v1.36.1 - Published by ILHS Software, LLC

Published by ILHS Software, LLC

• ScreenCap Docs is now published by ILHS Software, LLC, which also holds its copyright. The app, its MIT license, and everything you can do with it stay the same.
• The installer, the About window, the Windows file details, the Linux packages, the software center listing, and winget name ILHS Software, LLC as the publisher.
• Nothing changes for an installed copy: the install folder, settings, and updates stay the same, and this update installs over the top as before.

New

• The project website is live at screencap.ilhs.app, with the downloads for every system and the apt and dnf repository setup in one place.

v1.36.0 - The screencap command in any terminal

The screencap command in any terminal

• The Windows setup adds the program folder to the PATH, so screencap works in any new PowerShell, Command Prompt, or Windows Terminal window. An install for all users adds it to the system PATH; an install for you only adds it to your own PATH. It is ticked by default on the Options page, works in silent installs, and is never added twice. Untick it, or use /MERGETASKS="!addtopath", to leave the PATH alone. Uninstalling takes it out again.
• winget lists the screencap command for this package.

New

• Every release file now has a build provenance record: gh attestation verify <file> --repo ILikeHostingServices/Screencap-Documentation-Tool proves it was built by this repository's GitHub Actions workflow and not changed afterwards.
• Each release lists the Windows app's components in a CycloneDX SBOM (*-windows-sbom.cdx.json), for vulnerability scanners and procurement reviews.
• ScreenCap Docs for IT and Security Teams: what the app does on a PC, its network use, files, administrator settings, and how to verify a download.
• CodeQL and the OpenSSF Scorecard check the repository on every change.
• CONTRIBUTING.md explains how to report bugs, propose a change, and what a pull request must include.

Improved

• The README leads with the Windows setup as the main way to install. The PowerShell install command is in its own section for power users and developers.

Fixed

• A recording failed to process when automatic blurring found text in the last pixel row or column of the screen, or when a blur box was only a few pixels wide: FFmpeg refused the box and the whole recording was marked Failed. Blur boxes are now kept inside the screen and are at least 8 pixels on each side, so they always render and still cover the text.

v1.35.1 - Security and privacy fixes

Security and privacy fixes

• Characters typed with AltGr (such as @ or a euro sign on many European keyboards) or with Option on a Mac were noted as shortcuts such as "Ctrl+Alt+Q", which could reveal characters of a password. They now only count as "type the text", like every other typed character. Real shortcuts such as Ctrl+Alt+Delete are still named.
• Administrators can switch noting keys off for everyone on a PC with the environment variable SCREENCAP_NO_KEYS=1 (Group Policy, Intune, or a login script). Clicks are still noted, and the Record window shows that the option is switched off.
• The Windows install script no longer leaves the program files changeable by every user on the PC. A new folder such as C:\DATA\Tools used to inherit the C:\ rule that lets any signed-in user change files in it. Now only administrators (and, for a just-me install, your own account) can change the program; the source and output folders stay writable. Running the install command again applies this to an existing install.
• The install scripts now install the newest release, which is tested and published, instead of the newest code on the default branch. Add -Ref HEAD (Windows) or SCREENCAP_REF=HEAD (Linux and macOS) to get the newest code.
• The Windows install script no longer clears the "downloaded from the internet" mark from files, which security tools watch for. It was not needed.
• Update Now refuses an installer whose file name contains a folder.

Improved

• The build workflows pin the third-party GitHub Actions they use to an exact commit, and Dependabot proposes updates to them.

Fixed

• Captions in Word exports are shown exactly as typed, like in the HTML, PDF, and PowerPoint exports. Characters such as *, _, [, or a backtick no longer turn into formatting or links, text read from the screen can no longer add Word markup, and line breaks in a caption are kept.
• The HTML export's footer no longer shows the recording's file name, which can reveal internal names. It reads "Created with ScreenCap Docs".

v1.35.0 - New name: ScreenCap Docs

New name: ScreenCap Docs

• The app is now called ScreenCap Docs (formerly Screencap Documentation Tool): in the window title, the About window, the Start Menu and desktop shortcuts, Apps & Features, the installer, the Linux application menu and software centers, the Snap Store, the Homebrew app, and exported documents.
• Only the shown name changed. The repository, the winget package ILHS.ScreencapDocumentationTool, the Linux and Homebrew package names, the Windows install folder, the program file Screencap Documentation Tool.exe, and the screencap and screencap-gui commands keep their names, so updates, scripts, taskbar pins, and application allow-lists keep working.
• Updating replaces the old Start Menu and desktop shortcuts with ones under the new name. Settings, presets, and chosen folders are kept.
• New installs keep recordings and output in Documents\ScreenCap Docs. A Documents\Screencap Documentation Tool folder from an earlier version keeps being used.
• Homebrew: the app is now ScreenCap Docs.app. Run the ln -sf command from the README again, and remove the old link with rm ~/Applications/"Screencap Documentation Tool.app".

