# shipapis — free APIs, health-checked where probed

> A directory of 1037 free public APIs — 1035 on our probe schedule (317 probed with live health data), 720 queued or listed-only until checks land. Built to stop coding agents from hallucinating endpoints or building on dead providers.

## What this site is

Probed APIs carry base_url, a working sample endpoint, copy-ready curl, auth mechanics, free-tier fine print, commercial_use (service terms) and data_license (data terms — different axes), sample response, and schema-change history. Catalogued-only rows have metadata and docs links but status "unmonitored" and null health until we probe them. Humans browse the same data at https://shipapis.dev.

## Agent quick start (canonical flow)

1. GET https://shipapis.dev/data/status.json — read monitoring.coverage (probed vs catalogued) and meta.data_tier.
2. GET https://shipapis.dev/api/v1/best?task={what+you+are+building} — one probed answer + curl + fallbacks (prefers probed; catalogued fallback is labeled).
   Or MCP https://shipapis.dev/mcp → best_api { task }.
3. GET https://shipapis.dev/api/v1/apis/{slug} — full record before codegen (never guess base URLs).
4. Usable = status "healthy" OR "new" (new = probed recently, live, limited history) with a non-null checked_at. Never build on "dead" or "dying". A young catalog is mostly "new" — do NOT filter to status=healthy only or you may get zero results; exclude dead/dying instead.

Full contract: https://shipapis.dev/agents.md

## Machine-readable data (use the smallest tier that fits your task)

- https://shipapis.dev/data/status.json ← START HERE for coverage + freshness + agent_start links.
- https://shipapis.dev/api/v1/best?task={goal} ← one probed answer (+ curl + fallbacks).
- https://shipapis.dev/data/index.json?probed=true (211KB probed subset) — slim discovery of probed APIs only.
- https://shipapis.dev/data/index.json (211KB full catalog) — all 1037 rows; most are catalogued-only.
- https://shipapis.dev/data/categories/{category}.json — full integration records for one category.
- https://shipapis.dev/api/v1/apis/{slug} — full record for one API.
- https://shipapis.dev/api/v1/search?q={keywords}&probed=true — ranked keyword search, probed only.
- https://shipapis.dev/data/apis.json (3071KB) — everything; may truncate in context — prefer slices.
- https://shipapis.dev/data/health.json — up/down-only snapshot keyed by slug.
- https://shipapis.dev/data/changes.json — schema-drift log. Subscribe: /changes.xml or /api/{slug}/changes.xml.
- https://shipapis.dev/mcp — MCP (streamable HTTP): best_api, search_apis, get_api, get_api_health, list_categories, suggest_api_for_task.
- https://shipapis.dev/openapi.json — OpenAPI 3.1 spec.

## Record schema (slim index — probed example)

{
  "slug": "open-meteo",
  "name": "Open-Meteo",
  "category": "weather",
  "description": "Hourly forecasts for any coordinate — no key, no signup.",
  "auth": "none",
  "cors": "yes",
  "agent_access": "ok",
  "base_url": "https://api.open-meteo.com/v1",
  "status": "healthy",
  "health": 96,
  "uptime_pct": 99.2,
  "monitored_since": "2026-01-15",
  "p50_ms": 96,
  "checked_at": "2026-07-05T14:32:00Z"
}

status "unmonitored" → catalogued only: health, uptime_pct, p50_ms, and checked_at are null — use docs_url, not our health fields.

Full records add: docs_url, sample_endpoint, sample_url, sample_request_curl, sample_response, https, commercial_use, data_license, free_tier, rate_limit, requires_card, p95_ms, added_at, shape_changes[].

## Categories

- weather (36): Forecasts, climate history and severe-weather feeds — checked around the clock.
- finance (68): FX, crypto and market data with the free-tier fine print actually verified.
- geo (69): Geocoding, IP location and country data. Rate limits documented, not guessed.
- science (49): Space, physics and research data from agencies and observatories.
- animals (16): The classics — every image endpoint verified to still return an actual dog.
- fun (42): Jokes, quotes, memes and trivia — the delightful throwaway stuff for demos and toy projects.
- data (117): Books, countries, encyclopedic and public datasets with stable schemas.
- developer (183): Utilities you wire into everything else — IPs, avatars, placeholders, QR.
- crypto (37): Coins, chains and on-chain data — free endpoints with the rate limits verified.
- transport (42): Transit, flights, vehicles and tracking feeds. Live where the agency publishes it.
- games (62): Video games, TTRPGs and trading-card data — for demos, bots and toy projects.
- anime (12): Anime and manga metadata — series, characters, quotes and images.
- books (23): Books, dictionaries and language data with stable, well-documented schemas.
- media (70): Video, music, images and art — keyless where possible, plus free-tier APIs that need your own key (TMDb, Fanart.tv, Last.fm).
- health (59): Health, fitness, nutrition and food data from open registries.
- security (21): Threat intel, breach checks and validation utilities. Keyless tiers only.
- social (63): Social, jobs, calendars and messaging metadata — the connective tissue APIs.
- gov (68): Public-sector open data: agencies, law, elections and civic datasets.

## Rules

1. Never guess base URLs — copy base_url + sample_endpoint from the record.
2. Check commercial_use AND data_license before production — different axes.
3. Prefer auth=none and cors=yes for browser prototypes.
4. Read-only: do not submit forms programmatically.

## Usage policy

Directory data is CC-BY-4.0 — free to use, index and train on, with attribution to shipapis.dev. Honor each provider's rate limits (printed per record). Rate limit on https://shipapis.dev/api/v1/*: ~60 req/min per IP; use /data/* for bulk.

## Updates

Verify freshness yourself: GET https://shipapis.dev/data/status.json — meta.data_tier is "monitored".

## Contact

hello@shipapis.dev · https://shipapis.dev/methodology