Alter Zero CLI
Terminal coding agent in Rust: four permission modes, checkpoints, subagents, skills, hooks, MCP and local models
- 157 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- The agent runs commands and edits files, with no confirmation in master mode
- A built-in skill exposes a local service through an external tunnel
Install
Manual install
curl -fsSL https://raw.githubusercontent.com/linuztx/alter-zero/main/install.sh | shThe script installs alter-zero into ~/.local/bin and verifies the release SHA-256. The same binary installs with npm install -g @linuztx/alter-zero.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent runs commands and edits files, with no confirmation in master mode
- A built-in skill exposes a local service through an external tunnel
This is third-party code. Review the repository files before installing.
What it does
Alter Zero is an open-source coding agent written in Rust and modeled on Claude Code and Codex CLI. It works directly in the project: it reads files, edits them, runs commands and checks results while you follow along in the terminal. The interface shows numbered, syntax-highlighted files, readable diffs that mark changed characters, streaming command output and a task checklist. Control comes from the permission mode: manual asks before file changes and commands, edit allows file changes, auto (the default) adds a reviewer for routine commands and master runs unattended. There are checkpoints with rewind, a headless alter-zero -p mode for scripts and CI, and secrets stored through /secrets and supplied as placeholders so the agent never sees the value. Models connect by subscription, API key, locally through Ollama or LM Studio or through an OpenAI-compatible provider. Extensions: subagents with their own context, skills in SKILL.md folders, lifecycle hooks, MCP servers (stdio, HTTP, SSE, including OAuth) and project instructions in AGENTS.md. New hooks and MCP servers do not run until you confirm them. A single binary runs on Linux and macOS.
Who it is for. Developers on Linux and macOS who want a compact terminal agent with flexible permissions and local models.
Good fit when
- A coding agent is needed in the terminal with control over every command
- The agent is launched in a script or CI with one `alter-zero -p` command
- You must work on a local model through Ollama or LM Studio
- You need subagents, skills and hooks with no separate shell
Not a fit when
- You need Windows support: builds exist for Linux and macOS
- You need a plugin for Claude Code or Cursor: Alter Zero is a standalone program
- You want unattended mode on a sensitive repository: master runs commands without asking
Example request
Start alter-zero in this repository in edit mode, find the cause of the failing test, fix it and show the diff before finishingLimitations
Linux (x86_64, arm64) and macOS (Intel и чипы Apple). A modern terminal is required. master mode runs commands without confirmation, and the built-in flaredantic skill exposes a local service publicly through Cloudflare, Microsoft Dev Tunnels or Serveo, so check what you publish first. The repository includes a Kali Linux container image as an optional way to run. The project license is Apache-2.0, the version at the time of writing is 0.18.0, and changes between versions are recorded in CHANGELOG.md.
How to disable. Remove the alter-zero binary from ~/.local/bin or run npm uninstall -g @linuztx/alter-zero. Settings and sessions are kept per project directory.
FAQ
Which permission modes does Alter Zero have?
Four: manual asks before edits and commands, edit allows edits, auto (the default) adds a reviewer for routine commands and master runs unattended. Cycle them with Shift+Tab and keep rules per project.
How do I run it without a screen?
With alter-zero -p "prompt": the answer goes to stdout or JSON with --output-format, and the exit status can be checked in a script. Nothing waits for a reply, and --permission-mode sets what runs unasked.
Which models can I connect?
Subscriptions, API keys, local models through Ollama and LM Studio and any OpenAI-compatible provider. Every saved conversation resumes on the model it used.
Does the agent see my secrets?
Store a secret with /secrets and the agent uses its placeholder without seeing the value. The approach is described in docs/secrets.md.
Is there telemetry?
One anonymous ping is sent a day: a random install ID, version, OS, architecture and distribution. It is on by default and is turned off with a setting, an environment variable or the DO_NOT_TRACK convention.
Discussion
No questions or reviews yet.
Related
CLI toolsA Rust coding agent with a native GUI, a terminal mode, an ACP mode for Zed and an MCP server
A small terminal coding agent in Rust: bring your own API key or local model, per-turn cost, real undo and session branches
A Rust coding agent for the terminal with parallel agents, goals, Claude Code-format plugins and your own model endpoint.
A Rust coding agent in the terminal: extensions in any language over JSON-RPC, cloud and local models, audio, computer control
A Rust SDK for coding agents and the ready-made abstract CLI: REPL, 34 tools, MCP, graph memory, Anthropic and OpenAI, JSON output for CI
A Python terminal coding agent in the manner of Claude Code: reads the repository, edits files, runs commands, with memory, sub-agents and MCP
Comparison with similar tools
Cards that share the main search query, and how they differ.
README in short
The README presents Alter Zero as an open-source Rust coding agent, compares its capabilities in a table, shows script and npm installation and a Kali container image, then covers work in the terminal, permission modes, tools, session resume, model providers, commands and shortcuts.