Fixed

• The README no longer says the Arch package is on the AUR yet: it is published there as soon as the AUR reopens new account sign-ups. Until then, install the .pkg.tar.zst from the release with sudo pacman -U.

v1.34.1 - Installer keeps the install mode

Fixed

• The Windows installer no longer asks "Install for all users or for me only?" when the tool is already installed: it updates the installed copy the same way (for everyone, or for you only). To switch, run the installer with /ALLUSERS or /CURRENTUSER.
• A silent update without a scope (for example from a deployment tool) now updates a copy installed for one user in place, instead of installing a second copy for everyone.

v1.34.0 - Update Now

New

• Update Now: on a copy installed with the Windows installer, the update window has an Update Now button. It downloads the new installer for your PC (x64 or ARM64), checks it against the release's SHA256SUMS.txt, closes the app, installs the update silently for the same scope (just you, or everyone on the PC), and opens the app again. Settings, presets, folders, and recordings are kept.
• A download that does not match the checksum list is deleted and never run, and only this project's own GitHub release files are downloaded. A copy installed for everyone still needs administrator rights, so Windows shows its permission prompt.
• The update window shows the release notes, with Remind Me Later and Skip This Version. For other installs it shows the command to copy (Homebrew, apt, dnf, the AUR, snap) or opens the download (the portable zip).
• SCREENCAP_NO_UPDATE_CHECK=1 turns Update Now off along with the update check, for PCs where updates are deployed centrally.

Testing

• On real Windows machines (x64 and ARM64), for one user and for everyone: v1.33.0 is installed, the new updater downloads this release from a local test server, checks it, installs it silently, and the updated app opens again with the folders kept.

v1.33.0 - Faster processing and fixes

Faster

• Processing is about three times faster on a 4-core PC: a 3-minute, 1080p recording with 40 steps now takes 27 seconds instead of 81, and a recording made with the built-in recorder 4 seconds instead of 11. The results are the same, screenshot for screenshot.
• Saving the screenshots, reading the clicked words for captions, scanning for sensitive text, and rendering now run on all CPU cores (up to 8 at once) instead of one screenshot at a time. Save Changes in the step editor is faster for the same reason.
• An unblurred copy of a screenshot with nothing to blur is now copied instead of being rendered a second time, and screenshots are compared row by row, skipping the rows that did not change.
• SCREENCAP_WORKERS (an environment variable) sets how many cores to use; SCREENCAP_WORKERS=1 does everything one at a time.

Fixed

• Tesseract OCR could stall for minutes on a busy PC: its own threads compete for the CPU with each other. It now runs one thread per scan, which is also faster on its own.
• A change made in the step editor while Save Changes was still running was marked as saved even though it was not in the saved files, so it could be lost on closing. It now stays marked as unsaved, and the status bar says to save again.
• If FFmpeg could not be started for a recording, the tool kept reading clicks and keys in the background until it was closed.
• A Word or PDF export that hangs now ends with a clear message ("did not finish within 3 minutes") instead of an internal error.
• screencap.py --help mentions --record.

v1.32.0 - PowerPoint export

New

• Export to PowerPoint: PowerPoint in the Export box (command line: --export pptx) saves the steps as a widescreen presentation for training sessions and handovers: a title slide with the title, version, date, author, and logo, then one slide per step with "Step N", the screenshot as large as fits, and the caption under it.
• What was said during a step (Captions From Narration) goes in that slide's speaker notes, so the presenter has the script; screenshots keep their alt text; a step with an animated clip gets one more slide where the clip plays.
• The logo (PNG, JPG, or GIF) is on every slide. The Unblurred copy option works as for the other exports and puts the warning on the title slide.
• No other program is needed: the presentation is written by the tool itself (pptxwriter.py, Python standard library only) and opens in PowerPoint, Keynote, LibreOffice Impress, and Google Slides.

Testing

• Every exported presentation in the tests is checked with Microsoft's Open XML SDK (the rules Office itself uses, for Office 2007, 2016, and Microsoft 365) and read back with python-pptx, on Windows and Linux.

Changed

• The screenshots in the README and the software center listing show the PowerPoint button, and a new one shows an exported presentation. From now on, every feature release updates the screenshots.

v1.31.2 - Better click captions, and screenshots in the store listings

Fixed

• First-draft captions name the clicked button even when the mouse pointer covers part of its words: the same place is read a moment before the click too, and the fuller reading is used when it holds the shorter one (so Click "Ne". is now Click "Next".).
• A checkbox or option button clicked on its box gets its label from the words to the right of it (Click "Create a desktop shortcut".), instead of "the marked spot".
• Text box borders and the text cursor, which OCR reads as | or I, no longer hide the label of an empty text box (Click next to "Product key"), and the border between two buttons no longer joins their words (Click "Back | Next").

New

• Screenshots of the app in the README and in the Linux software center listing (GNOME Software, KDE Discover), plus a Snap Store button and download and Homebrew badges at the top of the README.
• packaging/screenshots/make_screenshots.sh makes the screenshots again after interface changes: it records a made-up setup wizard with the built-in recorder on a virtual display, so the click rings, blurring, and captions in the pictures are the tool's own.

v1.31.1 - Snap Store fix

Fixed

• The snap of v1.31.0 was held for manual review by the Snap Store, which asks for the audio-playback permission together with the microphone (audio-record) permission the recorder added. The snap now has both, so it is published to the Snap Store again. Nothing else changed; the other downloads are the same as v1.31.0.