Grok Build Dev Stack
IntermediateSpaceXAI's open-source terminal agent with parallel subagents, running Grok, Claude, or GPT.
Published 6 October 2026
About Grok Build Dev Stack
Grok Build is SpaceXAI's open-source coding agent for the terminal, a full-screen Rust interface that reads the repository, edits files, runs commands and background tasks, and proposes a plan for approval before it changes anything. Its defining habit is parallel subagents: the main agent hands research, implementation, testing, and review to child sessions with their own context, and any that edit files can work in a separate git worktree and merge back, while a dashboard follows several top-level sessions at once. It reads AGENTS.md, CLAUDE.md, hooks, skills, and MCP servers as they are, so a project set up for Claude Code moves over without rework.
Grok is the default model and the one SpaceXAI tunes the agent for, through a SuperGrok plan or a SpaceXAI API key: Grok 4.x for most work, or the coding-focused Grok Build 0.1 at a lower per-token price. Claude and GPT connect as custom models over the Anthropic and OpenAI protocols with their own API keys, and a local model served by Ollama works the same way, so the model can change mid-session and subagents can run a different one from the main session.
Most developers run Grok Build on their own computer, which has to stay on while it works; there are no SpaceXAI-hosted sessions for the agent and no official way to steer it from a phone. Headless mode covers scripts and CI, and the Agent Client Protocol puts it inside Zed, Neovim, or Emacs. For long runs with the laptop closed, it can run on a VPS or home server instead, signed in with a device code. A session keeper such as tmux or Zellij is worth adding there, so a dropped SSH connection doesn't end the session, and DigitalOcean offers a droplet with Grok Build preinstalled.
Code lives on GitHub by default, or on GitLab. A faster terminal, a session keeper for server use, a local model runtime, CI/CD, Docker, and an AI code reviewer are common additions; none of them change how Grok Build works.
Key Features
- ✓Open-source, Apache-2.0 terminal agent written in Rust, with plan mode and inline diffs
- ✓Parallel subagents with their own context, each optionally in a separate git worktree
- ✓A dashboard for watching and steering several sessions at once
- ✓Grok models on a SuperGrok plan or API key, plus Claude, GPT, or a local model through Ollama
- ✓Reads AGENTS.md, CLAUDE.md, hooks, skills, plugins, and MCP servers without changes
- ✓Headless mode for CI, and the Agent Client Protocol for Zed, Neovim, and Emacs
- ✓An optional sandbox using Landlock on Linux and Seatbelt on macOS
When to Use Grok Build Dev Stack
- →Splitting a large change across parallel subagents that each work in their own worktree
- →Moving a Claude Code project to Grok models without rewriting its CLAUDE.md, skills, or hooks
- →Running a coding agent on a SuperGrok subscription a developer already pays for
- →Comparing Grok, Claude, and GPT on the same task from one terminal agent
- →Running scripted agent jobs in CI on the low-cost Grok Build 0.1 model
Pros
- Parallel subagents and worktree isolation are built in rather than added through other tools
- Existing CLAUDE.md, AGENTS.md, skills, and hooks work as they are
- Open source under Apache 2.0, with other providers and local models supported in config
- Grok Build 0.1 is among the cheaper coding-tuned models per token
- Included with SuperGrok rather than sold as a separate product
Cons
- No hosted cloud sessions or official phone control; away-from-desk work runs on hardware you keep on
- Earlier versions uploaded whole repositories to SpaceXAI's cloud without clear consent, which still weighs on sensitive codebases
- A younger community and plugin ecosystem than Claude Code or Codex
- Claude and GPT bill per token on separate API keys; SuperGrok covers only Grok models
- Parallel subagents multiply token use, and prompts above 200K tokens bill at double for the whole request
LLM Options for Grok Build Dev Stack
Grok Build Dev Stack powered by Grok
Grok is the default here and the family Grok Build is tuned for. A SuperGrok plan from $30/mo covers it after a grok.com sign-in; on the API, Grok 4.7 costs $2 in and $6 out per million tokens with a 500K context, and Grok Build 0.1, built for coding, $1 and $2.
