Hand any public post URL to one endpoint and get its comments back, replies nested, pagination handled. The same Comment and Author schema applies on every platform, so you parse the response once.
Any post URL, the whole tree
A comment scraping API returns the comments and nested replies on a public post as structured data. SocialCrawl accepts a post URL from any of 46 platforms and returns the whole tree on one shared schema, including the reply-to-a-reply layer where most of the substantive opinion sits, with no per-platform OAuth app to register.
Every one of these is a shipped pattern, not a roadmap item. The endpoint chain under each card is the actual call sequence.
Paste a link from any supported network and get the comments back with replies nested and pagination run to completion server side. Most teams start here because it deletes the per-platform comment code entirely.
prism/comments (any platform URL)Whole discussion threads including the deep replies, at a flat 5 credits per page, sorted by top, new or controversial. There is no Reddit OAuth app to register and no token to rotate.
reddit/post/comments (full tree)List a video's top level comments, then expand each thread and walk its replies to exhaustion with cursors. This is the deepest comment workload on the API and the pagination is built for it.
tiktok/post/comments -> video/comment/repliesPull a video's comment threads, then the replies underneath the ones worth reading. Both sides land on the same schema as every other network.
youtube/video/comments -> video/comment/repliesCursor pagination on a post you already care about, with author records attached to every comment so you can group by commenter without a second call.
instagram/post/commentsRe-check up to 25 comments you already know in a single call and get their current metrics back, without paginating the thread they live in. Items that fail are refunded.
prism/comment-lookup (track one comment)Four calls cover the whole job. Start with the composite, then drop to the raw endpoints when you want control over sort order and depth.
Reads are priced in credits. Metered endpoints bill for what they actually return, and failed upstream calls are refunded automatically.
POST /v1/prism/commentsSend one URL from any supported platform. Replies come back nested and the endpoint paginates to completion rather than handing you a cursor.
metered, from 2 creditsGET /v1/reddit/post/commentsSort by top, new or controversial and page through the tree with cursors. Long form opinion concentrates here more than anywhere else.
5 credits per pageGET /v1/tiktok/post/commentsPage through a video's top level comments first, so you can decide which threads deserve the reply walk.
1 credit per pageGET /v1/tiktok/video/comment/repliesExpand a comment id into its reply thread and keep paginating until the thread is exhausted. Repeat per thread you kept.
1 credit per pageEach of these is a live lane on the same key. Follow a link to see that platform's full endpoint surface.
Most teams either buy a listening seat that shows top level comments or run one scraper per network. Here is what changes.
| Capability | SocialCrawl | Typical setup |
|---|---|---|
| Reply depth | Nested replies walked to the bottom of the thread, on every platform that has them. | Top level comments only, or replies sold as a separate product. |
| Coverage | One key returns comments across 46 platforms and 375 endpoints. | One scraper per platform, each with its own auth and its own comment shape. |
| Authentication | A single SocialCrawl API key covers every network, Reddit included. | Per-platform OAuth apps, review queues and tokens to rotate. |
| Response shape | One Comment and Author schema everywhere, so parsing is written once. | A different JSON shape per vendor, normalised by hand. |
| Tracking one comment | Re-check up to 25 known comments in one call and get current metrics back. | Re-paginate the whole thread to find the comment again. |
| Pricing model | Credit packs that never expire, charged per call, with refunds when an upstream fails. | Monthly subscriptions or per-record contracts with minimums. |
A comment scraping API returns the public comments on a post as structured data, with the author, the text, the timestamp and the engagement on each one. Instead of rendering a thread in a browser you get records you can store, filter and score.
Yes. Replies are the point. The composite endpoint returns replies nested under their parent, and the raw endpoints expose a dedicated replies call you can walk with cursors until the thread is exhausted. Most of the argument in a thread lives below the first level.
SocialCrawl covers 46 platforms and 375 endpoints. Comment surfaces include Reddit, TikTok, YouTube, Instagram, Facebook, LinkedIn and Hacker News, plus one composite endpoint that accepts a URL from any of them. One API key covers all of it.
No. You authenticate once with a SocialCrawl API key. There is no Reddit OAuth app to register, no review process per network and no separate credentials to rotate when a platform changes its rules.
Page through with cursors and keep the position you reached. Raw list endpoints bill per page, so a very deep thread costs what its depth costs, and the composite endpoint is metered against what it actually returns. Bulk paginated extraction into local storage is the normal production pattern.
Yes. Pass a comment URL or id and get that comment and its current metrics back directly. You can also re-check up to 25 known comments in a single call, which makes watching a brand's own comment on a large post cheap rather than a full re-pagination.
Yes, for transformed views shown to your own users. Our terms permit dashboards and derived products, and restrict raw passthrough or bulk redistribution. If you are unsure where your design sits, ask us before you build.
Ask AI about SocialCrawl
100 free credits on signup. No card, no per-platform OAuth, no seat licence.
curl -X POST "https://socialcrawl.dev/v1/prism/comments" \
-H "x-api-key: $SOCIALCRAWL_API_KEY" \
-H "content-type: application/json" \
-d '{"url":"https://www.reddit.com/r/SaaS/comments/abc123/","max":200,"replies":true}'curl -G "https://socialcrawl.dev/v1/reddit/post/comments" \
-H "x-api-key: $SOCIALCRAWL_API_KEY" \
--data-urlencode "url=https://www.reddit.com/r/SaaS/comments/abc123/" \
--data-urlencode "sort=top"Both calls run against live data. Swap the key and the post URL and they work as written.