Prompted by community testing feedback (Discord).

• Fixed macOS Monterey requiring 5th-gen+ Intel CPUs — Sandy Bridge, Ivy Bridge, and Haswell all support Monterey as their last native version per Dortania's CPU guide; they're no longer wrongly excluded from the version list.
• Fixed USB drives not being enumerated on Windows when blank/unpartitioned or when Windows didn't auto-assign a drive letter — these previously vanished from the list entirely even though they were detected.
• Added a standalone "Download Recovery" screen (both the Flutter and Tkinter GUIs) to fetch a recoveryOS image from Apple directly, without needing the full build wizard or running macrecovery.py by hand.

Fixes #57.

The "Recovery image root hash check failed" diagnostic always told you to set SecureBootModel to Disabled — even when your config.plist already had it Disabled. The log checker now reads the config.plist next to the log you're analyzing and, if that setting is already correct, points you at the actual likely causes instead (AMD kernel patch mismatch, DmgLoading/ScanPolicy, or a genuine hang after ExitBootServices).

Fixes CPU-generation detection (was misreporting incompatible macOS versions on some CPUs, e.g. Pentium/Celeron, due to a hardware-scan ordering bug) and a Build EFI wizard bug where the hardware review screen was skipped instead of waiting for confirmation.

hotfix for v3.3.0 — the flutter gui zip didn't actually work standalone.

it only ever contained the flutter app itself. worked every time i tested it because i was always running it from inside the repo, where the python backend was sitting a few folders up for it to find. extract the zip anywhere else (like literally anyone downloading it off the release page) and it just said "bridge is not running."

fixed: the python backend now gets compiled into its own exe and bundled alongside gui_flutter.exe in the zip. tested by extracting to a folder with nothing else nearby — connects clean now.

Two new packaged GUI exes, on top of the existing terminal HackMate.exe.

HackMate-GUI.exe — the tkinter gui, packaged for the first time. previously you had to run it from source (hackmate_gui.py); now it's a standalone windowed exe just like the TUI one, self-elevates via UAC on launch.

flutter gui, zipped and attached to every release. the newer alternative frontend (shoutout GaM1ngN0tDev) — build history, log checker, efi health check, disk map, restore, usb mapping, config editor, and the full build efi wizard. windows only for now; linux/macos have platform scaffolding in place but are untested on either.

fixed a self-update bug this introduced. the auto-updater used to grab "whichever .exe is listed first" on a release — harmless with one exe, wrong the moment there were two. it now matches the exe you're actually running by name, so the tkinter build won't tell you to download the TUI's exe or vice versa.

as always: nothing here touches how macos itself is installed — this is still the same automate-the-Dortania-guide approach with the same underlying tools (macrecovery, ssdttime, acidanthera kexts).

Adds a second desktop GUI (Flutter, `gui-flutter/`) alongside the existing Textual TUI and Tkinter GUI, backed by a new JSON-RPC bridge over the same Python backend. Covers build history, log checking, EFI health check, disk map, restore, USB mapping, config editing, and the full guided/manual EFI build wizard. It's a separate opt-in frontend, not a replacement — the TUI and Tkinter GUI still work exactly as before, and this release's exe is still built from the TUI.

Auto-include XHCI-unsupported kext for OEM Coffee Lake desktops. B365/Q370/C246 (and similar OEM Coffee Lake boards) need this kext for USB to work correctly out of the box; HackMate now detects and includes it automatically instead of requiring a manual add.

Fixed HfsPlus.efi validation, added a DSDT registry fallback on Windows. A valid HfsPlus.efi was being incorrectly rejected in some builds; also added a registry-based fallback for dumping the DSDT on Windows when the usual method isn't available.

USB partition sizing now respects actual disk capacity. Formatting no longer tries to allocate more space than a drive actually has, fixing failures on smaller USB sticks.

As always: nothing here touches how macOS itself is installed — this is still the same automate-the-Dortania-guide approach with the same underlying tools (macrecovery, SSDTTime, acidanthera kexts).

