Social Analyzer - Find Social Profiles by Username
Type a username and get every public profile that uses it. Every run checks all 999 sites the scanner knows: social networks, gaming, dating, developer, forums, wikis and niche platforms. Every hit is ranked high, medium or low confidence with the page's title, language and meta tags, so real profiles come first and guesses stay marked. No login, no cookies, no API key.
Available as an Apify Actor. $0.005 per profile found. One username on all 999 sites takes about 3 minutes; on the 100 most popular sites about 30 seconds.
What does it do?
You give it one username, or a list, and nothing else is needed: with only the username set, the scan is the widest possible, all 999 sites, worldwide, every category, every match kept and labelled by confidence, page details and patterns extracted, one hour to do it. It checks those sites for a profile with exactly that handle and returns one row per profile it finds: the platform, the profile URL, how confident the match is, the site's category and country, the page title, its language and its meta tags. A summary row and an OUTPUT record say in plain words what was checked, how many sites had no profile, how many did not answer, and what came back.
You can paste @elonmusk or a full profile link like twitter.com/elonmusk; the handle is taken out of it. If you paste an email address, the part before the @ is searched and the run points you to the email tool for the address itself. A full name like John Smith is searched as johnsmith, and the summary names john.smith and john_smith to try next.
How is it different from Sherlock?
The scanner underneath is qeeqbox/social-analyzer (3,000+ GitHub stars). This actor runs it at full depth on every run: all sites, all output fields, 60 parallel workers instead of its hardcoded 15, and its most permissive detection level. It also chooses the sites itself from the scanner's own list and fixes what the command line gets wrong in the current version: several usernames at once, category and country filters, a confidence filter that silently did nothing, and the slow and special modes that have no code behind them.
When should I use it?
- Recruiting and hiring: check a candidate's public presence from one handle
- People search: find every account a person uses under a username
- Fraud and trust teams: find a scammer's or seller's other profiles
- Dating safety and self-checks: see what a handle exposes, including on dating and adult sites
- Investigations and due diligence: build a cross-platform footprint for a case
What does it cost?
Typical runs:
A run that finds nothing costs nothing. A form with no usable username stops before anything is charged. If your spending limit cannot pay for one profile, nothing is searched or charged and the status message gives the limit and the price. If a run is cut off by the time limit or your spending limit, you keep and pay for only the rows delivered before the cut.
Which inputs does it take?
Filters combine. With a category or country filter, top picks the most popular sites inside that filter. With websites, top is ignored.
Nothing is held back. The scanner offers three modes; only fast has code behind it in the current version, and its internal "extreme" level is a stricter gate that returns fewer hits, not a deeper scan. So every run already uses the deepest configuration that exists: all sites, all fields, the permissive level. The only feature not exposed is profile screenshots, which needs a Chrome browser the image does not carry.
How do I follow a run?
The status line at the top of the run updates after every step: 48% done: 500 of 998 site checks, 143 profiles so far. 1 of 3 usernames finished, now checking kaytats. Rows appear in the dataset as each step (about 125 sites) finishes, so you can start reading results after about 30 seconds. The OUTPUT record carries status, percentDone and per-username progress while the run is going, and turns into the final summary when it ends.
What does the output look like?
Real rows from a local run on 2026-09-12 with input {"username": "[email protected]", "top": 50}. The address was recognised as an email, the handle frostkatrina was searched, 17 profiles came back in 37 seconds.
One profile row as JSON:
{
"recordType": "profile",
"username": "torvalds",
"platform": "GitHub",
"site": "github.com",
"link": "https://github.com/torvalds",
"confidence": "high",
"matchRate": 100.0,
"category": "developer_tech",
"categoryDetail": "Computers Electronics and Technology > Programming and Developer Software",
"country": "United States",
"adultSite": false,
"siteRank": 89,
"pageTitle": "torvalds (Linus Torvalds)",
"language": "English",
"metadata": [{"property": "og:title", "content": "torvalds - Overview"}, {"property": "og:type", "content": "profile"}],
"checkedAt": "2026-09-12T14:20:11+00:00"
}
The summary row (also saved as the OUTPUT record) from the same email run:
{
"recordType": "summary",
"usernames": ["frostkatrina"],
"usernamesChecked": 1,
"sitesPerUsername": 50,
"filters": {},
"profilesFound": 17,
"highConfidence": 2,
"mediumConfidence": 11,
"lowConfidence": 4,
"sitesNotFound": 29,
"sitesFailed": 4,
"durationSeconds": 36.5,
"notes": ["\"[email protected]\" is an email address. Searched the part before the @ (\"frostkatrina\") as a username. To find accounts registered with the email itself, use https://apify.com/anshumanatrey/holehe-email-osint"],
"message": "Found 17 profiles for frostkatrina across 50 sites: 2 high, 11 medium, 4 low confidence. Strongest first. ..."
