Smart run

Name the job, not the tool. Send capability plus a small canonical input to POST /v1/run and Cracked picks the best live endpoint for it, translates your input into that endpoint's schema, runs it and falls through to the next candidate if the upstream fails. Same auth, same billing, same run record as a direct run.

curl -s https://cracked-ai-pearl.vercel.app/v1/run -H "Authorization: Bearer ck_live_..." -H "content-type: application/json" \
  -d '{"capability":"instagram-profile","input":{"username":"nike"}}'

Why

No schema hunting. Every capability has a few well-named fields (username, query, url). You never read an actor's input schema.
Automatic fallback. If the first tool is blocked, times out or returns a 5xx, 408 or 429, the next candidate runs. Up to three attempts per call.
Live ranking. Candidates are ordered by verified status, recent health (healthy, stable, unknown, degraded; outages are skipped) and price. Keyless public endpoints are kept as last resort.
Transparent. The response tells you exactly which endpoint served it (routedTo) and what was tried (attempts).

Request

POST /v1/run
{
  "capability": "web-search",     // or "query": "instagram followers" (fuzzy match on name/tags)
  "input": { "query": "peptide therapy austin", "limit": 10 },
  "wait": true,                    // false = 202 + runId for long scrapers, poll GET /v1/runs/{runId}
  "webhook_url": "https://..."     // optional
}

When provider and endpoint are present the request is a normal direct run and capability is ignored. Unknown capability ids return 404 CAPABILITY_NOT_FOUND with the five closest ids in the message.

Response

A standard run record plus three fields:

{ "runId": "…", "status": "COMPLETED", "output": [ ... ],
  "provider": "apify", "endpoint": "/apify/instagram-profile-scraper",
  "providerResponse": { "httpStatus": 200, "error": null },
  "billing": { "units": 1, "providerUsd": 0.0033, "platformFeeUsd": 0.001, "totalUsd": 0.0043, "byok": false },
  "capability": "instagram-profile",
  "routedTo": { "provider": "apify", "endpoint": "/apify/instagram-profile-scraper" },
  "attempts": [
    { "provider": "apify", "endpoint": "/abe/tiktok-profile-scraper", "status": "FAILED", "httpStatus": 502, "runId": "…", "error": "Actor run failed" },
    { "provider": "apify", "endpoint": "/apify/instagram-profile-scraper", "status": "COMPLETED", "httpStatus": 200, "runId": "…" }
  ],
  "hints": { "capability": "GET /v1/capabilities?id=instagram-profile" } }

If every candidate fails you get the last run record (status FAILED, BLOCKED or TIMED_OUT) with the full attempts list. Failed attempts are recorded as runs but are not billed. With wait: false the first candidate that accepts the job (HTTP 202, status RUNNING) is treated as success; poll it with GET /v1/runs/{runId}?wait=30.

Listing capabilities

GET /v1/capabilities is public and returns every capability with its input schema and the candidates ranked for this deployment, including which are live right now. GET /v1/capabilities?id=web-search returns one. In MCP the same surface is list_capabilities and run_capability.

curl -s https://cracked-ai-pearl.vercel.app/v1/capabilities?id=web-search

Capabilities

36 capabilities. Required fields are marked. Candidates are listed in curated order; live ranking may differ.

