Free ATS dataset API

Endpoints

Every row carries a verified flag — true means the ATS was confirmed against the employer's live apply-URL host. Published percentages (Workday 38.0%, Greenhouse 12.6%, top-3 vendors ~60%) are computed from the 704 verified rows only.

Copy-paste examples

curl -s https://withresumeai.com/api/v1/ats/anthropic

{
  "meta": {
    "updated": "2026-07-28",
    "source": "https://withresumeai.com/reports/state-of-ats-2026",
    "license": "MIT"
  },
  "company": {
    "company": "Anthropic",
    "slug": "anthropic",
    "industry": "Technology",
    "ats": "Greenhouse",
    "verified": true,
    "hiring_volume_tier": "mid",
    "top_roles": ["software-engineer", "product-manager", "data-analyst"],
    "source_url": "https://withresumeai.com/ats-checker/anthropic"
  }
}

curl -s "https://withresumeai.com/api/v1/ats?ats=workday&verified=true"

{
  "meta": {
    "total": 738,
    "verified": 704,
    "returned": 269,
    "updated": "2026-07-28",
    "source": "https://withresumeai.com/reports/state-of-ats-2026",
    "license": "MIT",
    "docs": "https://withresumeai.com/developers",
    "filter_ats": "workday",
    "filter_verified": true
  },
  "companies": [
    {
      "company": "Salesforce",
      "slug": "salesforce",
      "industry": "Enterprise Software",
      "ats": "Workday",
      "verified": true,
      "hiring_volume_tier": "mega",
      "top_roles": ["software-engineer", "account-executive", "customer-success-manager"],
      "source_url": "https://withresumeai.com/ats-checker/salesforce"
    },
    ...268 more
  ]
}

curl -s https://withresumeai.com/api/v1/ats | jq '.meta'

{
  "total": 738,
  "verified": 704,
  "returned": 738,
  "updated": "2026-07-28",
  "source": "https://withresumeai.com/reports/state-of-ats-2026",
  "license": "MIT",
  "docs": "https://withresumeai.com/developers"
}

MCP server for AI agents

The same dataset — plus the production resume-scoring engine behind the ATS checker — is available as a remote Model Context Protocol server, so Claude, Cursor, and any other MCP-capable agent can query it natively. Streamable HTTP, no auth, no key:

https://withresumeai.com/api/mcp

claude mcp add --transport http resumeai-ats https://withresumeai.com/api/mcp

{
  "mcpServers": {
    "resumeai-ats": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://withresumeai.com/api/mcp"]
    }
  }
}

{
  "mcpServers": {
    "resumeai-ats": {
      "url": "https://withresumeai.com/api/mcp"
    }
  }
}

The three data tools are unlimited (same fair-use ask as the JSON API). check_resume runs a real scoring pass through the same engine as the web checker, so it's limited to 10 checks per day per IP — unlimited checks stay free on the site.

Agent-readable resumes (consent-gated)

A resume built on ResumeAI can be exposed as a structured, machine-readable career record that an AI agent — a recruiter's sourcing assistant, a candidate's job-search copilot — can fetch on the owner's behalf. It's served in the open JSON Resume schema, paired with an ats_context block that says how to render the profile for the owner's target applicant tracking system — recommended file format, layout, and that vendor's parsing rules, straight from the portal-verified State of ATS 2026 data above.

Consent is the whole design. A resume is returned only against a token the human owner explicitly minted and can revoke at any time. No token, no data — there is no way to browse, search, or enumerate resumes, and the endpoints are noindex. This is a candidate's verified record, exposed on their terms.

The HTTP endpoint mirrors the get_agent_resume MCP tool above — same shared logic, same consent gate:

GET /api/v1/resume/{token}   # token minted & shared by the resume's owner

{
  "standard": "JSON Resume (https://jsonresume.org/schema/)",
  "consent": "Shared by the resume's human owner via a revocable token.",
  "resume": {
    "basics": { "name": "...", "label": "...", "summary": "...", "location": {...} },
    "work": [ { "name": "...", "position": "...", "startDate": "...", "highlights": [...] } ],
    "education": [ ... ],
    "skills": [ { "name": "...", "keywords": [...] } ]
  },
  "ats_context": {
    "target_company": "Stripe",
    "ats": "Greenhouse",
    "verified": true,
    "company_page": "https://withresumeai.com/ats-checker/stripe",
    "ats_vendor": "Greenhouse",
    "recommended_format": "PDF (preferred) or .docx",
    "recommended_layout": "single-column or clean two-column",
    "parsing_rules": [ "..." ]
  }
}

How per-ATS rendering works. When the owner set a target employer on the resume, the response resolves that employer to its portal-verified ATS and attaches that vendor's parsing rules — so the agent renders the same underlying record the way Workday, Greenhouse, or iCIMS each parse best, instead of guessing at format. No target set? Look one up with which_ats first. The agent should keep every claim the owner wrote intact — this is their record, not a draft to rewrite.

Rate limits and caching

The dataset endpoints have no hard rate limit and no API key (the one exception: the check_resume scoring tool is capped at 10 checks per day per IP). Responses are cached at the edge for 24 hours (Cache-Control: public, s-maxage=86400, stale-while-revalidate), which matches how often the dataset can change. Please be nice: if you need the full dataset repeatedly, fetch the CSV once or use a mirror below instead of polling the API.