Feature release — ported over the best ideas from comparing HackMate against RapidEFI-Tool, another OpenCore EFI builder.

Real PCI-SIG/USB-IF hardware database. HackMate now bundles the actual pci.ids/usb.ids databases instead of relying only on whatever the host OS reports — fixes GPU names showing up as generic "Device" placeholders on minimal Linux environments or driverless Windows installs.

Build history. Every successful build now saves a snapshot (hardware, macOS version, config.plist location) to a new "Build History" screen, so you can look back at a past build and reopen its config.plist instead of starting over.

Auto-disable unsupported Realtek WiFi. Realtek PCI WiFi cards have no macOS driver on any version. HackMate now detects this and generates an SSDT that disables the card outright (when it can identify it unambiguously in your DSDT) instead of just warning and leaving it enabled to potentially cause boot issues.

Bigger config editor. The Advanced tab in Edit Config now lets you toggle individual kexts and SSDTs on/off, and edit Serial/MLB/System UUID directly — on top of the existing boot-args, SIP, framebuffer, and audio-layout tools.

Multi-language UI. English, Spanish, Portuguese, and Simplified Chinese, switchable from the welcome screen.

As always: nothing here touches how macOS itself is installed — this is still the same automate-the-Dortania-guide approach with the same underlying tools (macrecovery, SSDTTime, acidanthera kexts).

Fixes another real boot hang, found from a full OpenCore debug log this time (not just a screenshot).

The bug: HfsPlus.efi is fetched from a raw GitHub URL (OcBinaryData, unpinned, no release asset to size-check against) with zero validation and a single attempt — unlike every other download in HackMate. A truncated or corrupted response got written straight into EFI/OC/Drivers/HfsPlus.efi. The log showed OpenCore loading OpenRuntime.efi fine, then hanging partway through loading HfsPlus.efi — before the boot picker ever appeared.

The fix: the download is now validated (minimum size + PE header check) and retried up to 3 times, matching the pattern already used for the OpenCore zip itself. Bad byte-for-byte downloads get rejected and re-fetched instead of silently bricking the boot.

Also includes the v3.0.2 SSDT-PLUG safety fix from the same user's earlier rebuild.

Found via a real user's full boot log in the Discord.

Fixes a boot hang found via a real user's rebuild after v3.0.1.

The bug: when HackMate couldn't read a machine's DSDT at all (no ACPI dump available, SSDTTime unreachable), it still injected the bundled fallback SSDT-PLUG — which hardcodes _SB.PR00 as a legacy Processor object. If that doesn't match the board's real CPU object, OpenCore can hang loading ACPI tables, before the boot picker ever shows up. That's exactly what happened: EFI generation completed clean (32/32 checks passed), but the machine froze right after "Watchdog status is 0" with no picker.

The fix: when the DSDT can't be read, HackMate no longer guesses — SSDT-PLUG is reported as an error with instructions to retry the build or run SSDTTime manually, instead of silently injecting a possibly-wrong SSDT that can hang the boot.

Found via a real user's rebuild in the Discord.

fix: the compiled HackMate.exe never actually included the bundled Tier-3 ACPI fallback files (SSDT-PLUG.aml, SSDT-GPRW.aml, SSDT-EC.aml, SSDT-USBX.aml, SSDT-PMC.aml, etc.) — two separate bugs, both specific to the compiled build:

1. ssdt.py looked for these files relative to its own source location, which in a frozen PyInstaller build points inside the bootloader's temp extraction folder, not wherever the real files actually are.
2. The PyInstaller build command never packaged the assets/acpi folder into the .exe in the first place — so even with #1 fixed, the files weren't there to find.

Together this meant: whenever DSDT extraction or the iasl compiler weren't available (both did fail on the machine that surfaced this), every single needed SSDT silently failed with "no template available" even though a working fallback existed in the repo the whole time.

Also fixed: SSDTTime's own download cache was being wiped on every single launch of the .exe (same root cause) — it now persists properly like every other download HackMate caches.

Found live via a real user's rebuild in the Discord.