An open-source project called OpenClicky has been published on GitHub as an alternative to Clicky, a GPT Realtime-powered companion app. The codebase, maintained by developer Jason Kneen, is publicly available and has already drawn interest from the open-source and Mac automation communities.
OpenClicky is positioned as a software platform for computer use, voice interaction, and browser control on macOS. The repository shows a fast-moving project with hundreds of commits, multiple branches and tags, and a steady stream of updates focused on reliability, routing, and user experience. The latest changes highlight a continuing push to refine how the app handles realtime voice, background computer-use tasks, and browser workflows.
Based on the repository documentation and commit history, OpenClicky combines several automation layers. It includes a local computer-use lane, browser workspace tools, overlay and cursor features, and integrations for voice response and text-to-speech. The project also exposes APIs and runtime components for handling agent instructions, screen capture, keyboard input, and web navigation.
Recent commits suggest the app is designed to manage multiple types of user requests, including voice-driven commands, browser tasks, and UI interactions. Some updates focus on making the system more resilient when the runtime is not immediately ready. Others refine how the app routes different kinds of instructions, such as reflective conversation versus action-oriented agent tasks.
The repository also shows work on OpenClicky widgets and a companion manager, which appear to support the app’s interface and control flows. The project’s evolving documentation references computer-use behavior, browser agent controls, and guidance for keeping certain actions in a background lane unless a foreground handoff is explicitly needed.
Many of the most recent commits are aimed at making the app more dependable. Changes include watchdogs for realtime voice handling, deduplication logic to prevent repeated agent starts, and safeguards to recover from states where processing appears stuck. Other updates improve runtime readiness checks before actions such as screen capture, typing, and key presses are triggered.
The project has also added protections around sensitive settings. According to the commit log, some secrets have been moved to the Keychain, and certain bridge or proxy features are now disabled by default or gated behind explicit user configuration. That suggests the developer is tightening control over risky behavior while preserving backward compatibility where possible.
A separate set of changes focuses on the UI. The repository history references updated captions for voice responses, revised status wording, clearer overlays, and visual polish for widgets and other interface elements. There are also signs of work on audio buffering and playback timing to reduce stutter during voice output.
OpenClicky’s release matters because it provides a visible open-source counterpart to a commercial or proprietary-style realtime assistant experience. By publishing the code on GitHub, Kneen makes the project available for inspection, modification, and community contribution. The repository has already accumulated stars and forks, indicating that other developers are exploring it.
The source material does not indicate a formal launch announcement or pricing changes for Clicky itself. What is clear is that OpenClicky is being developed as a standalone alternative, not just a minor patch or plugin. Its codebase, documentation, and commit history show an active effort to build a broader platform around voice, computer use, and browser automation.
For now, OpenClicky appears to be a work in progress with frequent updates. But its public release gives developers another option in the growing field of AI-driven desktop automation tools, especially for those who want open code, local control, and a closer look at how the system works.