Capability	Input	Candidates
`web-search` Search the web and get ranked results with titles, URLs and snippets. {"query":"peptide therapy"}	`query`* string — Search query `limit` number = 10 — Max results	`serper/search` `tavily/search` `brave/web` `exa/search` `apify/apify/google-search-scraper` `apify/apify/rag-web-browser` — also fetches page content `duckduckgo/instant-answer` (last resort) — instant answer only, not a full SERP
`web-scrape` Fetch one URL and return its readable content (text or markdown) plus title and metadata. {"url":"https://example.com"}	`url`* string — Absolute http(s) URL `max_chars` number = 20000 — Truncate content to this many characters	`firecrawl/scrape` `jina/read` `web/fetch` `brightdata/unlock` — residential proxy for anti-bot sites `apify/apify/web-fetch`
`web-crawl-map` List the URLs of a website (sitemap or crawl map) so you can pick pages to scrape. {"url":"https://example.com"}	`url`* string — Site root URL `limit` number = 200 — Max URLs	`firecrawl/map` `web/sitemap` `apify/apify/website-content-crawler` — full crawl: returns page content, slower
`news-search` Recent news articles matching a query with source, date and URL. {"query":"peptide therapy"}	`query`* string — Query `limit` number = 20 — Max articles	`serper/news` `newsapi/everything` `brave/news` `apify/data_xplorer/google-news-scraper-fast` `gdelt/articles` (last resort)
`instagram-profile` Profile details for an Instagram account: followers, following, bio, verified, recent post count. {"username":"nike"}	`username`* string — Instagram username or profile URL	`apify/apify/instagram-profile-scraper` `apify/apify/instagram-scraper` `apify/coderx/instagram-profile-scraper-bio-posts`
`instagram-posts` Recent posts from an Instagram account with captions, likes, comments and media URLs. {"username":"nike"}	`username`* string — Instagram username or profile URL `limit` number = 20 — Max posts	`apify/apify/instagram-post-scraper` `apify/apify/instagram-scraper`
`tiktok-profile` Profile details for a TikTok account: followers, likes, bio, video count. {"username":"nike"}	`username`* string — TikTok username (with or without @) or profile URL	`apify/abe/tiktok-profile-scraper` `apify/clockworks/tiktok-profile-scraper` `apify/clockworks/tiktok-scraper`
`tiktok-videos` Recent videos from a TikTok account with views, likes, shares and captions. {"username":"nike"}	`username`* string — TikTok username or profile URL `limit` number = 20 — Max videos	`apify/clockworks/tiktok-profile-scraper` `apify/clockworks/tiktok-scraper` `apify/clockworks/free-tiktok-scraper`
`youtube-search` Search YouTube videos by keyword with title, channel, views and URL. {"query":"peptide therapy"}	`query`* string — Search query `limit` number = 10 — Max videos	`youtube/search` `apify/streamers/youtube-scraper`
`youtube-transcript` Transcript (captions) of a YouTube video as text. {"url":"https://example.com"}	`url`* string — YouTube video URL `language` string = en — Caption language code	`apify/pintostudio/youtube-transcript-scraper` `apify/starvibe/youtube-video-transcript` `apify/karamelo/youtube-transcripts` `apify/topaz_sharingan/Youtube-Transcript-Scraper-1`
`x-profile` Profile details for an X account: followers, following, bio, verified, tweet count. {"username":"nike"}	`username`* string — X handle (with or without @) or profile URL	`apify/apidojo/twitter-user-scraper` `apify/apidojo/tweet-scraper` — returns the latest post; author object carries profile stats
`x-search` Search recent posts on X by keyword, hashtag or advanced query. {"query":"peptide therapy"}	`query`* string — Search terms (X advanced search syntax allowed) `limit` number = 20 — Max posts	`apify/apidojo/tweet-scraper` `apify/apidojo/twitter-scraper-lite` `apify/kaitoeasyapi/twitter-x-data-tweet-scraper-pay-per-result-cheapest`
`reddit-search` Search Reddit posts by keyword, optionally inside one subreddit. {"query":"peptide therapy"}	`query`* string — Search query `subreddit` string — Optional subreddit name without r/ `limit` number = 25 — Max posts	`reddit/search` — Reddit public JSON, rate-limited `apify/trudax/reddit-scraper-lite` `apify/fatihtahta/reddit-scraper-search-fast`
`linkedin-profile` Public LinkedIn profile: headline, location, experience, education, skills. {"url":"https://example.com"}	`url`* string — LinkedIn profile URL or public identifier, e.g. satyanadella	`apify/harvestapi/linkedin-profile-scraper` `apify/apimaestro/linkedin-profile-detail` `apify/dev_fusion/Linkedin-Profile-Scraper` `pdl/person-enrich`
`linkedin-company` LinkedIn company page: industry, size, headquarters, description, follower count. {"url":"https://example.com"}	`url`* string — LinkedIn company URL or slug, e.g. anthropic	`apify/harvestapi/linkedin-company` `apify/apimaestro/linkedin-company-detail` `apify/automation-lab/linkedin-company-scraper`
`google-maps-places` Local businesses and places matching a query: name, address, rating, phone, website, coordinates. {"query":"peptide therapy"}	`query`* string — What to find, e.g. 'med spa' `location` string — Where, e.g. 'Austin, TX' (can also be part of query) `limit` number = 20 — Max places	`serper/places` `apify/compass/crawler-google-places` `apify/compass/google-maps-extractor` `apify/lukaskrivka/google-maps-with-contact-details` — adds emails and social profiles
`google-maps-reviews` Reviews for one Google Maps place: rating, text, author, date. {"place":"ChIJN1t_tDeuEmsRUsoyG83frY4"}	`place`* string — Google Maps place URL or place_id (ChIJ...) `limit` number = 50 — Max reviews	`apify/compass/google-maps-reviews-scraper` `apify/web_wanderer/google-reviews-scraper`
