LocalClicky icon

LocalClicky

LocalClicky is a macOS voice assistant that controls apps, files, reminders, and browser actions locally by voice, with wake-word listening, local transcription, Ollama reasoning, and optional screen vision.

LocalClicky

LocalClicky at a glance

LocalClicky is a macOS voice assistant built to control the Mac locally from the menubar. Its README describes an offline design that keeps transcription, reasoning, screen analysis, and speech output on the user’s machine rather than sending voice or screenshots to cloud services.

The project is designed around wake-word sessions and natural language commands. After a trigger phrase starts recording, the assistant can process requests, call tools, inspect the screen when needed, and continue the conversation until you end the session or it times out from inactivity.

Core capabilities

Menubar-first interface

Runs as a menubar app on macOS so it stays out of the Dock while remaining available for voice control.

Session-based voice control

Uses a wake phrase to begin a session, then stays active until you end it or silence expires, so follow-up commands do not require repeating the trigger phrase.

Local model pipeline

Combines offline transcription, local AI reasoning, and optional screen vision to interpret spoken requests and act on them.

Mac automation actions

Supports system actions such as app control, volume changes, file operations, shell commands, Spotify control, reminders, and Chrome JavaScript injection.

Screen-aware interaction

Takes screenshots on demand and uses the vision model to identify targets for clicking or describing what is on screen.

Automatic recording stop

Uses voice activity detection and a hard cap fallback to stop recordings when possible, reducing the need for manual cutoffs.

Practical use cases

  • Hands-free Mac control

    Use LocalClicky when you want to open apps, change volume, manage files, or launch browser actions by voice while keeping the workflow local to the machine.

  • Screen-guided interaction

    Use the assistant to ask what is on your screen, find a target element, and click it when a command depends on current UI state.

  • Everyday command execution

    Use natural language requests for reminders, Spotify control, and shell commands when you want a mixed personal-assistant and automation workflow.

  • Multi-step voice sessions

    Use the session mode for back-to-back follow-up instructions, such as refining a task after the first command runs, without repeating the wake phrase.

  • Local AI assistant prototype

    Use the repository as a reference implementation for an offline, local-first voice assistant stack that combines wake-word detection, transcription, model inference, and automation.

Pros and Cons

Pros

  • Runs locally on the user’s machine with no cloud APIs, API keys, or subscriptions mentioned in the README.
  • Combines wake-word detection, transcription, tool use, and screen vision in a single voice-driven workflow.
  • Supports a broad set of Mac tasks, from system control and file handling to reminders, Spotify, and browser actions.
  • Includes session memory and multi-round tool calling for follow-up commands and iterative requests.
  • The README provides concrete installation steps, model names, and permission requirements, which helps with reproducible setup.

Cons

  • The README depends on several local components, including Whisper, Ollama, and optional silence detection, so setup is more involved than a single-app install.
  • Screen interaction and automation require macOS permissions for microphone, screen recording, and accessibility access.
  • The documentation is strongest on setup and capabilities; it gives less detail about collaborative workflows or long-term project limits.

FAQ

What do I need to run LocalClicky?

LocalClicky is set up to run on macOS with a local Whisper transcription backend, Ollama for reasoning and vision, and macOS built-in text-to-speech. The README lists additional macOS permissions for microphone, screen recording, and accessibility access for the Python binary in its virtual environment.

How does the voice session workflow work?

The README describes LocalClicky as a menubar app that starts recording after the wake phrase, stops automatically on silence when voice activity detection is available, and can continue through a session without repeating the wake word.

What kinds of actions can LocalClicky handle?

It can open and quit apps, adjust volume, control Spotify, manage files, run shell commands, inject JavaScript into Chrome, create reminders, and click or identify on-screen UI elements using a screenshot and vision model.

Does LocalClicky keep conversation history?

The project README says the assistant keeps the last 10 exchanges as conversation memory during a session.

Is LocalClicky free to use?

The repository is presented as a local, offline project with no cloud APIs, no API keys, and no subscriptions mentioned in the README. The GitHub pricing page confirms GitHub itself offers a free plan and paid plans, but the LocalClicky repository does not describe a separate product price.

Quick Facts

Category
Developer Tool
Platform
macOS
Primary use
Offline voice control for Mac automation
Source domain
github.com
Runtime
Python with local Whisper and Ollama components
Pricing
No separate product pricing is described in the repository