YouTube API — ReplyNodes

Quickstart

# Search videos
curl -X POST $BASE/v1/youtube/fetch \
  -H "Authorization: Bearer $REPLYNODES_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"operation":"search","params":{"term":"MCP servers","limit":5}}'

# Video details + transcript
curl -X POST $BASE/v1/youtube/fetch \
  -H "Authorization: Bearer $REPLYNODES_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"operation":"transcript","params":{"id":"dQw4w9WgXcQ"}}'

Operations

Operation	Params	Returns
search	`term, language?, limit?`	Ranked public videos for a query term.
video	`id, language?`	Full video record: title, channel, duration, view/like counts, description.
channel	`id, language?`	Channel profile with metadata.
comments	`id, limit?`	Comment thread for one video.
playlist	`id, language?`	Playlist contents in order.
related	`id, language?`	Videos related to one video.
transcript	`id, language?`	Caption/transcript text — the tool agents use most for summarization.

Data shape

Entities use yt:-prefixed URNs (yt:channel:…, yt:comment:…). Transcripts return structured segments so agents can cite timestamps, not blobs of text.