Skip to main content

FPF Module MCP Tool Specification

Version: v1.1.9 | Status: Active | Last Updated: February 2026

Overview​

This document specifies MCP (Model Context Protocol) tools for AI integration with the FPF module. These tools allow AI agents to interact with FPF specifications programmatically.

Tools​

fpf_fetch_spec​

Fetch the latest FPF specification from GitHub.

Parameters:

  • repo (string, optional): GitHub repository in format "owner/repo" (default: "ailev/FPF")
  • branch (string, optional): Branch name (default: "main")
  • output_path (string, optional): Path to save the fetched file

Returns:

  • content (string): The fetched specification content
  • path (string): Path where content was saved (if output_path provided)

fpf_parse_spec​

Parse a local FPF specification file.

Parameters:

  • file_path (string, required): Path to FPF-Spec.md file
  • extract_concepts (boolean, optional): Whether to extract concepts (default: true)
  • extract_relationships (boolean, optional): Whether to extract relationships (default: true)

Returns:

  • spec (object): Parsed FPFSpec object with patterns, concepts, relationships
  • pattern_count (integer): Number of patterns extracted
  • concept_count (integer): Number of concepts extracted
  • relationship_count (integer): Number of relationships extracted

fpf_search_patterns​

Search for patterns in a parsed FPF specification.

Parameters:

  • query (string, required): Search query
  • file_path (string, optional): Path to FPF-Spec.md (if not already parsed)
  • filters (object, optional): Filter object with:
    • status (string, optional): Filter by status (Stable, Draft, Stub, New)
    • part (string, optional): Filter by part (A, B, C, etc.)

Returns:

  • results (array): List of matching Pattern objects
  • count (integer): Number of results

fpf_get_pattern​

Get a specific pattern by ID.

Parameters:

  • pattern_id (string, required): Pattern identifier (e.g., "A.1")
  • file_path (string, optional): Path to FPF-Spec.md (if not already parsed)
  • include_related (boolean, optional): Include related patterns (default: false)
  • depth (integer, optional): Relationship depth (default: 1)

Returns:

  • pattern (object): Pattern object
  • related (array, optional): Related patterns if include_related is true

fpf_export_json​

Export FPF specification to JSON.

Parameters:

  • file_path (string, required): Path to FPF-Spec.md
  • output_path (string, required): Path to output JSON file
  • format (string, optional): Export format (default: "json", only "json" supported)

Returns:

  • output_path (string): Path to exported file
  • size (integer): File size in bytes

fpf_build_context​

Build context string for prompt engineering.

Parameters:

  • file_path (string, required): Path to FPF-Spec.md
  • pattern_id (string, optional): Pattern ID to build context for
  • concept (string, optional): Concept name to build context for
  • filters (object, optional): Filters for context building
  • depth (integer, optional): Relationship depth (default: 1)
  • output_path (string, optional): Path to save context

Returns:

  • context (string): Context string
  • output_path (string, optional): Path where context was saved

fpf_visualize​

Generate visualization of FPF patterns.

Parameters:

  • file_path (string, required): Path to FPF-Spec.md
  • type (string, required): Visualization type ("hierarchy" or "dependencies")
  • output_path (string, optional): Path to save visualization

Returns:

  • diagram (string): Mermaid diagram string
  • output_path (string, optional): Path where diagram was saved

fpf_get_concepts​

Get all concepts from a parsed specification.

Parameters:

  • file_path (string, optional): Path to FPF-Spec.md (if not already parsed)
  • pattern_id (string, optional): Filter by pattern ID
  • concept_type (string, optional): Filter by concept type

Returns:

  • concepts (array): List of Concept objects
  • count (integer): Number of concepts

fpf_get_relationships​

Get relationships for a pattern or all relationships.

Parameters:

  • file_path (string, optional): Path to FPF-Spec.md (if not already parsed)
  • pattern_id (string, optional): Get relationships for specific pattern
  • relationship_type (string, optional): Filter by relationship type

Returns:

  • relationships (array): List of Relationship objects
  • count (integer): Number of relationships

Tool Registration​

These tools should be registered with the MCP server using the standard MCP tool registration format. Each tool should include:

  • Name
  • Description
  • Input schema (JSON Schema)
  • Output schema (JSON Schema)

Usage Example​

# Register tools
mcp_server.register_tool("fpf_search_patterns", {
    "name": "fpf_search_patterns",
    "description": "Search for patterns in FPF specification",
    "inputSchema": {
        "type": "object",
        "properties": {
            "query": {"type": "string"},
            "file_path": {"type": "string"},
            "filters": {"type": "object"}
        },
        "required": ["query"]
    }
})

# Use tool
result = mcp_server.call_tool("fpf_search_patterns", {
    "query": "holon",
    "filters": {"status": "Stable"}
})