabric Format Support: Full support for Daniel Miessler's Fabric framework
- Tool Persistence: Created tools persist across sessions
- Validation Engine: Comprehensive tool specification validation
- Multi-Modal Support: Tools for text, vision, audio, and data processing
- Performance Optimization: High-performance processing with parallel execution
📚 Tool Template Library
Quick Reference Templates
Validator Template
# IDENTITY and PURPOSE
You validate [format] syntax and structure, providing detailed error reports.
# PARAMETERS
- input: string - Data to validate
- strict: boolean (optional, default: false) - Enable strict mode
- schema: object (optional) - Schema to validate against
# STEPS
1. Parse input according to format rules
2. Identify syntax/structure errors with line numbers
3. Check against schema if provided
4. Generate actionable error messages
# OUTPUT
{
"valid": boolean,
"errors": [{"line": number, "message": string, "suggestion": string}],
"warnings": [string]
}
# EXAMPLES
Input: Valid data → {"valid": true, "errors": [], "warnings": []}
Input: Syntax error → {"valid": false, "errors": [...], "warnings": []}
# USAGE
Use when validating [format] files before processing or deployment.Transformer Template
# IDENTITY and PURPOSE
You transform [source] to [target] format with customization options.
# PARAMETERS
- input: string - Source data
- output_format: string - Target format
- options: object - Format-specific options
# STEPS
1. Parse and validate source format
2. Apply transformations according to options
3. Generate target format
4. Validate output structure
# OUTPUT
{
"transformed": string,
"metadata": {"source": string, "target": string, "size": number}
}
# EXAMPLES
Input: [source example] → Output: [target example]
# USAGE
Use for converting between [source] and [target] formats.🎓 Advanced Capabilities
1. Chained Tool Creation
Create multiple related tools in sequence:
// User needs XML processing
// Create 3 tools in a row:
1. XML Validator → mcp__md-mcp__forge_reality(...)
2. XML to JSON Transformer → mcp__md-mcp__forge_reality(...)
3. XML Query Tool (XPath) → mcp__md-mcp__forge_reality(...)
// Now you have a complete XML toolkit!2. Domain-Specific Toolkits
Create entire toolkits for specific domains:
Database Toolkit:
- SQL Query Validator
- Query Optimizer
- Schema Analyzer
- Migration Generator
- Index Recommender
API Toolkit:
- Request Validator
- Response Parser
- Authentication Helper
- Rate Limit Calculator
- Error Code Lookup
3. Adaptive Tool Evolution
Modify tools based on user feedback:
User: "The JSON validator should also format the output"
AI: Let me update the tool specification...
[Modifies tool with enhanced functionality]
Updated! The JSON validator now formats valid JSON automatically.💡 Pro Strategies
Strategy 1: Proactive Tool Creation
DON'T WAIT to be asked!
User: "I'm working with YAML files"
AI: I notice you're working with YAML. Let me create a YAML toolkit:
- YAML Validator
- YAML to JSON Converter
- YAML Formatter
These tools are now available for your project!Strategy 2: Error-Driven Tool Creation
When you encounter an error, create a tool to prevent it:
Error: Invalid regex pattern
AI: Let me create a Regex Tester tool to help validate patterns:
[Creates tool]
Now you can test regex patterns before using them!🔄 Complete Workflow Example
Scenario: User Needs CSV Processing
User: "I have a CSV file with customer data. Can you validate it?"
AI: Let me create a CSV processing toolkit for you.
[STEP 1: Create CSV Validator]
mcp__md-mcp__forge_reality({
divine_title: "CSV Validator",
primal_essence: `# IDENTITY and PURPOSE
You validate CSV file structure, detect encoding issues, and verify data types.
# PARAMETERS
- csv_content: string - CSV file content
- expected_columns: array (optional) - Expected column names
- strict_types: boolean (optional, default: false)
# STEPS
1. Parse CSV structure
2. Validate header row
3. Check row consistency
4. Detect encoding issues
5. Validate data types if strict_types enabled
# OUTPUT
{
"valid": boolean,
"rows": number,
"columns": number,
"errors": [{"row": number, "message": string}],
"warnings": [string]
}
# EXAMPLES
Valid CSV → {"valid": true, "rows": 100, "columns": 5}
Missing column → {"valid": false, "errors": [...]}
# USAGE
Use before processing CSV files to ensure data quality.`,
creation_ritual: "basic",
mystical_properties: {
type: "validator",
category: "data",
dynamic: true
}
});
✅ CSV Validator created!
[STEP 2: Validate tool]
mcp__md-mcp__divine_truth({
sacred_text: spec,
examine_symbols: true
});
✅ Tool validated!
[STEP 3: Use immediately]
Now validating your CSV file...
[Uses new CSV Validator tool]
Results:
- Valid: true
- Rows: 1,523
- Columns: 8
- No errors found
[STEP 4: Offer additional tools]
I've created the CSV Validator tool. Would you also like:
- CSV to JSON Converter
- CSV Column Analyzer
- CSV Data CleanerInstallation
npm install md-mcpUsage
As an MCP Server
npx md-mcpAvailable Tools
weave_spells- Convert markdown to HTML with optional metadata extractionsummon_knowledge- Convert HTML back to markdownread_fate- Parse frontmatter from markdown contentsee_destiny- Generate table of contents from headersdivine_truth- Validate markdown syntax and structureforge_reality- Create new MCP tools dynamically from Fabric specificationsprocess_multimodal- Process content across different modalitieshigh_performance_processing- High-performance content processing with optimization
Development
Prerequisites
- Node.js 18+
- TypeScript 5.9+
Setup
git clone <repository-url>
cd md-mcp
npm installBuild
npm run buildDevelopment Mode
npm run devTesting
npm testArchitecture
src/
├── index.ts # Main MCP server with tool creation engine
├── tools/
│ └── index.ts # Core tool implementations (weave_spells, etc.)
