Dated NL + US-pilot snapshot
JOBS MCP SERVER / PRIVATE BETA
A job posting MCP server forblue-collar jobs.
Connect an AI assistant to blue-collar jobs surfaced from hard-to-find public pages—including jobs from small businesses outside the standard ATS footprint. The private-beta jobs MCP will turn natural-language questions into source-linked searches across employers, postings and dated market snapshots.
Hosted remote MCP Read-only tools Source-linked results
Planned beta connectors
Claude
Cursor
WindsurfBLUE COLLAR JOB DATA MCP● PRIVATE BETA
YOU ASK
Which metal employers in the Netherlands have job records observed this week? Include the source pages.
ASSISTANT SELECTS TOOL
find_hiring_employers- market
- NL
- category
- metal
- observed_after
- 2026-08-02
STRUCTURED RESULTEmployer evidence + record countsSource URLs · observation dates · field limitations
Small, discoverable tool surface
No posting, applying or messaging
Trace results to public pages
01 / BUILT FOR AI-ASSISTED DATA WORK
Ask a hiring-market question. Get inspectable job data back.
A job listing MCP server gives the assistant a controlled route to current product data instead of asking it to guess, browse arbitrary pages or work from pasted spreadsheets.
01
Search listings conversationally
Describe the market, trade, employer or observation window in plain language. The assistant maps the request to typed filters.
02
Keep every answer traceable
Return public source routes and observation context so a user can inspect the evidence behind the result.
03
Constrain what the agent can do
Expose a small read-only tool set with server-side limits instead of raw collection-system or database access.
02 / HOW THE JOBS MCP WORKS
One connector between the AI client and the job-data layer.
The planned beta uses a hosted remote MCP service. Your assistant discovers the available tools, chooses one for the question and receives a bounded structured response from a sanitized read model.
01
Connect the AI client
Add the beta MCP URL to a supported client and complete the supplied authentication flow.
02
Discover the tool schemas
The client reads the tool names, descriptions and accepted arguments from the MCP server.
03
Run a validated query
The server validates market, category, dates and limits before querying a sanitized job-data view.
04
Return evidence
The client receives structured records, source links, observation dates and explicit limitations.
CLAUDE · CHATGPT · CURSOR · WINDSURFREMOTE MCPVALIDATED READ MODELSOURCE-LINKED JOB DATA
03 / CONNECTION SHAPE
Beta access supplies the URL, credentials and tested setup steps.
A supported client stores one remote server entry. On first use, it authenticates, discovers the tools and asks for permission before making a call according to that client's controls.
- Authentication scoped to the beta account
- Read-only tools and bounded result sizes
- Hosted connection—no collection infrastructure or database installed locally
ILLUSTRATIVE CLIENT CONFIGVALUES ARRIVE WITH BETA ACCESS
{
"mcpServers": {
"blue-collar-job-data": {
"url": "<BETA_MCP_URL>",
"headers": {
"Authorization":
"Bearer <BETA_ACCESS_TOKEN>"
}
}
}
}Placeholder only. The production URL, transport and authentication method will be documented after beta interoperability testing.
04 / PRIVATE BETA TOOLSET
A small tool set for job listings, employers and market snapshots.
The private beta starts with four focused tools. Each does one job and returns structured evidence; schemas will be refined from real beta workflows before general access.
search_blue_collar_jobsSearch job postings
Find observed records by market, broad category, employer, place context and observation date.
Returns a bounded record set with source URLs.get_job_recordInspect one job record
Retrieve the available source, observation and classification fields for one result.
Returns field values plus record-level caveats.find_hiring_employersFind hiring employers
Group matching open-status records by employer inside a defined market slice.
Returns counts, representative records and sources.summarize_market_snapshotSummarize a market
Create an aggregate view for one explicit snapshot, market and category scope.
Returns counts, denominator, date and limitations.05 / WHAT YOU CAN BUILD
Move from question to source-linked market evidence inside the assistant.