`amazon-product` Amazon product details or search results: title, price, rating, ASIN, images. {"query":"peptide therapy"}	`query`* string — Search term, ASIN, or Amazon product/search URL `limit` number = 10 — Max products	`apify/junglee/amazon-crawler` `apify/junglee/free-amazon-product-scraper`
`amazon-reviews` Customer reviews for one Amazon product: rating, title, text, date, verified purchase. {"product":"B00E9M4XEE"}	`product`* string — ASIN or Amazon product URL `limit` number = 50 — Max reviews	`apify/junglee/amazon-reviews-scraper` `apify/web_wanderer/amazon-reviews-extractor`
`company-enrich` Firmographics for a company by website domain: name, industry, size, location, description, socials. {"domain":"anthropic.com"}	`domain`* string — Company website domain, e.g. anthropic.com	`apollo/enrich-company` `pdl/company-enrich` `web/fetch` (last resort) — homepage title, description and Open Graph only
`email-find` Most likely work email for a person at a company domain. {"domain":"anthropic.com","first_name":"Jane","last_name":"Doe"}	`domain`* string — Company domain `first_name`* string — First name `last_name`* string — Last name	`hunter/email-finder` `apollo/enrich-person` `pdl/person-enrich`
`email-verify` Deliverability verdict for an email address. {"email":"jane@example.com"}	`email`* string — Email address	`hunter/email-verifier` `domain/email-check` (last resort) — syntax + MX heuristics, no SMTP probe
`geocode` Coordinates, country and timezone for a city, address or place name. {"query":"peptide therapy"}	`query`* string — Place name or address `limit` number = 5 — Max matches	`open-meteo/geocode` `openstreetmap/geocode`
`weather-forecast` Current conditions plus daily forecast for coordinates. Use geocode first for a city name. {"latitude":30.27,"longitude":-97.74}	`latitude`* number — Latitude `longitude`* number — Longitude `days` number = 3 — Forecast days 1-16 `units` string = fahrenheit — celsius or fahrenheit	`open-meteo/forecast`
`stock-quote` Latest price, change and volume for a stock ticker. {"symbol":"AAPL"}	`symbol`* string — Ticker symbol, e.g. AAPL	`alphavantage/quote` `polygon/aggregates` — last 7 daily bars `apify/architjn/yahoo-finance` — Yahoo Finance history
`crypto-price` Spot price of a cryptocurrency in a fiat currency with 24h change and market cap. {"coin":"bitcoin"}	`coin`* string — Coin id or symbol, e.g. bitcoin, ethereum, solana `currency` string = usd — Fiat currency	`coingecko/simple-price` `defillama/prices` — USD only
`domain-lookup` WHOIS/RDAP registration data for a domain: registrar, created and expiry dates, nameservers, status. {"domain":"anthropic.com"}	`domain`* string — Domain, e.g. example.com	`domain/rdap` `domain/dns` (last resort) — nameservers only
`github-repo` Repository metadata: description, stars, forks, language, license, default branch, topics. {"repo":"vercel/next.js"}	`repo`* string — owner/name or repository URL	`github/repo` `web/fetch` (last resort) — page text, not structured
`jobs-search` Job listings matching a title or keyword, optionally in a location. {"query":"peptide therapy"}	`query`* string — Job title or keywords `location` string — City, state or country `limit` number = 25 — Max listings	`apify/misceres/indeed-scraper` `apify/valig/indeed-jobs-scraper` `apify/curious_coder/linkedin-jobs-scraper` `apify/api-empire/linkedin-jobs-scraper` — LinkedIn jobs (formerly bebity/linkedin-jobs-scraper) `remotive/jobs` (last resort) — remote jobs only, ignores location
`real-estate-listings` Homes for sale in a city, neighbourhood or ZIP code: price, beds, baths, sqft, address, listing URL. {"location":"Austin, TX"}	`location`* string — City and state, e.g. 'Austin, TX', or a 5-digit ZIP `limit` number = 20 — Max listings	`apify/maxcopell/zillow-scraper` `apify/maxcopell/zillow-zip-search` — ZIP codes only
`image-generate` Create an image from a text prompt. Returns image URL(s) or base64 depending on the model. {"prompt":"A lighthouse at dawn, watercolor"}	`prompt`* string — Image prompt `size` string = landscape — square, landscape or portrait	`fal/flux-schnell` `openai/image` `kie/flux-kontext` `stability/sd3` `fal/flux-pro`
`video-generate` Create a short video clip from a text prompt. Asynchronous: send wait=false and poll the run. {"prompt":"A lighthouse at dawn, watercolor"}	`prompt`* string — Video prompt `aspect_ratio` string = 16:9 — 16:9 or 9:16	`kie/veo3-video` `minimax/video` `fal/run`
`text-to-speech` Convert text to spoken audio. Returns base64 MP3 or an audio URL. {"text":"Hello world"}	`text`* string — Text to speak `voice` string — Provider voice id (optional)	`elevenlabs/tts` `minimax/tts`
`transcribe-audio` Speech to text for a hosted audio or video file URL, with optional speaker labels. {"url":"https://example.com"}	`url`* string — Public audio/video URL `diarize` boolean = false — Speaker labels `language` string — Language code (optional)	`deepgram/transcribe-url`
`llm-chat` Generate text with a frontier or open model. Cheapest live model first. {"prompt":"A lighthouse at dawn, watercolor"}	`prompt`* string — User prompt `system` string — Optional system prompt `max_tokens` number = 1024 — Max output tokens `json` boolean = false — Ask for a JSON object response	`openai/chat` `anthropic/messages` `gemini/generate` `openrouter/chat` `groq/chat` `mistral/chat` `xai/chat` `huggingface/chat` `minimax/chat`
`translate` Translate text between languages. {"text":"Hello world","to":"es"}	`text`* string — Text to translate `to`* string — Target language code or name, e.g. es `from` string — Source language code (optional)	`mymemory/translate` — expects ISO codes; max ~500 chars per call `openai/chat` `anthropic/messages` `gemini/generate`

