Overview

Datasets

Datasets are the logical data domains exposed through Lightcast MCP tools. They help AI agents and applications choose the right labor market lens before selecting dimensions, metrics, filters, and tools.

Lightcast MCP currently documents three validated datasets:

Dataset	Identifier	Primary Lens	Best For
Job Posting Dataset	`job_postings`	Labor demand	Hiring activity, employer requirements, skills demand, compensation signals, and posting trends
Profile Dataset	`profiles`	Talent supply	Workforce composition, skills, education, experience, employer affiliation, and career background
Core LMI	`government_occupations`, `government_unemployment`, `government_industry`	Government-sourced labor market information	Occupational employment, earnings, unemployment, labor force participation, industry activity, and workforce trends

Datasets are logical analytical domains exposed through MCP tools. They are not physical databases, backend services, or API endpoints.

Why Datasets Matter

Datasets are central to how Lightcast MCP tools behave. The selected dataset determines which capabilities are available and how requests are validated.

Dataset-Driven Decision	Why It Matters
Supported Tools	Not every tool operates on every dataset.
Available Dimensions	Filters and ranking dimensions differ by dataset.
Available Metrics	Metrics are dataset-specific and should not be reused across datasets unless documented.
Validation Behavior	Tool inputs are validated against the selected dataset and tool contract.
Analytical Meaning	The same broad question can mean demand, supply, or official occupational baseline depending on the dataset.

Logical Datasets vs. Physical Sources

Lightcast MCP exposes logical datasets so users can think in terms of analytical questions rather than backend implementation details.

Logical Dataset Concept	Physical-Source Concept
I need hiring demand.	Which underlying service contains job advertisement data?
I need talent supply signals.	Which underlying service contains profile records?
I need government occupation baselines.	Which underlying service contains official occupational labor market data?

This abstraction allows MCP tools to expose normalized dimensions, metrics, and response structures across datasets.

Choosing a Dataset

Use the business question to determine which dataset best fits the analysis.

If the question is about...	Use this dataset
Employer hiring activity, advertised requirements, posting salaries, or demand trends	`job_postings`
Skills people have, education background, employer affiliation, workforce composition, or estimated wages	`profiles`
Government-sourced occupational employment, earnings, unemployment, labor force participation, or industry activity	`Core LMI`

📝
when an agent is unsure which dataset to use, start with dataset discovery before selecting an analysis tool.

Relationship to Dimensions, Metrics, and Tools

Datasets organize the rest of the Lightcast MCP analytical model.

Concept	Relationship to Datasets
Dimensions	Dimensions are available only where a dataset supports them.
Metrics	Metrics are scoped to datasets and tools.
Tools	Tools operate only on supported dataset and tool combinations.
Filters	Filters are composed of supported dataset dimensions.

Dataset Comparison

Dataset	Measures	Supported MCP tools
`job_postings`	Labor demand represented by deduplicated online job advertisements	`count`, `distribution`, `estimate`, `list`, `timeseries`, `top`
`profiles`	Talent supply represented by professional profile records	`count`, `distribution`, `estimate`, `top`
`Core LMI`	Government-sourced occupational, unemployment, and industry labor market measures	`estimate`, `timeseries`, `top` and `distribution` is also supported for Government Occupations

💡
Core LMI is a category, not a dataset identifier that you pass directly to an MCP tool. When making a request, use the specific identifier: government_occupations, government_unemployment, or government_industry.