Fetch video transcripts and metadata, extract structured data from video, search YouTube, and scrape or crawl the web through Supadata.
Supadata ships in the w6w first-party pack. It declares 19 actions, 3 health checks, and the host runs its code in a sandbox that never sees the credential.
io.w6w.supadataSupadata turns videos and web pages into text. This app exposes its REST API to a workflow, authenticated with an API key.
A workflow can fetch the transcript of a YouTube, TikTok, Instagram, X (Twitter) or Facebook video, or of a public media file, falling back to AI transcription when no native transcript exists. It can read unified metadata for any of those posts, ask AI to extract structured data from a video with a prompt or JSON Schema, scrape a web page to Markdown, map a site’s URLs, and crawl a whole site. For YouTube it can also search, read channels and playlists and their video ids, translate a transcript, and run batch jobs over many videos. Long-running work (generated transcripts, extraction, crawls and batches) is asynchronous: each has a start action that returns a job id and a matching action that polls it.
Every call spends credits from the account’s monthly allowance; the Get Account action reads the plan and the credits used.
Three routes to the same 19 actions. The Workflow tab is generated from Supadata's own manifest and carries its real ids, so it is copy-pasteable; the Code and CLI examples are the same call for any action on any app, so every app-specific value in them is a blank you fill in.
account-get Read the organization id, plan and credits used in the current billing period. Free, and exempt from the rate limit.
extract-job-get Poll a video-extraction job. `status` is `queued`, `active`, `completed` or `failed`; the extracted `data` is present once completed. Polling is free.
extract-start Use AI to extract structured data from a video (YouTube, TikTok, Instagram, X, Facebook or a media file URL). Give a prompt, a JSON Schema, or both. Returns a job id to poll with Get Extraction Job. 5 credits per started minute, minimum 5.
metadata-get Get unified metadata (title, author, stats, media, tags, publish time) for a YouTube, TikTok, Instagram, X (Twitter) or Facebook post. 1 credit.
transcript-get Get the transcript of a YouTube, TikTok, Instagram, X (Twitter) or Facebook video, or of a public media file URL. Answers with the transcript, or — when it must be generated and is too long to wait for — with a job id to poll with Get Transcript Job (`pending` is true).
transcript-job-get Poll an asynchronous transcript job started by Get Transcript. `status` is `queued`, `active`, `completed` or `failed`; the transcript is present once completed. Polling is free.
web-crawl-get Get a crawl's status (`scraping`, `completed`, `failed`, `cancelled`) and a page of results. When `next` is returned, call again with Skip set to the `skip` value in that URL.
web-crawl-start Start an asynchronous crawl that extracts the content of every page on a website. Returns a job id to poll with Get Crawl. 1 credit per crawled page.
youtube-batch-get Get the status and a page of results of a transcript or video-metadata batch job. Each result carries a `videoId`, its `transcript` or `video`, and an `errorCode` when that video failed. Polling is free.
youtube-channel-get Get a YouTube channel's name, handle, subscriber, video and view counts, and images.
youtube-channel-videos List the video, Short and live-stream ids of a YouTube channel.
youtube-playlist-get Get a YouTube playlist's title, video and view counts, owner and last update.
youtube-playlist-videos List the video, Short and live-stream ids of a YouTube playlist.
youtube-transcript-batch-start Create a batch job that fetches transcripts for a list of videos, a playlist or a channel. Returns a job id to poll with Get YouTube Batch. 1 credit per video, charged on submission.
youtube-transcript-translate Translate a YouTube video's transcript into another language.
youtube-video-batch-start Create a batch job that fetches metadata for a list of videos, a playlist or a channel. Returns a job id to poll with Get YouTube Batch.
A workflow step names the app and the action, and the editor fills in the
connection when you pick one. This is the Step shape from the
workflow spec, carrying Supadata's real ids.
{
"manifestVersion": "2",
"name": "supadata-example",
"steps": [
{
"id": "extract-job-get",
"uses": {
"app": "io.w6w.supadata",
"action": "extract-job-get",
"connection": "conn_YOUR_CONNECTION_ID"
},
"with": {
"jobId": "<jobId>"
}
}
]
}extract-job-get account-get extract-start metadata-get transcript-get +14 more actions available
Every app-specific value here is a blank you have to fill in. An
app action is reached through the connection that authenticates it, so the
address is a connection id, not the app id — and connections belong to your account,
so a public page cannot know yours. Create one for Supadata, then fill in
the three blanks: conn_YOUR_CONNECTION_ID, the action key, and the
parameters that action declares. The call itself is real — the shape is transcribed
from the studio's own snippet builder, which prints the same kind of blanks — but
nothing in it is specific to Supadata. The Workflow tab is where this app's
real ids are.
npm install @w6w/sdkyarn add @w6w/sdkpnpm add @w6w/sdkdeno add npm:@w6w/sdkimport { W6wClient, isActionRun } from "@w6w/sdk";
// Reads W6W_BASE_URL and W6W_TOKEN from the environment when omitted.
const client = new W6wClient();
const envelope = await client.run({
urn: "conn_YOUR_CONNECTION_ID",
action: "extract-job-get",
payload: {
jobId: "<value>",
},
});
if (isActionRun(envelope)) console.log(envelope.value); npm install -g @w6w/cli w6w run conn_YOUR_CONNECTION_ID --action extract-job-get --payload '{"jobId":"<value>"}' Give an AI agent Supadata — without giving it Supadata's credentials. One MCP endpoint exposes every app, function and workflow the caller is entitled to, as tools it can discover and run. Access is granted per team while we onboard.
One tool call{
"name": "w6w_invoke",
"arguments": {
"ref": "app:io.w6w.supadata#extract-job-get",
"input": {
"jobId": "<jobId>"
}
}
}
Every tool names its target with a single ref. The
app: form above doesn't name a connection at all — the
host resolves which of the caller's Supadata connections to sign
with, and refuses rather than guesses when the answer is ambiguous.
The token is attached host-side, at the moment of the call. It is never a tool argument, never in the model's context, and never in a transcript — so a prompt injection has nothing to exfiltrate.
Tools are derived per end user from what that person has actually connected and is entitled to — not one shared bot identity carrying the union of everyone's access.
Multi-step work runs on the workflow engine and returns a run handle the agent can poll — retries, branching and state survive the conversation that started them.
Supadata's declared health checks are on the surface too, so an agent can tell "the vendor is down" from "your credential expired" before it burns a retry on either.
The MCP surface is part of the hosted platform. Supadata itself is MIT, and the runtime that executes it is source-available (FSL).
Supadata declares its own checks, so its health is a property of the app rather than something the host guesses at.
The `Transcript API` resource on status.supadata.ai (Better Stack). The dashboard resource is shown but does not drive the verdict.
Unauthenticated GET /v1/me. Supadata's own JSON `unauthorized` refusal passes — it proves the application itself is answering.
Credits used against the plan's allowance for the current billing period.