When to use a direct run instead

You need an input the canonical schema does not expose (date ranges, cookies, proxy country). Use discover, inspect and run with the provider and endpoint.
You want a specific tool for repeatability. routedTo from a smart run tells you which provider and endpoint to pin.

Examples

# fuzzy capability lookup
curl -s https://cracked-ai-pearl.vercel.app/v1/run -H "Authorization: Bearer $CRACKED_API_KEY" -H "content-type: application/json" \
  -d '{"query":"tiktok followers","input":{"username":"nike"}}'

# long scraper: start and poll
curl -s https://cracked-ai-pearl.vercel.app/v1/run -H "Authorization: Bearer $CRACKED_API_KEY" -H "content-type: application/json" \
  -d '{"capability":"google-maps-places","input":{"query":"med spa","location":"Austin, TX","limit":20},"wait":false}'
curl -s "https://cracked-ai-pearl.vercel.app/v1/runs/<runId>?wait=30" -H "Authorization: Bearer $CRACKED_API_KEY"

# keyless: weather, crypto, geocode
curl -s https://cracked-ai-pearl.vercel.app/v1/run -H "Authorization: Bearer $CRACKED_API_KEY" -H "content-type: application/json" \
  -d '{"capability":"crypto-price","input":{"coin":"bitcoin"}}'