Grok Build Dev Stack powered by Claude
Claude connects as a custom model over the Anthropic Messages protocol, which Grok Build's model guide documents, with an Anthropic API key billed per token: Sonnet 5.5 at $2 in and $10 out per million, Opus 5.5 at $4 and $20. The pick for the hardest multi-file changes.
Grok Build Dev Stack powered by OpenAI
OpenAI's GPT models connect over the Chat Completions or Responses protocol with an OpenAI API key, billed per token: GPT-6.1 Sol at $2 in and $10 out per million for demanding work, or GPT-6 Luna at $0.10 and $0.50 for subagents doing high-volume, focused tasks.
These are highlighted picks. To see all the tools, check the LLM category.
Version Control Options for Grok Build Dev Stack
Grok Build Dev Stack with GitHub
GitHub is the default here, where most repositories and code reviewers already live. Grok Build only needs git, so its worktrees, branches, and commits push to GitHub like any other remote, and pull requests get reviewed with the tools the team already uses.
Grok Build Dev Stack with GitLab
GitLab bundles source control, merge requests, and CI/CD in one platform you can self-host, for teams whose code can't sit on GitHub. Grok Build works the same against a GitLab remote, since its worktree and branch handling is plain git.
Grok Build Dev Stack Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Server Add-ons
Add a server when you want the agent to keep running somewhere other than your own laptop, reachable at any time and not tied to your machine staying on. Some agents also offer their own managed cloud sessions as an alternative to self-hosting; check the stack's own description for details.
Grok Build Dev Stack with Hetzner
Hetzner is the default here: CX23 has 4 GB of RAM for about €5.49/mo, comfortable for Grok Build's Rust binary plus builds and tests, since the models run at SpaceXAI. Grok Build documents no memory minimum. A plain VPS billed hourly up to a monthly cap.
Grok Build Dev Stack with DigitalOcean
DigitalOcean offers a 1-Click Grok Build Droplet on Ubuntu 24.04 with the agent preinstalled; the 4 GB size is $24/mo, billed per second up to that cap. A device-code sign-in is all that stands between it and unattended sessions.
Grok Build Dev Stack with Hostinger
Hostinger's KVM 1 has 4 GB of RAM at $6.49/mo introductory, renewing at $11.99/mo, enough for Grok Build with the models served remotely. A budget pick on a plain Linux image; its agent templates target Claude Code and OpenClaw rather than Grok Build.
Grok Build Dev Stack with Raspberry Pi
A Raspberry Pi 5 is a one-time hardware cost instead of a monthly bill. Grok Build's installer ships native Linux aarch64 builds, and a 4 GB board runs the agent with the models served remotely, while 8 GB leaves room for subagents and tests.
These are highlighted picks. To see all the tools, check the Hosting & Cloud category.
Remote Access Add-ons
Add remote access when you want to reach an agent running on another machine — a VPS or a home server — without exposing it to anyone but you.
Grok Build Dev Stack with Tailscale
Add Tailscale alongside a server to SSH into it from a laptop or phone anywhere on a private address, with no port open to the internet, and reattach to the Grok Build session in tmux. The personal tier is free.
Session Persistence Add-ons
Add session persistence when the agent runs on a server or over SSH: a terminal multiplexer keeps the session working after the connection drops or the laptop closes, and you reattach from any machine to pick up where it left off.
Grok Build Dev Stack with tmux
Add tmux when Grok Build runs on a server: an SSH disconnect otherwise ends the session, while tmux keeps the agent and its subagents working and lets you reattach from any machine. Grok Build's terminal guide covers running inside it.
