explaingit

raynan00/clacky

Analysis updated 2026-05-18

56PythonAudience · generalComplexity · 3/5LicenseSetup · moderate

TLDR

An open source voice assistant for Windows that sees your screen, points at things, and completes tasks on your PC using Claude.

Mindmap

mindmap
  root((Clacky))
    What it does
      Sees your screen
      Points at UI elements
      Acts on your PC
    Tech stack
      Python
      Claude
      Deepgram
    Use cases
      Ask about your screen
      Run multi step tasks
      Background research agent
    Audience
      Windows users
      AI assistant fans

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Ask what's on your screen and get a spoken answer with a buddy pointing at the relevant element.

USE CASE 2

Tell Clacky to open apps, click buttons, and complete multi-step tasks on your Windows PC by voice.

USE CASE 3

Have a background agent research a topic while you keep talking, then hand you back real files.

What is it built with?

PythonClaudeDeepgramPyInstaller

How does it compare?

raynan00/clackyeadmin2/jarvis_aigreatvishal27-rc/ai-resume_analyzer
Stars565656
LanguagePythonPythonPython
Setup difficultymoderatehardmoderate
Complexity3/54/52/5
Audiencegeneraldevelopervibe coder

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Requires an Anthropic API key and ideally a Deepgram key, collected by a first-run setup wizard.

Free to use, modify, and distribute, including commercially, as long as you keep the copyright notice.

In plain English

Clacky is an open source voice companion app for Windows that watches your screen, points at things on it, and can carry out actions on your computer when you ask it to. It is inspired by a macOS tool called Clicky and aims to bring the same idea to Windows, since the original and its closed source agent mode are Mac only. To use it, a person holds a hotkey and talks. Clacky can answer questions about what is currently on screen, moving an on screen buddy character to point at the exact thing being discussed. It can also act directly, opening applications, clicking buttons, typing text, and completing multi step tasks using Claude's computer use feature. Clacky remembers information across sessions, such as stated preferences, and can learn routines that a user teaches it by voice, saving them as editable text files following an open standard called Agent Skills. For longer jobs, Clacky can run a background agent that keeps working on a task, such as research, while the user keeps talking, and it leaves behind real files rather than just a spoken summary. These background agents can also connect to outside services such as Notion or other tools through the Model Context Protocol, so results can land directly in a user's other apps. A separate, simpler feature called the file organizer tidies a folder using a single call to a language model, and every change it makes can be undone. Clacky's brain is Anthropic's Claude model, using Sonnet 5 and its computer use capability, while speech recognition runs through Deepgram with a free text to speech voice, or a local fallback if no Deepgram key is provided. Getting started requires an Anthropic API key, and ideally a Deepgram key, both collected by a first run setup wizard. It can be installed as a downloadable Windows executable or run from source with Python 3.11 or newer. The README describes it as an early, rough build, and it stops before irreversible actions like sending a message or making a purchase. The project is released under the MIT License.

Copy-paste prompts

Prompt 1
Show me how to install Clacky on Windows using the downloadable exe and set up my Anthropic and Deepgram keys.
Prompt 2
Explain how Clacky's voice agent decides when to act directly versus stop and ask for permission.
Prompt 3
Walk me through teaching Clacky a new routine by voice and saving it as a SKILL.md file.
Prompt 4
How do I run Clacky's file organizer to tidy a folder and undo the changes if needed?

Frequently asked questions

What is clacky?

An open source voice assistant for Windows that sees your screen, points at things, and completes tasks on your PC using Claude.

What language is clacky written in?

Mainly Python. The stack also includes Python, Claude, Deepgram.

What license does clacky use?

Free to use, modify, and distribute, including commercially, as long as you keep the copyright notice.

How hard is clacky to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is clacky for?

Mainly general.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.