Practical recipes¶

The snippets below are MCP tool arguments. Start your server using the quick start, then use these through an MCP client.

Find specialized Python skills¶

Tool: search_skills

{"query": "Python", "limit": 5, "skill_type": "ST1"}

Use a returned ID with get_skill_details or get_related_skills. Avoid guessing skill IDs from names.

Understand a role¶

Tool: normalize_title_and_get_skills

{
  "raw_title": "Senior software engineer",
  "max_occupations": 3,
  "max_skills_per_occupation": 10,
  "confidence_threshold": 0.6
}

Check workflow_metadata.status. A partial result contains useful completed steps and an errors list; it is not a complete skill profile. If no title or occupation matches, that status is returned explicitly.

Compare two occupations¶

Tool: get_career_skill_gap

{
  "source_id": "11-1021",
  "destination_id": "11-1011",
  "dimension": "soc",
  "nation": "us",
  "limit": 10
}

Both IDs must belong to the same taxonomy. Use get_pathways_metadata to confirm supported dimensions.

Benchmark a LOT occupation¶

Tool: get_occupation_benchmark

{
  "occupation_id": "231715",
  "dimension": "lotocc",
  "metrics": ["taxonomicalInfo", "salary", "demand"],
  "region": {"nation": "us"}
}

Read the response’s top-level warnings: requested datasets can be unavailable even when the HTTP request succeeds.

Investigate demand in a city¶

Tool: get_posting_statistics

{
  "filters": {"when": "active", "city_name": ["Seattle, WA"], "title": ["ET6850661D6AE5FA86"]},
  "metrics": ["unique_postings", "median_salary"]
}

The when filter is required. Use "active" for current postings or an object with start and end months for a bounded study. Confirm available months with get_postings_metadata. The native filter contract is documented in US Job Postings.

Call through Python¶

import asyncio
from fastmcp import Client
from mcp_lightcast.server import mcp

async def main():
    async with Client(mcp) as client:
        result = await client.call_tool(
            "search_skills", {"query": "Python", "limit": 3}
        )
        # List-returning tools are wrapped in {"result": [...]} by MCP.
        print(result.structured_content)

asyncio.run(main())

Employment, wages, industries, and education¶

See the Core LMI guide for verified query examples: compare software-developer employment and earnings across states, combine counties into a custom region, request location quotients and shift-share components, and examine computer-science completions at an institution.