Discover new creators on TikTok, Instagram or YouTube from one seed: a creator handle (walks who they follow on TikTok, who they collaborate with, tag or mention on Instagram), a hashtag, or a search phrase. Every account surfaced is screened live with one provider call (recent posts: how many reached 5k+ views, posting activity, newest post) and judged for topic relevance from its bio and captions, then returned ranked with recommended = qualified and on-topic. Accounts already in the database come back flagged known with their profileId. Nothing is crawled or added automatically: call crawl_profile for the ones you want. Priced up to 25 credits per call: 5 base plus 1 per account actually screened, the rest refunded (nothing found = fully refunded). Pass topic to say what related means for your use case. Not for looking up a known creator (use search_profiles).
Every creator database is a snapshot of what someone already decided to track. This skill looks outward: from one seed creator, hashtag or search phrase it surfaces accounts on the same platform, checks each one live for real reach and recent activity, judges whether it is actually about your topic, and hands back a ranked shortlist. It is the discovery step that turns "who else is like this?" into handles you can add with crawl_profile, and it runs synchronously in one call.
Connect the Viral Outliers MCP server once and the find_related_social_media_profiles tool becomes available to your agent alongside all other skills:
claude mcp add --transport http viral-outliers https://viraloutliers.com/api/mcp \
--header "Authorization: Bearer so_live_YOUR_KEY"In Claude.ai or ChatGPT, add a custom connector pointing at https://viraloutliers.com/api/mcp. Your agent then calls find_related_social_media_profiles on its own whenever the task needs it.
| Name | Type | Description |
|---|---|---|
platform* | string | "tiktok" | "instagram" | "youtube". |
handle | string | Seed creator. TikTok: the accounts they follow. Instagram: collaborators, tagged accounts and mentions in their recent posts. Not available on YouTube. Exactly one of handle, query, hashtag. |
query | string | Search phrase; the creators behind the top results become candidates. |
hashtag | string | Hashtag without the #; creators posting under it become candidates. |
topic | string | What "related" means for the relevance judge (e.g. "home workouts for women"). Defaults to the seed itself. |
limit | number | How many candidates to screen and return. Default 10, max 20. |
A list of candidate accounts for the platform you asked about. Each carries the handle and profile URL, display name, follower count and bio, a screening block (posts sampled, posts in the last 90 days, posts over 5k views, best and median views, newest post date), a qualified flag, a relevance verdict with confidence and a one-line reason, and a recommended flag that is true only when the account both qualifies on reach and is judged on-topic. Accounts already in the database are flagged known with their profileId so you can query them right away instead of crawling again.
The qualification rule is deliberately about reach, not size: at least one recent organic post at 5k+ views (pinned posts are ignored), at least five posts in the last 90 days, and a post within the last 45 days. Followers do not matter. That surfaces small accounts whose content works, which is exactly what outlier research is looking for. A call is priced at up to 25 credits (1 credit = $0.01): 5 for the seed lookup and the relevance judgement plus 1 per account actually screened, so a seed that only yields 8 screenable accounts costs 13 and the rest is refunded on the same request; a seed that yields nothing is fully refunded. The response carries a billing block that spells out the charge.
You pass a platform and one seed. A handle seed walks the creator graph: on TikTok the accounts that creator follows, on Instagram the accounts they collaborate with, tag or mention in their recent posts (Instagram no longer exposes a similar-accounts list to anyone, so this is the real signal that remains). A hashtag seed takes the authors of recent posts under the tag; a query seed takes the creators behind the top search results. YouTube supports query and hashtag seeds only.
Every account surfaced gets one screening call for its most recent posts, which is where the reach and activity numbers come from, and the whole batch is then judged for relevance in one model call from bios and captions against your topic. Results are ranked recommended first, then qualified, then known. A creator that surfaces on a later call within three days is not screened again. Nothing is added to the database by this skill: you choose, then call crawl_profile.
Building a watchlist for a niche you are new to: seed with the one creator you know, take the recommended handles, crawl them, then run outlier searches across the set. Finding lookalike creators for a brand: seed with a competitor's ambassador and a topic line describing the product, and use the relevance reasons to explain each pick. Widening a thin category: seed with the hashtags that category actually uses and let the screening throw out dormant and off-topic accounts before you spend a crawl on them.
Because the answer includes both the fresh finds and the accounts you already track, one call also tells you how much of a niche you have covered.
Doing this by hand means hitting a platform search or a following list, then opening every profile to see whether anyone actually watches it and whether it is about your topic. That is dozens of requests, most of them wasted on brands, dormant accounts and one-hit wonders whose only big number is a pinned post.
Here one authenticated call does the fan-out, the screening and the relevance judgement, prices it flat, and returns a list you can act on with a single follow-up skill. It is also the only discovery surface wired into the same database the outlier scores live in: known accounts are flagged, new ones become searchable minutes after you crawl them.
25 credits ($0.25) per call in prepaid credits (1 credit = $0.01). For example, 4 calls per dollar. Subscriptions include monthly credits; top-up packs start at $15. Failed asynchronous jobs are refunded automatically, and calls stop at a zero balance, never a surprise bill. See the full pricing table.
Call find_related_social_media_profiles with the platform and that creator's handle, plus a topic line if "similar" means something specific. You get back screened, relevance-ranked accounts; add the recommended ones with crawl_profile.
They cleared the reach and activity rule but the relevance judge did not see your topic in their bio and captions. Read the reason field; a niche account with a vague bio can land here.
No. It returns candidates only. Accounts already tracked are flagged known with a profileId; for the rest you decide, then call crawl_profile.
The list price of 25 credits is taken up front and the unused part is refunded on the same request: 5 credits base plus 1 per account screened. Nothing found means a full refund. The billing block in the response states the final charge.
YouTube does not expose who a channel follows or collaborates with. Use a search phrase or a hashtag; the creators behind the most-viewed recent shorts become the candidates.
Also available as a REST API endpoint.
Related topics: creator discovery API · similar creators finder · influencer lookalike search · niche creator research · competitor creator mapping