How The Tordex Search Engine Works: Crawling & Index

Quick answer: how does Tordex work?

Tordex crawls pages it can reach through Tor, stores their text and links in an index, and matches a user's keywords against that stored text. Each stage introduces gaps, so the result list is always a partial view.

Stage one: discovery

A crawl needs somewhere to begin. Tordex can start from addresses already known to it, links found on pages it has fetched, and addresses that users or operators submit. Each fetched page can add new addresses to the queue. The queue is therefore shaped by earlier discoveries, which is why some clusters of services are over-represented and others never appear.

Stage two: storage

Once a page is fetched, the engine extracts text, headings, links, and sometimes metadata such as a title or a content hash. It stores enough to answer keyword queries later. Storage decisions matter: a page whose text is not extracted cannot be found by a content search, and a page whose links are not parsed cannot lead the crawler onward.

Stage three: matching a query

When you search, the engine compares your terms with the stored index and returns a ranked list. Ranking may weigh title matches, term frequency, link relationships, or recency. The exact recipe is usually unpublished, and it can change. A first-place result means the engine considered it a match, not that it is the best or most reliable service.

Why onion crawling is slow

Every fetch travels through the Tor network. That adds latency and failure modes that clear-web crawlers do not face. A page that times out may be skipped, retried later, or quietly dropped. A service that blocks automated clients may never enter the index even if people use it every day.

Why coverage is uneven

Well-linked pages are discovered more often. Text-heavy pages are indexed more fully. Pages behind a login, built with heavy client-side code, or prone to errors are harder to capture. The result is a catalogue with uneven depth, not a uniform sample of the network.

What this means for your reading of results

Use the result list to generate leads, then verify. Record when you ran the query and what the engine showed you. If you later compare engines, the differences will tell you more about coverage than any single list can.

Tordex FAQ

What is a crawl seed?

A seed is a starting address the crawler knows about. New seeds can come from links on pages already crawled or from addresses submitted by users.

Does Tordex crawl continuously?

That depends on the operator and is usually not published. Treat any index as a snapshot with an unknown delay.

Why do results change between searches?

The index may be updated, the ranking may change, or the service may have gone offline. Record the date and query so you can tell which explanation fits.

Can Tordex see private onion services?

No. Services that are unlinked, private, offline, or protected from automated access are unlikely to appear in a public crawl.

Last updated: October 2026. Site map.