Free API and MCP server

Try it

Open any of these in a browser, or fetch them from your code. They're plain JSON files.

• Every endpoint, the MCP URL and the licence
  • https://footballapis.com/api/v1/index.json
• All 17 providers
  • https://footballapis.com/api/v1/providers.json
• One provider (football-data.org)
  • https://footballapis.com/api/v1/providers/football-data-org.json
• What changed, newest first
  • https://footballapis.com/api/v1/changes.json
• The latest daily reachability check
  • https://footballapis.com/api/v1/checks.json

Each provider also has its own file at /api/v1/providers/<slug>.json, where the slug is the last part of its page URL (e.g. /p/football-data-org/).

Add to your AI assistant

Our MCP server lets assistants like Claude, Cursor or ChatGPT look up providers, compare them and check what changed, with sources for every fact. The address is:

• https://footballapis.com/mcp

It's read-only and needs no login. Opening the address in a browser shows an error: it's meant for assistants, not people.

claude mcp add --transport http footballapis https://footballapis.com/mcp

Claude desktop app or claude.ai: Settings → Connectors → Add custom connector, then paste the address above (where your plan supports custom connectors).

{
  "mcpServers": {
    "footballapis": {
      "url": "https://footballapis.com/mcp"
    }
  }
}

ChatGPT: where custom MCP connectors are supported (developer mode, depending on your plan), open Settings → Connectors (also called Apps), add a custom connector, paste the address above and choose no authentication.

{
  "mcpServers": {
    "footballapis": {
      "type": "http",
      "url": "https://footballapis.com/mcp"
    }
  }
}

Menu names change often; if yours differ, look for "MCP", "connectors" or "tools" in your assistant's settings. Some tools call the transport "Streamable HTTP" or "HTTP".

What your assistant can do with it

• list_providers

  List providers, optionally only those with a free plan (or trial), under a monthly budget, covering a league or offering a data type.

• get_provider

  Everything we know about one provider: every price tier, free plan, leagues, data types, links and sources.

• compare_providers

  Two to five providers side by side.

• recent_changes

  What changed recently: price corrections, confirmed prices, changed source pages and site updates.

• provider_health

  Our daily check: was the provider’s public page (or keyless endpoint) reachable, and how fast. Not API uptime.

• recommend_provider

  Three picks for a need (free to start, live scores, big European leagues, betting odds, cheapest paid plan), by the same fixed rule as our home page.

Try asking:

• "Which football data APIs have a free plan and cover the Premier League?"
• "Compare football-data.org, Sportmonks and API-Football."
• "What is the cheapest paid plan for betting odds data?"

Every answer includes the sources and the date each group of facts was last checked, plus "Facts as of 2026-10-19. Verify on the vendor page." Recommendations use the same fixed rules as Pick by what you need; nobody pays for placement. Provider names are forgiving ("API-Football", "opta", "statsbomb"); if one is ambiguous, the answer lists the closest matches. A data type that costs extra reads "yes (paid add-on)", with the price when the vendor states it.

Licence

The data is free to use, including commercially, under CC BY 4.0. Please credit FootballAPIs.com with a link to footballapis.com, for example: "Data: FootballAPIs.com (CC BY 4.0)".

Prices, plans and coverage belong to each vendor and change without notice. We take care, but can't guarantee the data is complete or current.

Please be kind to it

• The data changes at most once a day, usually far less. Cache it — once a day is plenty.
• Keep to about one request per second, and don't hammer the server in loops.
• Need everything? Fetch providers.json once instead of every provider file.

There's no API key and no hard limit today. If abuse appears we may add rate limits.

Daily checks: how we check

Every day at 05:17 UTC we send one request to one public address per provider and record whether it answered and how fast. Provider pages show the latest result, e.g. "Checked today: public page reachable · 420 ms".

What we measure

• Reachable: the address answered with a success or redirect status (HTTP 2xx or 3xx) within 15 seconds. We don't follow redirects, so each check is exactly one request.
• Response time: milliseconds until the first response headers arrived, measured from Cloudflare's network (the location varies), so treat it as rough.
• The last 90 days: days checked, % reachable and the median response time, in checks.json and the MCP provider_health tool.

What we don't measure

• For 14 of 17 providers we check a public page (their status page if they have one, else their docs, else their homepage). A page that loads doesn't prove the API works, and we never call an API that needs a key.
• For 3 we call a keyless endpoint their docs allow without sign-up: OpenLigaDB, StatsBomb Open Data (Hudl), TheSportsDB. That is a real request, but one endpoint, once a day — not uptime.
• Not data freshness, accuracy, rate limits or the speed of paid plans.

For maintainers: each morning we also run node scripts/mcp-smoke.mjs, which calls every MCP tool against the live server and fails on any error.

Politeness

Our requests identify themselves as FootballAPIsBot/1.0 (+https://footballapis.com/about/). We read each site's robots.txt (at most once a week) and skip any address it disallows, showing "skipped (robots.txt)". We read at most the first 64 KB of each answer.

What's in a provider record

Field names are stable within /api/v1/; anything that breaks them gets a new version. Every file also carries apiVersion, generatedAt, dataAsOf, licence and disclaimer. "unknown" always means "not confirmed on public pages", never "no".

changes.json: {date, type, provider, summary, sources, url}, where type is price_correction, price_confirmed, source_changed or site and provider is a slug or null. checks.json: {generatedAt, schedule, label, userAgent, latest, last90}; each result is {slug, target, kind, checkedAt, status, ok, ms} plus error or skipped when relevant.

Older download links

The original files still work and are updated with every build: /data/providers.json (our raw data file, snake_case fields) and /data/providers.csv (one row per provider, for spreadsheets). For new projects use /api/v1/.