mcp-lightcast 0.2.0

MCP server for Lightcast API integration using FastMCP

MCP Lightcast Server

A production-ready Model Context Protocol (MCP) server that provides seamless integration with Lightcast APIs for job titles, skills analysis, and career data. Built with FastMCP and modern Python development practices.

🎯 Current Status: v0.1.0 - Production Ready with 61% API Coverage (11/18 endpoints working)

🚀 Features

✅ Working APIs & Endpoints (11/18 total)

🎯 Skills API (6/10 endpoints) - Version 9.33, 41,139 skills

• ✅ Skills Search - Search with filters (type, category, subcategory)
• ✅ Individual Skill Retrieval - Get detailed skill information by ID
• ✅ Skills Extraction from Text - Extract skills from job descriptions with confidence scores
• ✅ Bulk Skills Retrieval - Efficient batch processing of multiple skills
• ✅ Version Metadata - Complete API version and statistics information
• ✅ Skills Metadata - General skills taxonomy information

🏷️ Titles API (5/8 endpoints) - Version 5.47, 73,993 titles

• ✅ Job Title Search - Search Lightcast's comprehensive job title database
• ✅ Individual Title Retrieval - Get detailed title information by ID
• ✅ Bulk Title Retrieval - Efficient batch processing of multiple titles
• ✅ Version Metadata - Complete API version and statistics information
• ✅ General Metadata - Latest version and attribution information

🔧 Core Functionality Working

• 🎯 Skills Extraction from Text - High accuracy skill identification from job descriptions
• 📊 Search & Discovery - Fast, filtered search across both skills and titles taxonomies
• ⚡ Bulk Operations - Efficient processing of multiple items in single requests
• 🔄 Version Management - Uses "latest" keyword with backward compatibility
• 🔐 OAuth2 Authentication - Secure authentication with Lightcast APIs

⚠️ Limited/Premium Features

• ❌ Title Normalization - Requires premium authentication scope
• ❌ Skills Categories - Endpoint not available in current API version
• ❌ Related Skills - Endpoint pattern differs from documentation
• ❌ Title Hierarchy - Endpoint pattern differs from documentation

🛠️ MCP Tools Available

Skills Tools

• search_skills - Search skills with advanced filters
• get_skill_details - Get detailed skill information
• extract_skills_from_text - Extract skills with custom confidence threshold
• extract_skills_simple - Extract skills with default settings
• bulk_retrieve_skills - Efficient bulk skill retrieval
• get_skills_version_metadata - API version information

Titles Tools

• search_job_titles - Search job titles
• get_job_title_details - Get detailed title information
• bulk_retrieve_titles - Efficient bulk title retrieval
• get_titles_version_metadata - API version information
• get_titles_general_metadata - General taxonomy information

Combined Workflow Tools

• normalize_title_and_get_skills - Complete workflow (where normalization is available)
• analyze_job_posting_skills - Comprehensive job posting analysis using skills extraction

🛠️ Installation

Prerequisites

• Python 3.12+ (required for uv-dynamic-versioning)
• uv package manager (recommended) or pip
• Lightcast API credentials (Client ID and Secret with emsi_open scope)

🚀 Quick Start with uvx (Recommended)

# Install and run directly from PyPI (no installation required)
uvx --from mcp-lightcast mcp-lightcast --help

# Run with environment variables
LIGHTCAST_CLIENT_ID=your_id LIGHTCAST_CLIENT_SECRET=your_secret \
uvx --from mcp-lightcast mcp-lightcast

# Use stdio transport for Claude Desktop
LIGHTCAST_CLIENT_ID=your_id LIGHTCAST_CLIENT_SECRET=your_secret \
uvx --from mcp-lightcast mcp-lightcast --transport stdio

📦 Install from PyPI

# Install globally
pip install mcp-lightcast

# Or with uv
uv tool install mcp-lightcast

# Run the server
mcp-lightcast --help

🔧 Development Installation

# 1. Clone the repository
git clone https://github.com/your-org/mcp-lightcast.git
cd mcp-lightcast

# 2. Set up development environment 
make setup

# 3. Configure your API credentials
# Edit .env with your Lightcast API credentials

# 4. Validate configuration
make validate-config

# 5. Run the server
make run

🐳 Docker Installation

# Pull the latest image (when available)
docker pull ghcr.io/your-org/mcp-lightcast:latest

# Run with environment variables
docker run --rm -it \
  -e LIGHTCAST_CLIENT_ID=your_id \
  -e LIGHTCAST_CLIENT_SECRET=your_secret \
  ghcr.io/your-org/mcp-lightcast:latest

# Or with environment file
docker run --rm -it --env-file .env ghcr.io/your-org/mcp-lightcast:latest

⚙️ Configuration

Environment Variables

Create a .env file with your Lightcast API credentials:

