Open source project
Open Interpreter
Runs model-authored code locally to operate the machine — a code-first rather than pixel-first approach. The original repository has been repurposed; the Python project continues as a community-maintained fork.
Open Interpreter predates the current wave and takes the other road: rather than reading the screen and moving a pointer, it has the model write code and shell commands — Python, JavaScript, Bash, PowerShell — and runs them locally. There is also an experimental OS-control mode. Where it gets complicated is provenance. The openinterpreter/open-interpreter repository is now a Rust harness based on OpenAI's Codex, and the classic Python project lives on as a community-maintained fork at endolith/open-interpreter, which is what this entry tracks.
Best for
Tasks expressible as code on a machine you control, where running a script beats clicking through a GUI.
Watch out
Generated code touches your files and system settings. Code is held for approval by default, but -y and auto_run bypass that, and the fork's own README calls the classic branch a vibe-coded mess and says to use it with caution. Note also the licence: the community fork is AGPL-3.0, not the Apache-2.0 carried by the repository that took over the original name.