@lxfater: Microsoft just open-sourced a project that can turn screen recordings into skills. You might think it's like RPA—it records your actions and replays them exactly. But actually, it analyzes "what you want to do" and finds the best way to turn that into a skill. For example, if it sees you using software to compress a video, it might consider using ffmpeg...

X AI KOLs Timeline 08/03/26, 03:30 AM Tools

open-source microsoft ai-agent automation screen-recording copilot github

Summary

Microsoft open-sourced Skill Recorder, a tool that records the screen and uses GitHub Copilot to analyze the user's intent and action steps, converting them into reusable Skills or automated workflows for AI agents. Unlike traditional RPA, it's smarter and tends to use native tools rather than replaying click operations.

Microsoft just open-sourced a project that can turn screen recordings into skills. You might think it's like RPA—it records your actions and replays them exactly. But actually, it analyzes "what you want to do" and finds the best way to turn that into a skill. For example, if it sees you using software to compress video, it might consider using ffmpeg to compress it. Rigid process-based RPA is a thing of the past—give this a try.

Original Article

View Cached Full Text

Cached at: 08/03/26, 05:48 PM

Microsoft just open-sourced a project that can turn screen recordings into skills. You might assume it works like RPA — recording your operations and replaying them verbatim. But in reality, it analyzes “what you want to do” and then finds the best way to turn that into a skill. For example: if it notices you’re using software to compress a video, it might consider using ffmpeg to do the compression instead. Rigid, flow-based RPA is a thing of the past — give this a try.

microsoft/skill-recorder

Source: https://github.com/microsoft/skill-recorder

Skill Recorder

Record yourself doing a task once, then turn it into a skill your AI agent can repeat.

Skill Recorder captures a real work session on your screen: the clicks, the app and window switches, the pages you visit, and (if you want) your spoken narration. It then uses the GitHub Copilot CLI to reconstruct what you actually did as a clear intent plus an ordered list of steps. From there, one step turns that single run into something an agent can reuse:

a Skill: a SKILL.md procedure an agent runs on demand, or
an Automation: the same procedure on a schedule or trigger.

Both prefer the agent’s native tools (like the gh CLI or web_fetch) over replaying UI clicks, and generalize from your one example, so recording yourself submitting one form can teach the agent to submit all of them.

How it works

🔴 Record. Hit record (or ⌘⇧R / Ctrl+Shift+R from anywhere) and just do your task. Skill Recorder captures your screen and activity locally, in the background.
🎛️ Control. While recording, a small always-on-top bar shows capture and microphone state. Mute, unmute, or switch mics on the fly, then finish, or discard (with a confirmation) if the take didn’t go to plan.
🧠 Analyze. Click Analyze and GitHub Copilot reconstructs one overall intent and an ordered list of steps. Review and edit until it reads right.
✨ Create. From an approved analysis, generate a reusable Skill and/or a scheduled Automation.

Get started

Skill Recorder is published as a source release: one command downloads a pinned Node.js runtime, builds the exact release commit on your machine, and adds a Skill Recorder (Source) app you can relaunch anytime. Nothing is installed globally. You’ll need a GitHub account with Copilot access; the Copilot CLI ships with the app.

macOS is the primary target. Windows 11 (x64 and ARM64) is supported too (see WINDOWS-VALIDATION.md).

Install it

Open the latest release (https://github.com/microsoft/skill-recorder/releases/latest) and copy the command for your platform. Each release pins an exact commit, so the real command looks like the patterns below with <40-character-release-commit> filled in.

macOS / Ubuntu

commit="<40-character-release-commit>"; curl -fsSL "https://raw.githubusercontent.com/microsoft/skill-recorder/$commit/install.sh" | SKILL_RECORDER_COMMIT="$commit" bash

The commit pins both the downloaded script and the source it builds. To keep the app running after the terminal closes, add SKILL_RECORDER_DETACHED=1 after the pipe:

commit="<40-character-release-commit>"; curl -fsSL "https://raw.githubusercontent.com/microsoft/skill-recorder/$commit/install.sh" | SKILL_RECORDER_COMMIT="$commit" SKILL_RECORDER_DETACHED=1 bash

On macOS this adds a Skill Recorder (Source) app to ~/Applications (relaunch from Spotlight, Launchpad, or the Dock). On Ubuntu it adds a matching application entry.

