What makes the JobDataPool MCP server different from a generic job search MCP?

It reads JobDataPool's public source catalog, resolves the current reviewed dataset/API JSON artifact, returns JobDataPool listing ids and normalized fields, and exposes JobDataPool-specific resources and prompts for market scans, company briefs, and dataset QA.

Does the JobDataPool MCP server expose unmasked job application URLs?

No. Listing outputs mask source and application URLs behind jobdatapool.com/jobrd?id=... and also sanitize external URLs embedded in scraped title or summary text.

Which clients can use the JobDataPool MCP server?

Any MCP-compatible client that can launch a local stdio server can use it, including Claude Desktop, Claude Code, Cursor, and Windsurf.

MCP Server

JobDataPool tools for AI agents.

The JobDataPool MCP server gives Claude, Cursor, Windsurf, Claude Code, and other MCP-compatible clients a local stdio bridge into the current reviewed job data snapshot. It is built around JobDataPool provenance: source catalog, contract URLs, dataset cache status, listing ids, masked redirect URLs, and guided analyst workflows.

Install locally View tools URL masking

Why it is JobDataPool-native

Not a generic job search wrapper

The server reads the public /v1/sources catalog, resolves the current monthly API JSON artifact, and keeps JobDataPool contracts visible to the client. Agents can ask for listings and also see where the snapshot came from.

Source-aware

Clients can inspect repository, DVC pointer, CSV/API artifact, schema, OpenAPI, and endpoint pointers through get_sources and jobdatapool://source-catalog.

Snapshot-aware

Search results include the dataset source artifact, and get_sources reports whether the local MCP cache has loaded rows.

Listing-id first

Agent responses are grounded in stable JobDataPool ids and masked /jobrd?id=... handoff URLs rather than raw origin URLs.

Analyst workflows

Prompts shape the MCP into market scans, company briefs, and dataset QA instead of leaving users to guess which tool to call first.

Local and private

The package runs locally over stdio. Your MCP client launches it; JobDataPool is queried only for the public source catalog and data snapshot.

Contract-compatible

The MCP surface points back to /openapi.json, JSON Schemas, API docs, and dataset releases.

Install

Install the published PyPI MCP server

The JobDataPool MCP server is published as jobdatapool-mcp on PyPI. It runs locally over stdio and requires Python 3.9 or newer.

Install from PyPI

python -m pip install --upgrade jobdatapool-mcp
jobdatapool-mcp

The package installs a local command named jobdatapool-mcp. Your MCP client launches that command when it needs JobDataPool tools.

Claude / Cursor / Windsurf config

{
  "mcpServers": {
    "jobdatapool": {
      "command": "jobdatapool-mcp",
      "env": {
        "JOBDATAPOOL_BASE_URL": "https://jobdatapool.com"
      }
    }
  }
}

No local repository checkout or cwd is required when installed from PyPI. Point JOBDATAPOOL_BASE_URL at localhost only when testing a local JobDataPool build.

One-command run with uvx

uvx jobdatapool-mcp

If your client supports launching commands through uvx, this can run the published package without a manual virtual environment.

uvx client config

{
  "mcpServers": {
    "jobdatapool": {
      "command": "uvx",
      "args": ["jobdatapool-mcp"],
      "env": {
        "JOBDATAPOOL_BASE_URL": "https://jobdatapool.com"
      }
    }
  }
}

Use this form when you want the MCP client to fetch and run the latest published package directly from PyPI.

Tools

Search, summarize, inspect, and compare listings

All listing outputs use public JobDataPool fields and masked handoff URLs.

Tool	Purpose	Good prompt
`search_jobs`	Search by keyword, skills, salary signal, remote signal, seniority, location, country, company, employment type, industry, posted window, and sort order.	Find remote senior Python security jobs in the US.
`get_job`	Return one full listing by JobDataPool id.	Tell me more about listing id `...`.
`get_company`	Summarize a company from matching listings, including locations, seniority, industries, and examples.	Brief me on Huntress in the current snapshot.
`get_filter_options`	Show high-count filter values for seniority, skills, employment type, industries, countries, locations, and companies.	What skills are common in this dataset?
`find_similar_jobs`	Find similar listings by lexical overlap across title, skills, company, industries, and summary.	Find jobs similar to this listing id.
`get_sources`	Return source catalog, API contract pointers, and MCP dataset cache status.	What snapshot is this MCP reading?

Resources and Prompts

A discoverable agent cockpit

Clients that support MCP resources and prompts get a more guided experience than a bare tool list.

Resources

jobdatapool://source-catalog — live /v1/sources catalog plus base URL.
jobdatapool://analyst-context — short guide explaining dataset scope, cache behavior, provenance, and caveats.
jobdatapool://contracts — OpenAPI, schemas, v1 endpoint URLs, and MCP tool names.

Prompts

jobdatapool_market_scan — role, skill, industry, or hiring-theme scan.
jobdatapool_company_brief — company footprint summary with examples and caveats.
jobdatapool_dataset_qa — source/catalog health and facet coverage inspection.

Privacy and Source Protection

Listing URLs stay masked

The MCP server is designed to preserve JobDataPool's publisher-protection posture. It does not hand agents raw source or application URLs from listings.

What gets masked

url, apply_link, posting_url, and source URL fields are returned as https://jobdatapool.com/jobrd?id=....
External URLs embedded inside scraped text fields, including titles and summaries, are sanitized before output.
Company examples and similar-job results use the same compact masked listing shape.

What remains visible

JobDataPool listing id, title, company, location, country, seniority, employment type, industries, salary text, posted date, skills, and summary.
Dataset provenance URLs such as the R2 API JSON artifact in source/cache metadata.
Public contract URLs such as /openapi.json, /schemas/, and /v1/sources.

Example Agent Requests

Use it like a JobDataPool analyst

Market scan

Find remote senior Python security jobs in the US, include representative JobDataPool ids, salary signals, companies, and the dataset source.

Company brief

Brief me on Huntress from the current JobDataPool snapshot. Include active listing count, location mix, seniority, industries, and examples.

Dataset QA

Inspect the current JobDataPool source catalog and show the top skills facet values. Tell me which artifact the MCP loaded.

Similarity

Given this JobDataPool listing id, find similar jobs and explain the overlap by title, skills, industry, and summary language.

Filter discovery

Show the highest-count countries and industries available in the current snapshot before I decide which market to analyze.

Provenance check

What source catalog, OpenAPI contract, schemas, and cache status is the JobDataPool MCP using right now?