Find employers associated with observed blue-collar hiring in a defined territory before account research.
Explore a dated slice by category and employer while keeping the denominator and limitations visible.
Prototype agent-led search and research features before committing to a larger API or bulk integration.
Ask an assistant to find and compare relevant blue-collar openings while keeping dates, employers and original source links attached.
06 / MCP VS API VS BULK
Use the same data contract through the right delivery path.
AI assistants and agent workflows
Natural-language exploration, tool discovery and source-linked follow-up questions.
Request beta accessDeterministic application logic
Controlled machine-to-machine queries for products, alerts and internal systems.
Explore the APIWarehouses and repeatable analysis
Defined CSV or JSON extracts for joins, models, QA and larger market studies.
Explore bulk data07 / PRIVATE BETA
Bring the workflow. We'll bring the connector.
Tell us the market, blue-collar categories, preferred AI client and two or three questions you need the jobs MCP to answer. Beta access will prioritize workflows the current dataset can support honestly, from bounded individual research to team use cases.
INCLUDE IN YOUR REQUEST
- MARKET
- Country or region
- CATEGORIES
- Trades or frontline roles
- CLIENT
- Claude, ChatGPT, Cursor or other
- QUESTIONS
- The outputs your workflow needs
08 / JOB POSTING MCP FAQ
Before you connect a jobs MCP server.
The beta is being built for read-only job search and market research by individuals and teams—not job publishing or automated applications.
Read the official MCP introduction01What is a jobs MCP or job listing MCP server?
A jobs MCP server exposes job-listing data as typed tools that an MCP-compatible AI assistant can discover and call. Instead of copying results into a chat, the assistant can request a defined market slice, receive structured records and keep the source evidence attached.
02How will I connect to the Blue Collar Job Data MCP server?
Beta users will receive a hosted remote MCP URL, an authentication method and tested client-specific instructions. After the connector is added, the client discovers the available tools automatically. The final URL and authentication flow will be published after interoperability and access-control testing.
03Is this job posting MCP for reading or publishing jobs?
It is designed for reading and analyzing source-linked job-posting data. The beta will not publish vacancies to job boards, create ATS requisitions, apply to jobs, message candidates or modify employer systems.
04Can individuals use this as a job search MCP?
Yes. An individual could ask an AI assistant to find and compare relevant blue-collar openings, employers, places and recent hiring signals while retaining links to the original sources. A bounded self-serve plan for individual research is a candidate launch tier; the beta is being used to validate usage and limits. The connector would support research, not submit applications or contact employers on the user's behalf.
05Which AI clients will the jobs MCP support?
Claude, ChatGPT, Cursor and Windsurf are target clients for beta interoperability testing because they support MCP-style connector workflows. A client will be listed as supported only after its connection and authentication flow has been tested.
06What tools will be available in the private beta?
The proposed first tool set searches blue-collar job records, retrieves one record with evidence, finds employers associated with observed hiring and summarizes a dated market snapshot. Tool names and schemas may change in response to beta evaluation.
07What will a job record contain?
A result can include the source title, employer, broad blue-collar category, reported or company-context place, first-seen and last-seen observations and the public source URL. Exact availability varies by source and is returned with field limitations.
08Does the MCP include candidates or contact data?
No. The planned beta is a read-only interface to employer and job-posting evidence. It does not promise candidate profiles, personal contact data, verified decision-makers, outreach actions or application automation.
09When should I use the MCP instead of the API or bulk data?
Use MCP when an AI assistant should choose a tool and explore the data conversationally. Use the API for deterministic application logic and bulk CSV or JSON for warehouses, repeatable analysis and larger extracts.
010How do I request beta access?
Email the market, blue-collar categories, intended MCP client and two or three questions the connector should answer. This lets the team match beta access to a concrete workflow and prioritize the right filters and outputs.