Skip to content

Repository files navigation

MultyCapture

Do something once; get a written procedure out of it.

MultyCapture records every mouse click and keystroke — each with a screenshot and the active window's context — and turns that into a numbered Word document with the screenshots in place:

Step 4 Click "Save" in "Orders". "Save changes?" appears.

Both lines are written without any AI: the first from the recorded event, the second by comparing the screenshots either side of the step.

User Guide — installing, where your files go, what is optional, and the AI features.

Platform support

Platform Status
Windows 10/11 Supported, no extra system dependencies
Linux / X11 Supported
Linux / Wayland Not supported — log into an Xorg session

Wayland does not let one application observe another's windows or read the pointer globally. MultyCapture detects it and refuses to start rather than recording something useless.

Quick start

Install the .deb or the Windows installer, then run MultyCapture — or from a source checkout:

python -m venv .venv
source .venv/bin/activate      # Windows: .\.venv\Scripts\Activate.ps1
pip install -e .
mc tray

Left-click the tray icon to start; Ctrl+Alt+Q stops. There is a 5-second start delay so you can get to the right window first. When you stop, the document is written to Documents/MultyCapture and opened.

Command line

mc record        # stop with Ctrl+Alt+Q or Ctrl+C
mc doc --last    # build a document from the most recent recording

Full options in the User Guide.

Optional extras

  • tesseract-ocr makes steps name what was clicked — Click "Save"… rather than Click in…. Recommended by the .deb, not required; without it nothing fails and the steps are simply less specific.
  • An AI model can improve the wording of the steps, and nothing else. Off by default, because it can send your captured text off the machine. Nothing needs installing for it: every backend (Ollama, Claude, OpenAI-compatible, Gemini) is reached over plain HTTP.

Documentation

USER_GUIDE.md How to use it, and what it does with your files
CAPTURE_SPEC.md The event stream and storage format
decisions.md Why it is built this way, and what was tried and dropped

Development

pip install -e ".[test]"
pytest                          # Linux: xvfb-run -a pytest

The version lives in src/multycapture/_version.py and nowhere else; the packaging scripts read it with installer/version.sh, and CI refuses to build a tag that disagrees with it.

About

Thi project aim to build a multi captures system that produce a documentation docxfile with the relevant screen cast

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages