What the sources actually say
The official documentation is direct: before running zapret you must discover a working bypass strategy, and blockcheck.sh automates that process. It first checks DNS, then tries many strategies to find the working ones. The script works on all systems zapret supports.
The documented DNS stage is mostly Russia-targeted: it resolves several pre-defined domains blocked in Russia, compares system DNS answers against public DNS answers, and verifies that returned answers are unique — an ISP that blocks DNS often returns a single IP for every blocked domain to serve an access-denied page. When DNS spoofing is detected, DoH servers are used automatically for the checks.
What blockcheck reports is grounded in the documented desync modes and filters of nfqws: the dpi-desync modes, hostlist and ipset filters, and port-range parameters define the space of strategies it probes. The documentation also records a hard limit: some conditions — such as certain IPv6 fragmentation firewall configurations on Linux — must be fixed manually, and blockcheck.sh cannot auto-fix them for you.
In the Windows world, the Flowseal build wraps the same idea in service.bat: Run Tests offers Standard tests, which check the sites listed in utils/targets.txt, and DPI checkers, which probe DPI behavior against providers such as Cloudflare and Amazon. Community Discussion #251, the long YouTube troubleshooting thread, reflects the same practice — participants recommend blockcheck and manual strategy selection instead of assuming any posted configuration will transfer, because the thread itself is full of user-specific DPI parameters that worked for one network and failed on another.
How to test strategies, step by step
- Establish a clean baseline first, following the guide checklist: no other bypass tools running, no unsupported test browser, and security software not interfering with winws.exe.
- On Linux, macOS or OpenWRT, run blockcheck.sh from the official repository and let its DNS stage finish before interpreting strategy results — a poisoned DNS answer invalidates everything after it.
- On the Flowseal Windows build, use service.bat, then Run Tests: Standard tests against the bundled target list, and DPI checkers when you need provider-level information.
- Test strategies one at a time and record which build version, strategy name and target site each result belongs to — the community discussion shows results are meaningless without that context.
- Confirm a candidate strategy against every service you use, not just the one that was failing, before installing it as a service.
- Re-test after build updates or network changes, because both the strategy space and the DPI behavior it probes can change.
What is source-backed vs community-reported
- Source-backed (official documentation): blockcheck.sh exists to automate strategy discovery, its DNS verification logic, the automatic DoH fallback on spoof detection, its cross-platform support, and the manual-fix limitation.
- Source-backed (build README): the service.bat Run Tests workflow with Standard tests and DPI checkers, and the README statement that strategy effectiveness depends on many factors.
- Community-reported (Discussion #251): the recommendation to use blockcheck and manual selection, and the individual parameter reports that demonstrate why copying a stranger's strategy often fails.
When the reported workaround may not apply
A blockcheck result is a measurement of your network at one moment, not a permanent property. Providers change DPI rules, and the documentation itself flags cases blockcheck cannot fix automatically. The documented DNS stage is Russia-oriented, so its predefined domain checks may be less meaningful elsewhere. Test tooling also cannot see application-layer conflicts — the guide's browser warning shows a case where zapret worked and the test still failed. No strategy, however carefully tested, is universally best or guaranteed across ISPs, browsers and regions.