2871 MCP Reference 9b2d06aa

MCP Reference - Skill Seekers

Version: 3.1.0
Last Updated: 2026-02-16
Complete reference for 26 MCP tools

Table of Contents

• Overview
  • What is MCP?
  • Transport Modes
  • Starting the Server
• Tool Categories
  • Core Tools (9)
  • Extended Tools (9)
  • Config Source Tools (5)
  • Config Splitting Tools (2)
  • Vector Database Tools (4)
  • Workflow Tools (5)
• Tool Reference
• Common Patterns
• Error Handling

Overview

What is MCP?

MCP (Model Context Protocol) allows AI agents like Claude Code to interact with Skill Seekers through a standardized interface. Instead of running CLI commands, you can use natural language:

"Scrape the React documentation and create a skill"
"Package the output/react skill for Claude"
"List available workflow presets"

Transport Modes

The MCP server supports two transport modes:

Starting the Server

# stdio mode (default)
skill-seekers-mcp

# HTTP mode
skill-seekers-mcp --transport http --port 8765

# With custom host
skill-seekers-mcp --transport http --host 0.0.0.0 --port 8765

Tool Categories

Core Tools (9)

Essential tools for basic skill creation workflow:

Extended Tools (9)

Advanced scraping and analysis tools:

Config Source Tools (5)

Manage configuration sources:

Config Splitting Tools (2)

Handle large documentation:

Vector Database Tools (4)

Export to vector databases:

Workflow Tools (5)

Manage enhancement workflows:

Tool Reference

Core Tools

list_configs

List all available preset configurations.

Parameters: None

Returns: Array of config objects

{
  "configs": [
    {
      "name": "react",
      "description": "React documentation",
      "source": "bundled"
    }
  ]
}

Example:

# Natural language
"List available configurations"
"What configs are available?"
"Show me the preset configs"

generate_config

Generate a configuration file from a documentation URL.

Parameters:

Returns: Config JSON object

Example:

# Natural language
"Generate a config for https://docs.django.com/"
"Create a Django config"
"Make a config from the React docs URL"

validate_config

Validate a configuration file structure.

Parameters:

Returns: Validation result

{
  "valid": true,
  "errors": [],
  "warnings": []
}

Example:

# Natural language
"Validate this config: {config_json}"
"Check if my config is valid"
"Validate configs/react.json"

estimate_pages

Estimate total pages for documentation scraping.

Parameters:

Returns: Estimation results

{
  "estimated_pages": 230,
  "discovery_rate": 1.28,
  "estimated_time_minutes": 3.8
}

Example:

# Natural language
"Estimate pages for the React config"
"How many pages will Django docs have?"
"Estimate with max 500 pages"

scrape_docs

Scrape documentation website and generate skill.

Parameters:

Returns: Scraping results

{
  "skill_directory": "output/react/",
  "pages_scraped": 180,
  "references_generated": 12,
  "status": "success"
}

Example:

# Natural language
"Scrape the React documentation"
"Scrape Django with enhancement level 3"
"Do a dry run of the Vue docs scrape"

package_skill

Package skill directory into uploadable format.

Parameters:

Returns: Package info

{
  "package_path": "output/react-claude.zip",
  "platform": "claude",
  "size_bytes": 245760
}

Example:

# Natural language
"Package the React skill for Claude"
"Create a Gemini package for output/django/"
"Package with streaming mode"

upload_skill

Upload skill package to LLM platform.

Parameters:

Returns: Upload result

{
  "success": true,
  "platform": "claude",
  "skill_id": "skill_abc123"
}

Example:

# Natural language
"Upload the React package to Claude"
"Upload output/django-gemini.tar.gz to Gemini"

enhance_skill

AI-powered enhancement of SKILL.md.

Parameters:

Returns: Enhancement result

{
  "success": true,
  "mode": "LOCAL",
  "skill_md_lines": 450
}

Example:

# Natural language
"Enhance the React skill"
"Enhance with security-focus workflow"
"Run enhancement in API mode"

install_skill

Complete workflow: scrape → enhance → package → upload.

Parameters:

Returns: Installation result

{
  "success": true,
  "skill_directory": "output/react/",
  "package_path": "output/react-claude.zip",
  "uploaded": true
}

Example:

# Natural language
"Install the React skill"
"Install Django for Gemini with no upload"
"Complete install of the Vue config"

Extended Tools

scrape_github

Scrape GitHub repository.

Parameters:

Example:

# Natural language
"Scrape the facebook/react repository"
"Analyze the Django GitHub repo"
"Scrape vercel/next.js with issues"

scrape_pdf

Extract content from PDF file.

Parameters:

Example:

# Natural language
"Scrape the manual.pdf file"
"Extract content from API-docs.pdf"
"Process scanned.pdf with OCR"

scrape_codebase

Analyze local codebase.

Parameters:

Example:

# Natural language
"Analyze the ./my-project directory"
"Scrape this codebase with comprehensive preset"
"Analyze only Python and JavaScript files"

unified_scrape

Multi-source scraping (docs + GitHub + PDF).

Parameters:

Example:

# Natural language
"Run unified scraping with my-config.json"
"Combine docs and GitHub for React"
"Multi-source scrape with claude-enhanced merge"

detect_patterns

Detect code patterns in repository.