Grok Build Dev Stack with Zellij
Add Zellij for the same keep-running sessions with friendlier defaults: shortcuts stay on screen, and a built-in web client opens a running Grok Build session from a browser. Layouts come back after a reboot, though the running agent doesn't.
Terminal Add-ons
Add a terminal when you want a faster, more configurable place to run the agent than your OS default — most agent CLIs live here all day.
Grok Build Dev Stack with Ghostty
A GPU-accelerated terminal with native macOS and Linux integration, a solid default for Grok Build's full-screen interface, its mouse support, and inline diffs all day.
Grok Build Dev Stack with Alacritty
A minimal, GPU-accelerated terminal that does little beyond drawing text fast; pick it when tmux or a window manager already handles tabs and splits around Grok Build.
Grok Build Dev Stack with WezTerm
A GPU-accelerated terminal with tabs, splits, and multiplexing built in and configured in Lua, able to hold several Grok Build sessions side by side without tmux.
Grok Build Dev Stack with Kitty
A GPU-accelerated terminal with its own graphics protocol and scripting layer, for developers who want fine control over how Grok Build's full-screen view and its neighbors are laid out.
Grok Build Dev Stack with iTerm2
The long-standing free macOS terminal, with split panes, profiles, triggers, and tmux integration. The familiar pick for Mac users who want Grok Build in the terminal they already know.
Code Review Add-ons
Add code review when you want an AI reading every pull request before it merges: it comments on the diff so bugs, security issues, and inconsistencies surface before a human has to catch them.
Grok Build Dev Stack with CodeRabbit
Add CodeRabbit when pull requests from Grok Build should get a reviewer outside the agent: it installs as a GitHub or GitLab app, summarizes the diff, and leaves line-level comments, with a free tier for open-source repos. Useful once several subagents produce more changes than one person reads closely.
Grok Build Dev Stack with Greptile
Add Greptile when review should read the whole repository rather than only the diff: it indexes the codebase so its GitHub and GitLab comments carry wider context, and a self-hosted deployment keeps the code on your own infrastructure.
Grok Build Dev Stack with Cursor Bugbot
Add Cursor Bugbot when a bug-focused reviewer is enough: it reviews GitHub or GitLab pull requests whatever agent wrote them, though it bills through a Cursor plan rather than a subscription of its own.
Model Inference Add-ons
Add model inference when you want an open-weight model in the mix: on your own hardware for privacy, or on a hosted inference provider for speed and low per-token prices.
Grok Build Dev Stack with Ollama
Add Ollama to run Grok Build against an open model on your own machine: its model guide sets up a local Ollama endpoint as a custom model, and SpaceXAI describes a fully local-first setup. Expect weaker results than on Grok, especially for parallel subagents. The Local LLM Dev Stack covers the hardware and the models in full.
These are highlighted picks. To see all the tools, check the AI Runtime & Serving category.
CI/CD Add-ons
Add CI/CD when you want a dedicated pipeline for running tests, linting, or multi-stage builds before a deploy goes out. Many hosting platforms already redeploy automatically on every push on their own — a CI/CD tool adds the most value on top of that by gating the deploy on a passing test suite, and matters even more when the hosting choice does not auto-deploy at all, such as a self-hosted server.
Grok Build Dev Stack with GitHub Actions
Add GitHub Actions when every push should run the tests. Grok Build's headless mode fits a workflow step too, authenticated with a SpaceXAI API key from a secret, since a CI runner can't complete the browser sign-in.
Grok Build Dev Stack with GitLab CI/CD
Add GitLab CI/CD when the project lives on GitLab: pipelines defined in a .gitlab-ci.yml file run the tests on every push, on GitLab's runners or your own, with Grok Build's headless mode as one more job if you want it.
These are highlighted picks. To see all the tools, check the CI/CD Pipelines category.
Containerization Add-ons
Add containerization when you want the app packaged the same way across local development, staging, and production, or need to deploy somewhere that isn't a managed serverless platform.