Windows (PowerShell)

$commit="<40-character-release-commit>"; $env:SKILL_RECORDER_COMMIT=$commit; irm "https://raw.githubusercontent.com/microsoft/skill-recorder/$commit/install.ps1" | iex

This adds Skill Recorder (Source) shortcuts to your desktop and Start Menu.

Then record

Grant Screen Recording. On first launch, macOS asks for Screen Recording permission; grant it and you’re ready to record.
Record, Analyze, Create. Do your task, then Analyze. The first time you Analyze, Skill Recorder offers Sign in to Copilot if you aren’t signed in yet.

To inspect the script before running it, set install options, update, or uninstall, see INSTALL.md.

⚠️ Keep secrets out of your recordings. Don’t record, type, paste, or narrate passwords, tokens, API keys, or other confidential info. Choosing Analyze sends recording data to GitHub’s cloud. Skill Recorder reminds you before every recording. Details in What gets captured.

Everything below is for people who want the details, or want to hack on the code.

What gets captured

Recording, storage, frame extraction, and optional narration transcription all happen on your computer; nothing leaves while you record. Only when you choose Analyze does Skill Recorder send the event timeline (window/document titles, URLs, and clipboard previews), extracted screen images, and narration text to GitHub’s cloud for Copilot to process.

The in-app “Records your screen and activity” panel spells out exactly what’s collected:

Window tracking: active-app / window switches.
Browser URLs: the page you’re on (macOS).
Screen video: recorded by Chromium; low-rate snapshots are kept only when the screen changes or a heartbeat is due.
Clipboard: short previews of copied text that tie steps together.
Narration (optional): spoken commentary, transcribed on-device in any of Whisper’s 99 supported languages (a one-time ~252 MB model download on first use).

⚠️ Please don’t capture secrets. Passwords, access tokens, API keys, credentials, and other confidential information should never be recorded, typed, pasted, shown, copied, or narrated during a session.

Develop from source

Requires Node.js 24. After checking out a release revision:

npm ci
npm run compliance:licenses
npm run dev

npm run dev starts Vite and launches the Electron app with hot-reload; ⌘⇧R (macOS) / Ctrl+Shift+R (Windows) toggles recording from anywhere.

Full manual setup, the build and dist scripts, and the licensing boundary between local source builds and redistributable packages are in INSTALL.md. Maintainers changing versions, dependencies, assets, or releases must follow RELEASING.md.

Evals

The Copilot describer and builders have a fixture-based eval suite; see evals/README.md.

npm run eval          # score the describer against synthetic recordings
npm run eval:builder  # score the skill/automation generalization

Documentation

INSTALL.md: install options, inspect-first install, updating, uninstalling, and manual developer setup.
RELEASING.md: maintainer release runbook.
evals/README.md: the describer / builder eval harness.
THIRD-PARTY-NOTICES.md: licenses for bundled dependencies.

Security

Please don’t report security vulnerabilities through public GitHub issues. See SECURITY.md for Microsoft’s coordinated-disclosure process and reporting channels.

Support

File bugs and feature requests through GitHub Issues (https://github.com/microsoft/skill-recorder/issues) (search existing issues first to avoid duplicates). Support is limited to the resources described in SUPPORT.md.

License

MIT

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit Contributor License Agreements (https://cla.opensource.microsoft.com).

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct (https://opensource.microsoft.com/codeofconduct/). For more information see the Code of Conduct FAQ (https://opensource.microsoft.com/codeofconduct/faq/) or contact [email protected] with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft’s Trademark & Brand Guidelines (https://www.microsoft.com/legal/intellectualproperty/trademarks/usage/general). Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party’s policies.