How authentication works

An agent authenticates the same way a script does: every request carries an Authorization: Bearer YOUR_API_KEY header. The key is created automatically the first time an account buys credits, and can be regenerated from the dashboard at any time. There's no OAuth flow, no session cookies, and no per-agent registration — one key is enough for all three services.

Example: an agent fetching a video, then its transcript

A common agent workflow — summarize a video it was given a link to — looks like two calls: fetch the file, then pull the transcript for the same URL.

curl -s "https://verinio.com/api/video-info?url=SOURCE_URL" \
  -H "Authorization: Bearer YOUR_API_KEY"

curl -s "https://verinio.com/api/transcript?url=SOURCE_URL&format=txt" \
  -H "Authorization: Bearer YOUR_API_KEY"

Both calls return JSON (or, for downloads, the file itself) so the agent can parse the result and decide its next step — no HTML scraping or headless browser required.

Designing an agent workflow around credits

Because every call has a fixed or size-based credit cost (see services overview and current pricing via GET /api/credit-config), an agent can estimate a job's cost before running it: check /api/video-info first to see the file size, decide whether the download is worth it, then call /api/download only if so. Failed requests are never charged, so retry logic doesn't waste credits. An action priced at 0 credits works without an API key at all; anything above 0 needs one.

Original vs. clean downloads

/api/download returns the source file exactly as provided by default. Add &clean=1 to instead get a re-encoded copy with identifying metadata and chapters stripped - same audio/video quality, but safe to re-upload, share, or hand to another tool without carrying the original source's embedded metadata. If your agent is going to re-publish or forward the file anywhere, prefer clean=1; for one-off local processing, the default (original) is fine and slightly faster since it skips the re-encode step.

Where this fits in a larger pipeline

Verinio is typically one step in a larger agent or automation pipeline — n8n, a LangChain/agent-SDK tool call, or a cron job — rather than the whole pipeline. Because responses are plain JSON and files are returned directly (not wrapped in a proprietary SDK), Verinio works as a tool call for any agent framework that can make an HTTP request with a header.