Excavator Tor Link, Tordex, DeepSearch: Judging Smaller Onion Search Engines
Why smaller onion search engines are worth evaluating carefully
The famous engines attract clones because of their fame. Smaller engines carry a different risk: nobody has looked at them closely, and the person recommending one in a forum post may be its operator. An engine controls what its users find, which makes running one attractive to anyone who wants to steer traffic toward particular sites. A search engine that ranks its own clones first, or that quietly rewrites addresses in results, is a more efficient phishing tool than any single fake page. This is not an accusation against any particular engine. It is the reason evaluation has to rest on evidence rather than reputation. An excavator tor link recommended in three comment threads has a reputation of three comments, which is nothing. What you can gather instead are observable properties of the engine itself, and those are the same regardless of name. A scenario: a researcher hears that a new engine indexes more than the established ones. They obtain an address from a forum, open it, and find it works well. They start using it daily. Months later they learn from a warning thread that results for several popular services had been pointing to clones the whole time. Nothing in their experience of the engine told them; only the missing evaluation would have.
Test one: does the excavator link tor page work at the safest security level
The first and cheapest test applies to any engine, including whatever excavator link tor address you are considering. Set Tor Browser's security level to safest, which disables JavaScript, and load the engine. Run a search. A search engine has no technical need for scripts to accept a query and return a list of links, and engines built with Tor users in mind work fine this way. An engine that shows a blank page, demands that you enable scripts or fails silently at the safest level is telling you one of two things. Either it was built carelessly, without regard for the browser its users are told to keep locked down, or it wants scripts running for its own reasons. Neither is disqualifying on its own, but both reduce trust and raise the question of why. A related test is what the engine does with your query. Some engines pass search terms in the address bar; some use forms. Either is fine. What is not fine is an engine that redirects you through intermediate addresses before showing results, or that loads resources from other onion services or the clearnet. Tor Browser's safest level blocks much of this, which is another argument for keeping it there while you evaluate.
Test two: do the tordex onion operators publish their address anywhere accountable
The single strongest signal of a trustworthy onion service, search engine or otherwise, is whether the operators publish their address somewhere that is hard to impersonate. For a tordex onion address, or any other, look for a clearnet page the operators control, a long-lived PGP key used to sign announcements, or documentation on a platform where the operator's identity has history. When such a source exists, you can obtain the address from it, verify it, bookmark it and stop worrying about clones of the entry point. When it does not, every address you will ever see for that engine is a third-party claim, and the engine's usefulness has to be weighed against that permanent uncertainty. Some engine operators do this well, publishing their address and a key and signing updates. Some do not bother. Some cannot, because they prefer complete anonymity, which is their right and also a fact you must price in. Our Useful Resources page explains how to check a signed announcement. A scenario: two engines return similar results for a test query. One has a clearnet page with a signed onion address that has been stable for years. The other exists only as addresses passed around in threads. A careful user bookmarks the first and treats the second as a curiosity, not because its results were worse but because its entry point can never be confirmed.
Test three: how the deepsearch onion results treat advertising and ranking
Run a query for a well-known, legal onion service and study the results page from any deepsearch onion engine or its peers. You are looking at how the engine treats money and manipulation. - Are sponsored results labelled clearly and separated from organic ones, or mixed in with subtle styling? - Do results that claim to be the same service with different addresses all appear, or does the engine show signs of deduplication? - Does the genuine service, whose address you already know from its operator, appear at all, and where? - Do keyword-stuffed spam pages dominate, or has the engine filtered some out? - Does the engine display any information about when a page was last crawled? No onion engine does all of this well, because the underlying corpus is small and hostile. What you learn is relative: how this engine compares to others on the same query. An engine whose top result for a famous service is a clone, with the genuine address absent, is either poorly maintained or complicit, and either way not a tool to rely on. An engine that surfaces the genuine address, labels ads and shows crawl dates is doing what can be done. Record the comparison in your notes so you do not repeat it.