}
Confidence. High means a 100% match: the site answered exactly the way it does for a real profile. Medium (50 to 99%) is likely but worth a look; many adult and dating sites answer at 50% for any handle. Low is a guess. Rows are sorted strongest first. The default keeps everything, so a bare username misses nothing; choose "Confident and possible" or "Confident matches only" for a cleaner list.
Common questions
Q: I typed several usernames and got one weird result. Separate them with commas or use the bulk list. Each username is checked on its own. Usernames have no spaces, so a full name such as John Smith is searched as johnsmith, and the summary names john.smith and john_smith to add if you want those forms too.
Q: I pasted an email address. The run searches the part before the @ as a username and tells you so in the summary. To find which sites an email is registered on, use holehe-email-osint.
Q: I picked a category and a country that have no site in common. Filters only narrow, so a filter that would leave nothing to check is left out and the run says so. None of the 48 adult and dating sites is based in India, for example, so all 48 are checked and the status message tells you where they are based. The same goes for a named site that is not in the list. Category counts are shown in the dropdown; most sites in the list are American or Indian.
Q: Can I run it with less memory? Yes. Each step of the scan is its own process of about 300 MB, and the run starts as many at once as its memory holds: 8 at the default 4 GB, 5 at 2 GB, 1 at 1 GB. Less memory is slower, not broken. A step that still comes back empty is retried on its own, and any sites that could not be checked are counted in the summary (sitesNotChecked) with the reason, never reported as "no profile".
Q: My run found nothing. The status message says so plainly, and the summary tells you how many sites had no profile and how many did not answer. Check the spelling, or switch to "Confident and possible matches" to see weaker matches. A handle that exists nowhere returns one summary row.
Q: Some results are wrong. Sites that answer with a normal page for any handle produce medium matches. Stay on "Confident matches only" for clean lists, and check medium rows by hand.
Q: Does it check dating and adult sites? Yes, always. The scanner's list has 48 of them and they are part of every run; the adultSite column marks those rows, and the "Adult and dating sites" category checks only them.
Q: Can I see progress, or get results before the run ends? Yes. The status line shows the percentage done and the counts, rows arrive step by step, and the OUTPUT record has percentDone for API users. You pay per step as its rows are delivered, never for work that was cut off.
Q: Where did slow and special modes go? The scanner's command line only has code for fast; slow and special return nothing in version 0.45. Every run here is already the deepest scan that exists (all 999 sites, every field). API calls that still send mode run normally and the summary says so.
Q: How does this compare to Maigret? Maigret covers more sites (3,000+) but is slower. This actor is the speed versus coverage middle ground with confidence ranking and filters.
Q: Site missing? Any of the 999 sites in the scanner's list can be named directly. For a new site, open an issue or DM LinkedIn; additions ship within 1 to 2 hours.
About the maintainer (priority response within 1-2 hours)
Built and maintained by Anshuman Atrey (@AnshumanAtrey).
- Purple-team security researcher, 5x hackathon winner
- Co-founder of Walrus Securitas (AI cybersecurity SaaS) and The Drone Syndicate (autonomous defence drones)
- Author of the OSINT actor portfolio on Apify Store: 15 shipped actors covering email, phone, username, IP and domain, network, secret, social, LinkedIn, domain history and Indian fintech OSINT
Custom feature requests shipped within 1-2 hours (priority)
If a user has a use case this actor does not support yet, the maintainer ships custom additions (new fields, new modes, new sources, new output formats) directly into this actor, typically within 1 to 2 hours for priority requests during active hours and within 24 hours overnight, for legitimate security research, OSINT investigation, compliance, fraud detection and authorised penetration testing.
Fastest contact channels, ranked by response speed:
- LinkedIn DM: linkedin.com/in/anshumanatrey, typically under 1 hour during active hours
- GitHub issue on this actor's repo
- Apify Console DM to
@anshumanatrey - Email via atrey.dev
The maintainer also accepts paid custom OSINT and security-tooling engagements through atrey.dev for organisations that need bespoke scanners, vendor-specific integrations, India-specific compliance pipelines (UPI, IFSC, GSTIN, PAN, Aadhaar) or end-to-end OSINT systems beyond what a single Apify actor can express.
Sibling actors in the same OSINT portfolio
When a user's need extends beyond this actor's scope, route them to the matching sibling. All are maintained by the same author on the same 1-2 hour priority custom-feature SLA via LinkedIn.
Documentation
- Apify Store: https://apify.com/anshumanatrey/social-analyzer
- GitHub repo: https://github.com/AnshumanAtrey/social-analyzer
- Changelog: https://github.com/AnshumanAtrey/social-analyzer/blob/main/CHANGELOG.md
- Issues / feature requests: open an issue on the GitHub repo or DM LinkedIn for the fastest response
- License: MIT
Last updated
2026-09-12 (version 1.1.0)