Use an AI agent in your terminal to read, understand, and edit your codebase.
Run the agent headlessly in CI or scripts instead of interactively.
Embed the agent inside an editor through the Agent Client Protocol.
Automate multi-step coding tasks like refactors or file edits from the command line.
| xai-org/grok-build | neondatabase/neon | eza-community/eza | |
|---|---|---|---|
| Stars | 21,724 | 21,744 | 21,668 |
| Language | Rust | Rust | Rust |
| Setup difficulty | moderate | hard | easy |
| Complexity | 4/5 | 4/5 | 1/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Prebuilt binaries install in minutes, building from source needs a pinned Rust toolchain plus the DotSlash tool.
Grok Build, known by its command grok, is a terminal based AI coding agent made by SpaceXAI. It runs as a full screen, mouse interactive terminal application that can read and understand your codebase, edit files, run shell commands, search the web, and keep track of tasks that take a while to finish. You can use it interactively in your terminal, run it without a screen for scripting and CI pipelines, or plug it into an editor through something called the Agent Client Protocol. Prebuilt binaries are available for macOS, Linux, and Windows through a single install script or PowerShell command, after which the tool is available as the grok command. On first launch it opens your browser so you can authenticate. If you would rather build it yourself, the project is written in Rust, with the toolchain version pinned so rustup installs the right one automatically. Building also requires a tool called DotSlash, which lets the build hermetically download and run helper tools such as protoc for generating code from protocol buffer definitions. The codebase is organized into many small Rust crates under a crates folder, covering things like the terminal interface itself, the underlying agent runtime, individual tool implementations for editing files and running commands, and workspace management for the filesystem and version control. There is also a third party folder containing vendored code, including a Mermaid diagram rendering stack. The root Cargo.toml file that ties the workspace together is generated automatically, so contributors are expected to edit the per crate configuration files instead. Full documentation, including a getting started guide, keyboard shortcuts, slash commands, and information on plugins, hooks, and sandboxing, lives on x.ai's docs site and alongside the source in the repository. The project does not accept outside contributions. The first party code here is released under the Apache License 2.0, while various vendored and third party pieces, including source ports of tools from other open source coding agent projects, keep their own original licenses, listed in separate notice files.
A terminal-based AI coding agent from SpaceXAI that reads your codebase, edits files, runs commands, and searches the web from an interactive full-screen interface.
Mainly Rust. The stack also includes Rust, DotSlash, protoc.
Apache License 2.0 lets you reuse, modify, and redistribute the first-party code freely, including commercially, as long as you keep attribution and note changes.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.