# Required - Lightcast API Configuration
LIGHTCAST_CLIENT_ID=your_client_id_here
LIGHTCAST_CLIENT_SECRET=your_client_secret_here

# Optional - API Configuration (with defaults)
LIGHTCAST_BASE_URL=https://api.lightcast.io
LIGHTCAST_OAUTH_URL=https://auth.emsicloud.com/connect/token
LIGHTCAST_OAUTH_SCOPE=emsi_open
LIGHTCAST_RATE_LIMIT=1000

# Optional - MCP Server Configuration
MCP_SERVER_NAME=lightcast-mcp-server
LOG_LEVEL=INFO
MASK_ERROR_DETAILS=true

Lightcast API Access

To use this server, you need:

1. 📝 A Lightcast API account
2. 🔑 Client ID and Client Secret for OAuth2 authentication
3. 🎯 Access to the following Lightcast APIs:
   • Titles API - Job title search and normalization
   • Skills API - Skills search and categorization
   • Classification API - Occupation code mapping
   • Similarity API - Skills and occupation relationships

Contact Lightcast for API access and credentials.

🎯 Usage

Command Line Interface

The server includes a comprehensive CLI with multiple options:

# Basic usage (uses streamable-http on port 3000 by default)
mcp-lightcast

# Use stdio transport (for Claude Desktop integration)
mcp-lightcast --transport stdio

# Use streamable-http transport with custom port
mcp-lightcast --transport streamable-http --port 8080

# With custom log level
mcp-lightcast --log-level DEBUG

# Validate configuration without starting server
mcp-lightcast --validate-config

# Use custom environment file
mcp-lightcast --env-file /path/to/custom.env

# Quiet mode (no logging)
mcp-lightcast --quiet

# Show help
mcp-lightcast --help

Development Commands

Using the included Makefile for easy development:

# Quick development setup and run
make dev

# Run with debug logging
make dev-server

# Run all quality checks
make check

# Run tests with coverage
make test-coverage

# Show Claude Desktop configuration
make claude-config

Claude Desktop Integration

Using uv (Recommended)

{
  "mcpServers": {
    "lightcast": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/path/to/mcp-lightcast",
        "mcp-lightcast"
      ],
      "env": {
        "LIGHTCAST_CLIENT_ID": "your_client_id",
        "LIGHTCAST_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

Using Docker

{
  "mcpServers": {
    "lightcast": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e", "LIGHTCAST_CLIENT_ID",
        "-e", "LIGHTCAST_CLIENT_SECRET",
        "ghcr.io/your-org/mcp-lightcast:latest"
      ],
      "env": {
        "LIGHTCAST_CLIENT_ID": "your_client_id",
        "LIGHTCAST_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

Using uvx (Isolated)

{
  "mcpServers": {
    "lightcast": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp-lightcast",
        "mcp-lightcast"
      ],
      "env": {
        "LIGHTCAST_CLIENT_ID": "your_client_id",
        "LIGHTCAST_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

🛠️ Available MCP Tools

🎯 Skills Tools

search_skills - Search skills with advanced filters

skills = await search_skills(
    query="python programming",
    limit=10,
    skill_type="Hard Skill",  # Optional: filter by skill type
    category="Information Technology",  # Optional: filter by category
    version="latest"  # Uses latest API version
)

extract_skills_from_text - Extract skills from job descriptions

# Extract skills with custom confidence threshold
skills = await extract_skills_from_text(
    text="Looking for Python developer with React and database experience...",
    confidence_threshold=0.7,
    version="latest"
)

extract_skills_simple - Extract skills with default settings

# Quick skills extraction with default confidence (0.5)
skills = await extract_skills_simple(
    text="We need Java developers with Spring Boot experience",
    version="latest"
)

bulk_retrieve_skills - Efficient bulk skill retrieval

# Get multiple skills in one request
skills = await bulk_retrieve_skills(
    skill_ids=["KS125LS6N7WP4S6SFTCK", "KS440C66FGP5WGWYMP0F"],
    version="latest"
)

🏷️ Titles Tools

search_job_titles - Search job titles

titles = await search_job_titles(
    query="software engineer",
    limit=10,
    version="latest"
)

get_job_title_details - Get detailed title information

title = await get_job_title_details(
    title_id="ET6850661D6AE5FA86",
    version="latest"
)

bulk_retrieve_titles - Efficient bulk title retrieval

titles = await bulk_retrieve_titles(
    title_ids=["ET6850661D6AE5FA86", "ETBF8AE9187B3810C5"],
    version="latest"
)

📊 Metadata Tools

get_skills_version_metadata - API version information

metadata = await get_skills_version_metadata(version="latest")
# Returns: version, skill_count, language_support, skill_types, etc.