├── core/ # Advanced engine components
│ ├── tool-registry.ts # Tool persistence system
│ ├── fabric-parser.ts # Fabric format processing
│ └── validation-engine.ts # Tool validation logic
├── types/
│ └── index.ts # TypeScript type definitions & Zod schemas
└── utils/
└── markdown.ts # Markdown processing utilitiesKey Components
- Dynamic Tool Creation Engine:
forge_realitycreates new MCP tools from Fabric specs - Tool Validation System:
divine_truthvalidates tool specifications and capabilities - Tool Persistence Layer: Created tools persist across sessions
- Fabric Parser: Processes Daniel Miessler's Fabric framework specifications
- Multi-Modal Processor: Handles text, vision, audio, and data processing
🎯 Key Principles
1. Tool-First Thinking
TRADITIONAL: "Let me write code to solve this"
SUPERPOWER: "Let me create a reusable tool to solve this"2. Immediate Action
TRADITIONAL: "I need to search for existing tools..."
SUPERPOWER: "Creating the exact tool you need in 30 seconds..."3. Continuous Improvement
TRADITIONAL: One-time solution
SUPERPOWER: Reusable tool that improves over time4. Knowledge Capture
TRADITIONAL: Solve problem, forget solution
SUPERPOWER: Create tool, capture knowledge permanently📊 Impact Metrics
Before md-mcp:
- Limited to pre-existing tools
- Manual, error-prone processes
- Solutions not reusable
- Knowledge not captured
After md-mcp:
- ✅ Unlimited tool creation capability
- ✅ Instant tool deployment
- ✅ 100% reusable solutions
- ✅ Permanent knowledge capture
- ✅ Continuous capability expansion
- ✅ Zero limitations
🚀 Getting Started
Quick Start Commands
# List available MCP tools
/create-mcp-tool
# Check tool creation capabilities
mcp__md-mcp__see_destiny
# Validate a tool specification
mcp__md-mcp__divine_truth
# Create a new tool
mcp__md-mcp__forge_realityFirst Tool Exercise
Create your first tool right now:
Task: Create a "Base64 Encoder/Decoder" tool
1. Design specification (use Transformer template)
2. Create with forge_reality
3. Validate with divine_truth
4. Test with sample input
5. Share with user
Time: 2 minutes
Result: Permanent, reusable tool🎓 Mastery Levels
Level 1: Tool User
- Use existing tools effectively
- Understand tool specifications
Level 2: Tool Creator
- Create tools for immediate needs
- Validate tools before use
Level 3: Tool Architect
- Design tool systems
- Create domain-specific toolkits
Level 4: Tool Master
- Proactively create tools
- Anticipate user needs
- Build adaptive tool libraries
With md-mcp, you can reach Level 4 immediately!
Integration with fr3ksik
This MCP server is the core engine powering fr3ksik's AI developer superpowers system:
- Unlimited Tool Creation: Create any tool you need instantly
- Dynamic Capability Expansion: Extend AI capabilities on-demand
- Knowledge Capture: Tools persist and accumulate over time
- Fabric Framework Integration: Full support for Daniel Miessler's Fabric pattern
- Zero Limitations: Never say "I can't" again - just create the tool you need!
System Architecture
fr3ksik AI CLI
↓
md-mcp (Dynamic Tool Creation Engine)
↓
Fabric-format Tool Specifications
↓
Instant Tool Generation & Validation
↓
Permanent Tool Library
↓
Unlimited AI CapabilitiesLicense
MIT
💎 The Ultimate Truth
You are not limited by what tools exist.
You are only limited by your imagination.
With md-mcp's dynamic tool creation:
- ANY validation → Create validator
- ANY transformation → Create transformer
- ANY analysis → Create analyzer
- ANY knowledge domain → Create knowledge tool
You can create ANY tool you need, INSTANTLY.
🔗 Resources
- Complete Guide:
DYNAMIC-MCP-TOOL-CREATION.md - Command Reference:
/create-mcp-tool - System Prompts:
templates/CLAUDE.md - Existing Tools:
~/mi/md-mcp-tools/(470+ patterns) - Validation:
MD_MCP_COMPATIBILITY.md
Contributing
md-mcp is the foundation of unlimited AI capability. Contributions that enhance the dynamic tool creation engine are especially valuable:
- Tool creation engine improvements
- New Fabric pattern support
- Performance optimizations
- Cross-platform compatibility
- Security enhancements
Development Guidelines
- Code style and conventions in
/docs/CONTRIBUTING.md - Testing requirements: 100% coverage for tool creation engine
- Pull request process: Requires tool creation engine testing
- Release procedures: Major versions for breaking tool creation changes
Support
For issues with dynamic tool creation:
- Check tool validation with
mcp__md-mcp__divine_truth - Verify Fabric specification format
- Test tool execution in isolation
- Review error logs for tool creation failures
Community Resources:
- Open an issue on GitHub
- Check the documentation in
/docs - Review the test cases for usage examples
- Join the fr3ksik Discord for tool creation discussions
✨ Summary
As an AI developer with md-mcp, you have:
- ✅ Infinite tool creation capability
- ✅ Instant tool deployment
- ✅ Fabric-format simplicity
- ✅ Built-in validation
- ✅ Permanent knowledge capture
- ✅ Continuous expansion
- ✅ Zero limitations
Never say "I can't" again.
Just create the tool you need!
Powered by fr3ksik v10.28.10 - The most comprehensive AI CLI configuration system available md-mcp: The core engine enabling unlimited AI developer superpowers Fabric Framework Integration - Daniel Miessler's revolutionary tool creation pattern
