A disk report you can keep and compare
smartmontools is useful when a drive problem needs more evidence than a single green status indicator. Its smartctl utility reads information supplied by supported storage devices and can request their built-in self-tests. The companion smartd program performs recurring monitoring according to its configuration. These are command-line and background tools, not a graphical disk-repair application with a one-click correction button.
The verified stable release is 7.5, published on April 30, 2025. The project is open source under GNU GPL licensing and supplies Windows executables for the appropriate architecture. Administrator access is commonly needed for device operations. The actual information available depends on the drive, controller and connection path, so installing the program alone cannot make an unsupported USB bridge expose every internal health counter.
Identify the physical device first
Start with device discovery rather than copying a disk address from someone else's example. The smartctl --scan-open command scans and attempts to open detected devices, reporting paths and types. A physical drive is not the same thing as a partition letter: several Windows volumes can reside on one device. Match the reported model and identity with the disk you intend to examine.
After identification, the -x information option provides a broad report of supported information for the selected device. Save the output with a date and a short description of the connection used. If you later move the same disk from an enclosure to an internal connection, the available fields may change. That difference can reflect the communication path rather than a sudden change in the condition of the drive.
Read the values in their own context
ATA attributes, SCSI logs and NVMe health information do not share one universal interpretation table. Raw ATA values can be vendor-specific, and a large number is not automatically a failure count. Consider the meaning of the field, the model and its trend. A general passed status is useful information, but it does not promise that every sector is readable or that future failure is impossible.
If a graphical overview is more convenient for occasional checks, CrystalDiskInfo provides another way to inspect supported disk-health data. Keep the source of each measurement clear when comparing tools: different labels, formatting or unavailable fields do not automatically mean that one program discovered damage which the other missed.
A test request is not a test result
A supported short or extended self-test runs inside the drive after it is requested. Check capabilities and the expected duration before starting, then inspect the self-test log after the relevant interval. A message confirming that a test started does not mean it finished successfully. Interruption, power changes or device limitations can affect the outcome, so retain the result rather than only the command used to launch it.
When a disk is already unstable and contains the only copy of important files, prioritize preservation of those files before lengthy testing. Diagnostics can provide evidence but cannot recover inaccessible documents or repair worn hardware. Neither a clean report nor an additional test substitutes for a separate backup.
Use recurring monitoring deliberately
smartd can monitor selected devices and report changes through configured mechanisms. Installing the binaries is not proof that a service is running or that an alert will reach you. Verify the device list, configuration and notification route, then keep a manageable reporting schedule. For automated processing, structured JSON output from smartctl can be more dependable than scraping visually aligned text columns.