get_titles_version_metadata - API version information

metadata = await get_titles_version_metadata(version="latest")
# Returns: version, title_count, removed_title_count, fields

⚠️ Limited Availability Tools

Some tools require premium authentication scopes or have endpoint limitations:

normalize_job_title - ❌ Requires premium scope

# Currently returns 401 Unauthorized with emsi_open scope
result = await normalize_job_title("sr software dev")

analyze_job_posting_skills - ✅ Working via skills extraction

# Uses skills extraction instead of normalization
result = await analyze_job_posting_skills(
    job_title="Software Engineer",
    job_description="Full job description text...",
    extract_from_description=True  # Uses working skills extraction
)

🎯 Example Workflows

1. Extract Skills from Job Description

# Analyze a job posting to extract relevant skills
job_description = """
We're looking for a Senior Software Engineer with expertise in Python, 
React, and cloud technologies. Experience with Docker, Kubernetes, 
and AWS is required. Strong communication skills and team collaboration 
abilities are essential.
"""

# Extract skills with high confidence
skills = await extract_skills_from_text(
    text=job_description,
    confidence_threshold=0.8,
    version="latest"
)

print(f"High-confidence skills found: {len(skills)}")
for skill in skills:
    print(f"- {skill['name']} (confidence: {skill['confidence']:.2f})")

2. Compare Skills Across Job Titles

# Search and compare skills requirements for different roles
titles = ["Data Scientist", "Machine Learning Engineer", "Software Engineer"]
title_skills = {}

for title in titles:
    # Search for the title
    title_results = await search_job_titles(query=title, limit=1)
    if title_results:
        title_id = title_results[0]['id']
        
        # Get detailed title information  
        title_details = await get_job_title_details(title_id)
        title_skills[title] = title_details
        
print("Job title comparison completed")

3. Bulk Skills Analysis

# Efficiently analyze multiple skills at once
skill_names = ["Python", "JavaScript", "Machine Learning", "Docker"]

# First search for skill IDs
skill_ids = []
for name in skill_names:
    results = await search_skills(query=name, limit=1)
    if results:
        skill_ids.append(results[0]['id'])

# Get detailed information for all skills in one request
if skill_ids:
    detailed_skills = await bulk_retrieve_skills(skill_ids)
    
    for skill in detailed_skills:
        print(f"Skill: {skill['name']}")
        print(f"Type: {skill.get('type', {}).get('name', 'Unknown')}")
        print(f"Category: {skill.get('category', 'Unknown')}")
        print("---")

4. API Version and Statistics

# Get comprehensive API information
skills_meta = await get_skills_version_metadata()
titles_meta = await get_titles_version_metadata()

print(f"Skills API v{skills_meta['version']}: {skills_meta['skill_count']:,} skills")
print(f"Languages: {', '.join(skills_meta['language_support'])}")
print(f"Skill types: {len(skills_meta['skill_types'])}")

print(f"Titles API v{titles_meta['version']}: {titles_meta['title_count']:,} titles")

🧪 Development

Prerequisites

• Python 3.12+ (required)
• uv package manager
• Docker (for containerized development)
• Make (for development commands)

Development Setup

# Clone and setup
git clone https://github.com/your-org/mcp-lightcast.git
cd mcp-lightcast

# Quick setup (installs dependencies, creates .env)
make setup

# Install development dependencies only  
make install-dev

# Run development server with debug logging
make dev-server

Project Structure

mcp-lightcast/
├── 📁 src/mcp_lightcast/           # Main package
│   ├── __init__.py                 # CLI entry point with Click
│   ├── __main__.py                 # Module execution entry
│   ├── server.py                   # FastMCP server instance
│   ├── 📁 auth/                    # Authentication modules
│   │   ├── __init__.py
│   │   └── oauth.py               # OAuth2 implementation
│   ├── 📁 apis/                    # API client modules  
│   │   ├── __init__.py
│   │   ├── base.py                # Base client with error handling
│   │   ├── titles.py              # Titles API client
│   │   ├── skills.py              # Skills API client
│   │   ├── classification.py      # Classification API client
│   │   └── similarity.py          # Similarity API client
│   ├── 📁 tools/                   # MCP tools registration
│   │   ├── __init__.py
│   │   ├── titles_tools.py        # Title-related MCP tools
│   │   ├── skills_tools.py        # Skills-related MCP tools
│   │   ├── workflow_tools.py      # Combined workflow tools
│   │   └── normalize_title_get_skills.py  # Core workflow logic
│   └── 📁 utils/                   # Utility functions
│       └── __init__.py
├── 📁 tests/                       # Test suite
│   ├── 📁 unit/                    # Unit tests
│   ├── 📁 integration/             # Integration tests
│   └── conftest.py                # Pytest fixtures
├── 📁 config/                      # Configuration management
│   └── settings.py                # Pydantic settings
├── 📁 .github/workflows/           # CI/CD pipelines
│   └── ci.yml                     # GitHub Actions workflow
├── 🐳 Dockerfile                   # Production container
├── 🐳 Dockerfile.dev               # Development container  
├── 🐳 docker-compose.yml           # Multi-service setup
├── 📋 Makefile                     # Development commands
├── 📦 pyproject.toml               # Project metadata & dependencies
├── 🔒 uv.lock                      # Dependency lock file
└── 📖 README.md                    # This file