Parameters:

Returns: Detected patterns

Example:

# Natural language
"Detect patterns in this codebase"
"Find architectural patterns"
"Show me the code patterns"

extract_test_examples

Extract usage examples from test files.

Parameters:

Returns: Test examples

Example:

# Natural language
"Extract test examples from tests/"
"Get Python test examples"
"Find usage examples in the test suite"

build_how_to_guides

Generate how-to guides from codebase.

Parameters:

Returns: Generated guides

Example:

# Natural language
"Build how-to guides for this project"
"Generate guides about authentication"
"Create how-to documentation"

extract_config_patterns

Extract configuration patterns.

Parameters:

Returns: Config patterns

Example:

# Natural language
"Extract config patterns from this project"
"Find configuration examples"
"Show me how this project is configured"

detect_conflicts

Find discrepancies between documentation and code.

Parameters:

Returns: Conflict report

{
  "conflicts": [
    {
      "type": "api_mismatch",
      "doc_signature": "foo(a, b)",
      "code_signature": "foo(a, b, c=default)"
    }
  ]
}

Example:

# Natural language
"Detect conflicts between docs and code"
"Find discrepancies in React"
"Compare documentation to implementation"

Config Source Tools

add_config_source

Register a git repository as a config source.

Parameters:

Example:

# Natural language
"Add my-configs repo as a source"
"Register https://github.com/org/configs as configs"

list_config_sources

List all registered config sources.

Parameters: None

Returns: List of sources

Example:

# Natural language
"List my config sources"
"Show registered sources"

remove_config_source

Remove a config source.

Parameters:

Example:

# Natural language
"Remove the configs source"
"Delete my old config source"

fetch_config

Fetch configs from a git source.

Parameters:

Example:

# Natural language
"Fetch configs from my source"
"Get the react config from configs source"

submit_config

Submit a config to a source.

Parameters:

Example:

# Natural language
"Submit my-config.json to configs source"
"Add this config to my source"

Config Splitting Tools

split_config

Split large configuration into smaller chunks.

Parameters:

Example:

# Natural language
"Split the large config into chunks"
"Break up this 500-page config"
"Split with 50 pages per chunk"

generate_router

Generate a router skill for large documentation.

Parameters:

Example:

# Natural language
"Generate a router for this large config"
"Create a router skill for Django docs"

Vector Database Tools

export_to_weaviate

Export skill to Weaviate vector database.

Parameters:

Example:

# Natural language
"Export React skill to Weaviate"
"Send to Weaviate at localhost:8080"

export_to_chroma

Export skill to ChromaDB.

Parameters:

Example:

# Natural language
"Export to ChromaDB"
"Send Django skill to Chroma"

export_to_faiss

Export skill to FAISS index.

Parameters:

Example:

# Natural language
"Export to FAISS index"
"Create FAISS index for this skill"

export_to_qdrant

Export skill to Qdrant.

Parameters:

Example:

# Natural language
"Export to Qdrant"
"Send skill to Qdrant vector DB"

Workflow Tools

list_workflows

List all available workflow presets.

Parameters: None

Returns:

{
  "workflows": [
    {"name": "security-focus", "source": "bundled"},
    {"name": "my-custom", "source": "user"}
  ]
}

Example:

# Natural language
"List available workflows"
"What workflow presets do I have?"

get_workflow

Get full YAML content of a workflow.

Parameters:

Returns: Workflow YAML

Example:

# Natural language
"Show me the security-focus workflow"
"Get the YAML for the default workflow"

create_workflow

Create a new workflow.

Parameters:

Example:

# Natural language
"Create a workflow called my-workflow"
"Save this YAML as a new workflow"

update_workflow

Update an existing workflow.

Parameters:

Example:

# Natural language
"Update my-custom workflow"
"Modify the security-focus workflow"

delete_workflow

Delete a user workflow.

Parameters:

Example:

# Natural language
"Delete my-old-workflow"
"Remove the test workflow"

Common Patterns

Pattern 1: Quick Documentation Skill

# Natural language sequence:
"List available configs"
"Scrape the react config"
"Package output/react for Claude"

Tools: list_configs → scrape_docs → package_skill

Pattern 2: GitHub Repository Analysis

# Natural language sequence:
"Scrape the facebook/react GitHub repo"
"Enhance the output/react skill"
"Package it for Gemini"

Tools: scrape_github → enhance_skill → package_skill

Pattern 3: Complete One-Command

# Natural language:
"Install the Django skill for Claude"

Tool: install_skill

Pattern 4: Multi-Source with Workflows

# Natural language sequence:
"List available workflows"
"Run unified scrape with my-unified.json"
"Apply security-focus and api-documentation workflows"
"Package for Claude"

Tools: list_workflows → unified_scrape → enhance_skill → package_skill

Pattern 5: Vector Database Export

# Natural language sequence:
"Scrape the Django documentation"
"Export to ChromaDB"

Tools: scrape_docs → export_to_chroma

Error Handling

Common Errors

Error Response Format

{
  "error": true,
  "error_type": "ConfigNotFoundError",
  "message": "Config 'react' not found",
  "suggestion": "Run list_configs to see available configs"
}

See Also

• CLI Reference - Command-line interface
• Config Format - JSON configuration
• MCP Setup Guide - Server configuration

For tool help: Ask the AI agent about specific tools