FlowCapture

Turn real workflows into documentation — locally, with your own AI key.

Website · Download · Quick start · Documentation · Report a bug

---

FlowCapture is an open-source, local-first desktop app that records your screen, mouse, keyboard, and window activity — then uses your own AI API key to produce step-by-step documentation, SOPs, and export-ready guides.

No cloud account required. Sessions stay on your machine in SQLite. You bring the AI provider (OpenAI, Claude, Gemini, or Ollama).

Download

Platform	Status	Link
macOS (Apple Silicon)	Available	Download v0.1.0 .dmg
Windows (x64)	Available	Download v0.1.0 .exe
Linux	Build from source	See Build from source

macOS may show an “unidentified developer” warning for unsigned builds. Open System Settings → Privacy & Security and choose Open Anyway, or right-click the app and select Open.

Features

• Rich capture — screen video, event-driven screenshots, mouse, keyboard, and window tracking
• Local-first storage — SQLite on your machine; optional redaction before AI processing
• BYOK AI — OpenAI, Claude, Gemini, and Ollama; you control keys and models
• Multi-stage pipeline — timeline compression, screenshot selection, writing, and review
• Exports — Markdown, styled HTML, PDF, and video
• Session replay — step through captured workflows before generating docs

How it works

1. Record — perform your workflow; FlowCapture captures events and screenshots as you go.
2. Review — inspect the compressed timeline and replay steps in the app.
3. Generate & export — run the AI pipeline, edit the Markdown, then export in your preferred format.

Quick start

1. Install FlowCapture from Releases (or build from source).
2. Open Settings and enable an AI provider with your API key.
3. Click Start Recording on the home screen and complete a workflow.
4. Use the overlay to Mark Step (Cmd+Shift+M on macOS) or Stop when finished.
5. Open the session, click Generate Documentation, review the output, and export.

For a full walkthrough, see the Quick start guide.

Build from source

All platforms require Node.js 20+ and Rust stable.

git clone https://github.com/Abhi6722/FlowCapture.git
cd FlowCapture

Then follow the steps for your OS below. The first tauri:dev run compiles the Rust backend and may take several minutes.

macOS

Requirement	Notes
Node.js 20+	Required
Rust (stable)	Installed by setup:install if missing
Xcode Command Line Tools	xcode-select --install

npm run setup          # check deps + npm install
npm run tauri:dev      # run the app

If Rust is not installed yet:

npm run setup:install
npm run tauri:dev

Runtime: Screen Recording and Accessibility permissions are required. See permissions docs.

Linux

Requirement	Notes
Node.js 20+	Required
Rust (stable)	Installed by setup:install if missing
Ubuntu 22.04+	Or any distro with WebKitGTK 4.1 (libwebkit2gtk-4.1-dev)
System libraries	Tauri (WebKitGTK) + X11 screen capture (xcb, dbus) — installed by setup:install

npm run setup:install       # Rust + system libs (needs sudo) + npm install
source "$HOME/.cargo/env"   # load cargo into current shell
npm run tauri:dev

Add this to ~/.bashrc or ~/.zshrc so cargo is available in new terminals:

. "$HOME/.cargo/env"

Runtime: On Wayland (Ubuntu 22.04 default), screenshots use GNOME Shell / portal capture. If captures are blank, install gnome-screenshot (sudo apt install gnome-screenshot) or log into an X11 session. ffmpeg is downloaded automatically on first build.

Common compile errors on Linux:

Error	Fix
webkit2gtk-4.1 / xcb / dbus-1 not found	Run npm run setup:install
spa_video_info_raw has no field named flags	Pull latest — Linux uses xcap 0.4.x without PipeWire. Run cargo clean in src-tauri and rebuild.

Windows

Requirement	Notes
Node.js 20+	Required
Rust (stable)	Installed by setup:install if missing
Visual Studio Build Tools	Install the Desktop development with C++ workload
WebView2	Usually preinstalled on Windows 10/11

npm run setup:install
npm run tauri:dev

Restart your terminal after Rust installs so cargo is on your PATH.

Runtime: Periodic frame capture encoded to MP4 via bundled ffmpeg.

Production build (all platforms)

npm run tauri:build

Packaged artifacts are written to:

src-tauri/target/release/bundle/

Development

Command	Description
npm run setup	Check prerequisites and install npm deps
npm run setup:install	Same as setup, plus install Rust / Linux system packages
npm run tauri:dev	Run the desktop app with hot reload
npm run build	Build the app frontend
npm run tauri:build	Build a production desktop bundle
npm run test:rust	Run Rust tests
npm run build:linux	Build/test in Ubuntu 22.04 Docker (requires Docker)
npm run build:linux:bundle	Linux installers → dist-linux/ (.deb, AppImage, etc.)
npm run website:dev	Run the marketing site and docs (port 4321)
npm run website:build	Build the static website

Project layout

FlowCapture/
├── src/              # React app (Tauri webview)
├── src-tauri/        # Rust backend (recording, AI, exports, SQLite)
├── website/          # Marketing site + documentation
└── docs/             # Legacy planning notes (PRD, dogfooding)

Documentation

Full docs are hosted at flow-capture.vercel.app/docs:

• Installation
• User guide
• AI providers
• Development setup

Run the docs locally:

npm run website:dev
# http://localhost:4321/docs/intro

Contributing

Contributions are welcome. To get started:

1. Fork the repo and create a branch from main.
2. Make your changes and run npm run build and npm run test:rust.
3. Open a pull request with a clear description of what changed and why.

See the development setup guide for environment details.

Platform notes

Platform	Recording	Export extras
macOS	In-app video at ~15 fps	PDF via headless Chrome
Linux	ffmpeg MP4; PipeWire/X11 dependent	PDF via Chrome or wkhtmltopdf
Windows	ffmpeg MP4 via periodic frame capture	PDF via Chrome or wkhtmltopdf

All platforms store sessions locally in SQLite. PDF export uses headless Chrome or wkhtmltopdf when available on the system.

License

This project is licensed under the MIT License.

---

Website · Built with Tauri · React · Rust