Development Workflow

Code Quality & Testing

# Run all quality checks (lint + type-check + test)
make check

# Individual quality checks
make lint           # Ruff linting
make type-check     # MyPy type checking  
make format         # Black + Ruff formatting

# Testing options
make test           # Run all tests
make test-coverage  # Tests with coverage report
make test-basic     # Basic functionality test

Docker Development

# Build Docker images
make docker-build       # Production image
make docker-build-dev   # Development image

# Run with Docker
make docker-run         # Run production container
make docker-dev         # Run development container

# Test Docker configuration
make docker-test        # Validate container setup

uv Package Management

# Dependency management
make uv-lock           # Generate lockfile
make uv-sync           # Sync from lockfile
make uv-update         # Update all dependencies

# Add dependencies
make uv-add PACKAGE=requests
make uv-add-dev PACKAGE=pytest-mock

API Reference

Rate Limits

• Default: 1000 requests per hour per API key
• Rate limit headers are included in responses
• Rate limit errors (429) are handled gracefully

Error Handling

• Authentication errors are automatically retried
• Rate limits include reset time information
• API errors include detailed status codes and messages
• Network errors are handled with appropriate timeouts

API Version Flexibility

The MCP server provides flexible version management:

• Default: "latest" - Always uses the newest available API version
• Backward Compatible: Users can specify any previous version (e.g., "5.47", "9.33")
• Future-Proof: Automatically gets new API features when Lightcast releases updates

Examples:

# Use latest version (default)
search_job_titles("software engineer")

# Use specific version for consistency
search_job_titles("software engineer", version="5.47")

# Use older version if needed
search_skills("python", version="9.32")

Current API Versions:

• Skills API: 9.33 with 41,139 skills (English, Spanish, French support)
• Titles API: 5.47 with 73,993 titles
• Both APIs use "latest" keyword for automatic version management

📈 Current Limitations & Future Enhancements

⚠️ Known Limitations (v0.1.0)

Authentication Scope Limitations

• Title Normalization: Requires premium authentication scope (currently 401 Unauthorized)
• Skills Categories: Endpoint not available in current API tier
• Related Skills: Endpoint pattern differs from documentation

Endpoint Coverage

• 61% Coverage: 11 out of 18 planned endpoints working
• Core Functionality: ✅ All essential features (search, extraction, bulk ops) working
• Premium Features: ❌ Some advanced features require paid Lightcast tiers

🚀 Planned Enhancements

Additional Lightcast APIs

• Classification API: Map concepts to occupation codes (O*NET SOC)
• Similarity API: Find similar occupations and skills
• Occupation Benchmark API: Industry benchmarking and compensation data
• Career Pathways API: Career progression and pathway analysis
• Job Postings API: Real-time job market data and trends

Feature Improvements

• Premium Authentication: Support for advanced Lightcast scopes
• Enhanced Error Handling: Better handling of rate limits and API errors
• Caching Layer: Response caching for improved performance
• Additional Export Formats: CSV, JSON, XML output options

📊 Production Readiness

✅ Ready for Production Use:

• Skills extraction from job descriptions
• Skills and titles search functionality
• Bulk data retrieval operations
• API version management and metadata

⚠️ Requires Premium Lightcast Access:

• Job title normalization workflows
• Skills categorization features
• Advanced relationship mapping

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

📚 Documentation & Support

API Documentation

• Lightcast API Documentation - Official Lightcast API reference
• FastMCP Documentation - FastMCP framework documentation
• Model Context Protocol - MCP specification
• Testing Summary - Comprehensive endpoint testing results

Project Resources

• PyPI Package - Official Python package
• GitHub Repository - Source code and issues
• GitHub Releases - Version history and changelog

Getting Help

• Issues: GitHub Issues for bugs and feature requests
• Discussions: GitHub Discussions for questions and community support
• Lightcast Support: Contact Lightcast for API access and credentials

Current Status

• Version: 0.1.0 (Production Ready)
• API Coverage: 11/18 endpoints (61%)
• Python: 3.12+ required
• License: MIT