RunTheTests
/

Interpreter tools

24 tools that run in your browser. Nothing to install, nothing uploaded.

Paste the output your own diagnostic tools produced — a ping run, a blue-screen code, a battery report — and get it explained in plain English. These interpreters parse everything locally; nothing you paste is uploaded, which matters because diagnostic output often contains serial numbers and hostnames. Crucially, this section never asks you to run or paste a command it gives you: it reads output from tools you already chose to run, because "paste this into your terminal" is exactly how the most common malware of today spreads.

You run the command; this reads the output

Every tool in this category works on text you already have. You run smartctl, or open a battery report, or copy a ping run, or write down a stop code — and paste the result here to have it explained. That direction is deliberate and it is the category’s single firmest rule: nothing here will ever hand you a command and ask you to run it. Talking somebody into pasting an unfamiliar command into a terminal is the mechanism behind the most successful consumer malware of the last few years, and a diagnostic site that normalises the habit is training its readers to be victims.

Why parsing locally matters for diagnostic output

Diagnostic reports are far more identifying than people expect. A SMART report carries the drive’s serial number, a DxDiag dump lists your hardware and often your machine name and user account, a battery report includes the manufacturer’s device serial, and email headers contain internal hostnames and addresses. All of it is parsed in your browser here, with no upload step, which is what makes it reasonable to paste a full report rather than hunting through it for the lines you think are relevant — and hunting selectively is how people paste the wrong five lines and get a confident answer to the wrong question.

A verdict is only as good as the sample behind it

These tools say when your data is too thin to support a conclusion, which is where most automated interpretation goes wrong. Ten ping packets cannot establish a packet loss rate — one lost packet reads as 10%, and the honest response is that the sample is too small, not that the connection is failing. A single SMART snapshot shows current counts without the trend that actually predicts failure, so a reallocated-sector count of four means something very different if it was zero last month. Where the input cannot carry the weight of a verdict, you are told that instead of being given one.

The attributes and codes that actually predict failure

A SMART report lists dozens of attributes and almost all of them are noise for a consumer decision. The ones that carry real predictive weight are few: reallocated sectors, pending sectors, uncorrectable sectors, CRC errors on the interface, and for an SSD the wear-levelling and total-bytes-written figures against the drive’s rated endurance. The interpreters here build their verdict from those and say so, rather than turning a wall of raw values into an unexplained score. Blue-screen stop codes work the same way: the code narrows the cause to a class of fault, and the useful next step is almost always driver or memory testing rather than a reinstall.

Frequently asked questions

Why will these tools not just give me the command to run?

Because talking somebody into pasting an unfamiliar command into a terminal is the delivery mechanism behind the most successful consumer malware of recent years, and a diagnostic site that normalises the habit is training its readers to fall for the next one. These tools read output from commands you already chose to run. Where you genuinely need to know how to produce a report, the instruction is to use your system’s own documented tool by name, not to copy a line from a web page.

Is the report I paste sent anywhere?

No — everything is parsed in your browser. That matters more here than in most categories, because diagnostic output is unusually identifying: SMART reports carry drive serial numbers, DxDiag dumps list your hardware and often your machine and user names, and email headers contain internal hostnames. Local parsing is what makes it reasonable to paste the whole report rather than picking out the lines you think matter, which is how people get a confident answer to the wrong question.

My drive shows a few reallocated sectors. Should I replace it?

Not on that number alone — the trend is what predicts failure. A stable count of a handful of reallocated sectors that has not moved in months is common and not urgent. The same count on a drive that read zero last month is a very different signal, as are pending or uncorrectable sectors, which indicate data the drive currently cannot read. Either way, a drive showing any of these is a drive whose backups you should verify today rather than plan for.

Why does the tool say my ping sample is too small?

Because a packet loss percentage from ten packets cannot support a conclusion. One lost packet in ten reads as 10% loss, which sounds alarming and means almost nothing — you need a few hundred packets before the figure is stable enough to act on. The analyser reports the sample size alongside the result and refuses to give a confident verdict where the data cannot carry one, which is the difference between a measurement and a guess with a decimal point.