Explore with less exposure.

Our philosophy

TorSeeker puts distance between curiosity and direct exposure. Search stored records, inspect observed availability and choose whether to reveal an archived screenshot. Our goal is to help you understand the darknet while reducing the need to connect to unknown services. We cannot promise risk-free browsing, anonymity or protection against every harmful image, false claim or scam.

Assume a service may be a scam

The darknet contains scams, impersonation, phishing and malicious services. Treat every listing as untrusted, including those marked online. We do not verify sellers, transactions or claims, and we do not have a validated estimate of what percentage of listed services are fraudulent. A professional-looking page, multiple mirror addresses or a long observation history does not establish legitimacy.

Do not send money or cryptocurrency, disclose personal details or passwords, download files, or follow instructions based on a listing. Cryptocurrency payments are typically irreversible; see the FTC's cryptocurrency scam guidance. Stored previews may still contain distressing, adult, violent or deceptive material. Leave content blurred if you do not want to see it.

Stored observations versus live visits

Searching and viewing our records or archived images does not open the source onion site in your browser. External onion links are different: following one leaves TorSeeker, requires Tor and exposes you to the destination's content and practices. TorSeeker is not a proxy, VPN, malware scanner or anonymity guarantee.

Discovery and verification

We collect v3 onion addresses from seed lists and links on crawled pages. Address version and checksum checks validate the address format, not the operator or content. The first discovery date is when we recorded a service, not when it was founded.

Reachability is not trustworthiness

Any HTTP response, including a 404 or 500, counts as reachable. One failed check marks a service degraded; three consecutive failed checks mark it offline. Workers pause when the Tor connection check fails. Displayed uptime is the proportion of successful sampled checks over 30 days, with the sample count shown. It is not a guarantee of continuous availability.

Bounded crawling and content groups

TorSeekerBot respects robots.txt, noindex and nofollow directives. Default crawl budgets are 25 pages, two link levels, ten minutes and 50 MB per service. HTML is processed without running scripts. Search queries do not initiate live Tor requests. Matching titles or matching or highly similar landing-page text connect services into groups before search, directory and recent-observation pagination. Title matching ignores capitalization and extra spaces. Groups are recalculated as observations change; a connection does not prove a shared operator. Other addresses remain inspectable; grouping alone is not a spam verdict.

Small, consent-gated screenshots

Eligible pages may have a small WebP copy of their first screen archived. Screenshot collection does not use source noarchive, noimageindex, nosnippet or max-image-preview directives, or robots.txt, as an archive permission gate. Search indexing and link discovery have separate rules. Rendering takes place in an isolated, offline browser with scripts disabled and a restricted set of fetched resources. It may differ from the live site. Previews are blurred on the server. The clear image is sent only after you accept that image's warning, with permission expiring after ten minutes. Image routes ask search engines not to index or archive them.

Ranking, language and limitations

Search uses text relevance across titles, headings, descriptions and body text, with reduced ranking for unreachable services. Our interface is English only. Source titles, descriptions and screenshots retain their original language and may contain unverified claims. Content-language filters describe the source, not the interface. Displayed titles are limited to 120 characters (100 in directory rows), descriptions to 500 and search snippets to 320; longer stored text remains searchable. Coverage, categorization and observations may be incomplete or stale.

Reporting and operator requests

Use Report content for abuse or rights concerns. Operators can restrict crawling through robots.txt or submit a removal request. Requests are reviewed; a report does not automatically remove another operator's service. Read the Terms of Use before using this service.