Analysis updated 2026-07-03 · repo last pushed 2026-07-02
SSH into a production server and have the CLI investigate and fix a bug without leaving your terminal.
Ask the CLI to read relevant files and draft a code fix, then apply it with your approval.
Start a coding session in the terminal and export it to the desktop app for a richer visual interface.
Use the CLI on a remote cloud machine where a full graphical application is not available.
| google-antigravity/antigravity-cli | kaelio/ktx | vibeforge1111/keep-codex-fast | |
|---|---|---|---|
| Stars | 1,445 | 1,450 | 1,456 |
| Language | — | TypeScript | Python |
| Last pushed | 2026-07-02 | 2026-07-03 | 2026-05-06 |
| Maintenance | Active | Active | Maintained |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | data | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires installing the CLI and configuring permissions for an AI agent that can read files and execute commands.
Antigravity CLI is a terminal-based AI coding assistant that understands your codebase, makes edits with your permission, and runs commands directly from the command line. Think of it as having a smart pair-programmer who lives in your terminal, can read your entire project, and can actually do work for you instead of just suggesting snippets. At a high level, the tool brings the same "agent engine" that powers Google's full Antigravity 2.0 desktop application into a lightweight, keyboard-driven terminal interface. That engine supports multi-step reasoning (the AI can plan out a sequence of actions), multi-file editing (it can change several files at once to accomplish a goal), tool calling (it can run commands and use external tools), and persistent history (it remembers what happened in previous sessions). You stay in control, it asks for your permission before making changes or executing commands. The primary audience is developers who live in the terminal, especially those working over SSH on remote servers. If you are connected to a cloud machine and cannot easily use a full graphical application, this tool is designed to give you the same AI assistance you would have locally. It is built for speed and low resource usage, so it runs well even in constrained environments. A practical example: you SSH into a production server, ask the CLI to investigate a bug by reading relevant files, and then have it draft and apply a fix, all without leaving your terminal session. One notable design choice is that the CLI and the full GUI application share the same underlying engine, settings, and permissions. That means if you start a session in the terminal, you can export it to the desktop app and keep going with a richer visual interface. Preferences sync both ways, so you do not have to configure things twice. The README does flag important caveats: AI coding agents carry security risks like autonomous code execution, data exfiltration, and prompt injection. You should monitor and verify everything the agent does.
A terminal-based AI coding assistant that reads your codebase, makes edits with your permission, and runs commands from the command line. It is lightweight, keyboard-driven, and works well over SSH on remote servers.
Active — commit in last 30 days (last push 2026-07-02).
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.