YouTube comment scraper, as an API
Paste a video, a Short, or a channel. Get the comment section back as JSON, every comment linked to its spot under the video.
Runs in your BuzzSearch account. Free to sign up, and the same read is one API call or one MCP tool call.
Comments we read under one source
One video, read in full. These are eight of the 150 comments we captured under it, as the API hands them back.
Every card is a real comment, quoted as written. This is what the API returns, before the quotes are tagged.
One call from your code. One tool call from your agent.
The search endpoint takes a link the same way it takes a question. Poll once with include=comments and the whole section comes back as JSON.
curl -X POST https://buzzsearch.ai/api/v1/search \
-H "Authorization: Bearer $BUZZSEARCH_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "query": "Read the comments on this YouTube video: https://www.youtube.com/watch?v=dQw4w9WgXcQ", "depth": "quick" }'
# 202 { "id": "srch_...", "status": "queued", ... }
curl https://buzzsearch.ai/api/v1/search/srch_...?include=comments \
-H "Authorization: Bearer $BUZZSEARCH_API_KEY"# any MCP client: Claude Code, Claude.ai, ChatGPT, Cursor, Codex
claude mcp add --transport http buzzsearch https://buzzsearch.ai/api/mcp
> Read the comments on this YouTube video: https://www.youtube.com/watch?v=dQw4w9WgXcQ
● buzzsearch - search (MCP)(query: "Read the comments on this YouTube video: https://www.youtube.com/watch?v=dQw4w9WgXcQ")
● buzzsearch - get_comments (MCP)(search_id: "srch_...")
● buzzsearch - get_quotes (MCP)(dimensions: ["emotional_painpoints"])What comes back
Not a CSV of rows. The comments, the quotes already tagged by what they mean, the sources with their counts, and a cited answer.
{
"id": "srch_...", "status": "completed", "credits_charged": 45,
"answer": { "markdown": "...", "citations": [{ "n": 1, "quote_id": "...", "source_url": "..." }] },
"coverage": { "threads": 1, "comments": 112, "sources": ["tiktok"] },
"quotes": [{ "quote": "...", "author": "...", "url": "...",
"dimension": "emotional_painpoints", "score": 53, "emotional_intensity": 4 }],
"sources_read": [{ "source": "tiktok", "title": "...", "url": "...", "comment_count": 112, "comments_read": 112 }],
"comments": [{ "source": "tiktok", "url": "...", "body": "...",
"comments": [{ "author": "...", "score": 53, "depth": 1, "body": "..." }] }],
"comment_count": 112
}comments[]
Every comment read, with author, like count, reply depth, and a per-comment URL (YouTube addresses single comments, so each quote cites the exact one).
quotes[]
The lines worth keeping, tagged pain point, failed solution, objection, desired outcome, or lingo, with a score and an intensity.
sources_read[]
Each video with its view count, channel, and how many comments were read.
answer.markdown
A cited read of the section, every claim pointing at a quote id.
coverage
Videos, comments, and sources in the run, so you can judge the read before you trust it.
credits_charged
The exact cost, reported after the run. Rereading a finished search is free.
What gets read
A watch link or a Short
The same video id either way. Replies are paged, not just top-level comments.
A channel link
The channel's recent videos, then the comments under each one.
A question
Search first, then read: the videos and Shorts that rank for the query, with their comment sections.
The description
Quotable as its own line, like a Reddit post body.
Comments turned off
Reported as comments_disabled on the source, never retried, never charged twice.
No API quota
No YouTube Data API key and no daily quota. The read is a page read.
More comment APIs
- TikTok comment scraperVideos, creators, and hashtags, as JSON.
- Reddit comment scraperA subreddit or a thread, as JSON.
Questions
Is there a free YouTube comment scraper here?
Reads run inside a BuzzSearch account. Signing up is free, and the input above drops your link into the chat, where the first read is the same one the API makes.
Does it work on YouTube Shorts?
Yes. A Shorts link and a watch link resolve to the same video id, and each comment links back to its spot under the video.
How many comments does one read cover?
Depth sets it: quick reads the top of the section, deep and exhaustive page further. The response reports comments_read per video.
Can I search YouTube comments?
Ask a question instead of pasting a link and BuzzSearch finds the videos first, then reads their comments. You can also filter a finished read's quotes by phrase with get_quotes. AI market research
Why would I read YouTube comments for research?
Objections under a competitor's review, questions under a how-to, and the exact phrases viewers use when they ask for a product. That language goes straight into copy. UGC script generator
How is this billed?
Credits from your BuzzSearch balance, charged after the run and reported as credits_charged. Reading a finished search, its quotes, its sources, and its comments costs nothing. See pricing