Test four: does the deep search onion link behave consistently over time
Trust in onion services is built over time, and a deep search onion link that passed every test today has still only passed them today. The last dimension is consistency, and it is the one most users skip because it requires patience. Revisit the engine periodically and repeat the same test query. Note whether the results for your known service change in ways that make sense, such as a new address appearing after a signed announcement, or in ways that do not, such as the genuine address vanishing while clones remain. Note whether the engine's own address changes, and whether any change was announced through the accountable channel you identified in test two. Note whether the interface starts requiring scripts, adding redirects or displaying new advertising. An engine that behaves the same way for months is not proven trustworthy, but it has given you something clones and traps rarely give: a track record you observed yourself rather than read about. Sudden changes without announcement are the strongest reason to stop using an engine and return to the operator's channel to find out what happened. Bookmarks help here too. If the bookmarked address stops working and there is no signed explanation, resist the urge to search for a new address on the clearnet, where the first results after an outage are frequently planted.
Applying the tests to a darksearch onion link, a dark search onion link or any new name
New engines will keep appearing, and each will arrive with a name, a handful of recommendations and an unverifiable address. Whether it is a darksearch onion link or something that does not exist yet, the routine is the same. 1. Do not open it from the recommendation. Look first for an accountable source of the address: a clearnet page, a signed announcement, documentation with history. If none exists, decide now whether you are willing to use a tool whose entry point can never be confirmed. 2. Open it at the safest security level. If it does not work without scripts, note that and weigh it. 3. Run a test query for a legal service whose genuine address you already hold. Study the results page for ads, duplicates, spam and the presence of the genuine address. 4. Record what you found: date, address, source, test results. 5. Revisit over weeks and repeat. Watch for unannounced changes. 6. Bookmark only after the engine has passed and stayed passed. Use the bookmark, never a directory, from then on. A scenario: a student evaluates two engines this way over a month. One passes everything and stays stable; the other fails the script test and changes address without explanation in week three. The student keeps one bookmark and a note explaining why the other was dropped. Neither decision relied on anyone's opinion.
A method outlasts any list of names
Excavator, Tordex, DeepSearch and DarkSearch are names people search for today, and in a few years the list will be different. What does not change is that an onion search engine controls what its users find, that its address cannot be verified by inspection, and that its operators may or may not have made themselves accountable. The four tests here measure those properties directly, and they apply to any engine regardless of what a forum thread says about it. The tests also share a theme with everything else on this site: replace reputation with evidence, and replace convenience with a check you can repeat. An engine that passes is a useful tool for discovering names and reading context. An engine that fails is a liability wearing a search box. Neither conclusion requires trusting anyone. One step to take today: pick whichever smaller onion search engine you have heard of most recently and, before opening it, spend ten minutes looking for an accountable source of its address. What you find, or fail to find, will tell you more about that engine than any review, and it is the first of the four tests in a routine you can now apply to all of them.
Frequently asked questions
is excavator a safe onion search engine?
No engine can be declared safe by name, because the address you reach may be a clone and the operators may or may not be accountable. Evaluate it with observable tests: works without JavaScript, address published somewhere accountable, honest treatment of ads and duplicates, and consistent behaviour over time.
what is the difference between tordex, deepsearch and darksearch?
They are separate onion search engines that index Tor sites by crawling and text matching. Their coverage, interface, advertising and operator accountability differ and change over time, so compare them with the same test query rather than relying on descriptions.
why should a dark web search engine work without javascript?
Returning a list of links for a query needs no scripting, and the Tor Project recommends the safest security level, which disables JavaScript, for users who need the most protection. An engine that insists on scripts is either carelessly built or wants them running for its own reasons.
how do i know if a search engine is showing me clones?
Run a query for a legal service whose genuine address you already have from its operator, then check whether that address appears and how many results claim the same name with different addresses. If the genuine address is missing while look-alikes rank, the engine is not reliable.
what should i do when a search engine changes its address?
Check the operators' accountable channel, such as a clearnet page or PGP-signed announcement, for an explanation. Do not search for a new address on the clearnet, because the first results after an outage are frequently clones planted for that moment.