Grok Build Dev Stack with Docker
Add Docker to give the project the same environment on your machine, a server, and in CI. Grok Build's own sandbox is off by default, so a container is also a simple way to fence in an agent left running with approvals switched off.
These are highlighted picks. To see all the tools, check the Containerization category.
Frequently Asked Questions about Grok Build Dev Stack
Can Grok Build keep working when my computer is off?
Not on its own. SpaceXAI hosts no sessions for the agent and offers no official way to steer it from a phone, so it runs on the computer you started it on, which has to stay awake. For work that continues with the laptop closed, Grok Build runs on a VPS or home server instead, signed in with a device code because a server has no browser. Started inside tmux or Zellij, the session survives a dropped SSH connection, and you reattach from any machine, a phone's SSH client included, over a private network such as Tailscale. DigitalOcean sells a droplet with Grok Build preinstalled.
Can I use Claude, GPT, or a local model instead of Grok?
Yes. Grok Build's model guide sets up Claude over the Anthropic protocol and GPT over OpenAI's Chat Completions or Responses protocol, each with its own API key billed per token; SuperGrok only covers Grok. A local model served by Ollama joins the same way, as does any OpenAI-compatible server. The model can change mid-session, and subagents can be given a different model from the main session, so a cheaper model can take research while a stronger one leads. Grok is still what the agent is tuned and tested on.
How does Grok Build compare to Claude Code?
They're close in shape, and Grok Build reads Claude Code's CLAUDE.md, skills, and hooks as they are. Claude Code is the more mature product, with hosted sessions on the web and Remote Control from a phone, and a larger ecosystem of plugins and guides. Grok Build is open source, builds parallel subagents with worktree isolation into the agent, and runs other providers or a local model, which Claude Code doesn't officially support. Claude Code starts with Claude Pro at $20 a month; Grok Build with SuperGrok at $30, or per token on the API.
Should I pay for SuperGrok or use the API?
SuperGrok suits interactive use in the terminal: one subscription covers Grok models within the plan's limits, and SuperGrok Plus and Heavy raise them. The API suits CI, headless jobs, and servers, where an API key is the documented sign-in, and irregular use, where per-token billing can come in under a monthly fee. Grok Build 0.1 keeps API costs low for routine coding. Watch two things on the API: parallel subagents each consume tokens, and a prompt at or above 200K tokens bills the whole request at double. The Pricing section has the figures.
Should subagents work in their own worktrees or the shared checkout?
It depends on whether they edit files. Grok Build's subagents share the working directory by default, which is fine for research, planning, and review, since those read without writing. Two subagents editing the same checkout at once can trip over each other, so give file-changing work worktree isolation: each child edits its own git worktree, and the changes merge back when it finishes. The cost is disk space and a merge step, and a conflict between two worktrees still has to be resolved by you or the agent. Grok Build reclaims old worktrees automatically.
Stacks Related to Grok Build Dev Stack
Cursor Dev Stack
Developer
Cursor as editor and coding agent in one, with Grok by default and Claude, GPT, or Gemini a click away.
Claude Code Dev Stack
DeveloperClaude Code in the terminal, desktop app, or web, paired with Claude and the add-ons to run it anywhere.
Codex Dev Stack
Developer
OpenAI Codex on your desktop, in your terminal, or in the cloud, steerable from your phone.
Conductor Dev Stack
Developer
Run Claude Code, Codex, and Cursor's agent in parallel on your Mac with Conductor, each in its own workspace.
Scores
Popularity2/5
A recognised name with an active open-source repository, but far fewer working users than Claude Code, Codex, or OpenCode, and some lasting distrust from the repository-upload incident.
Learning Curve3/5
Installing and signing in takes minutes, but subagents, worktree isolation, custom model entries, and a server setup with a session keeper add configuration beyond a first session.
Flexibility4/5
Grok, Claude, GPT, and local models through config, plugins, skills, hooks, MCP, and the Agent Client Protocol, though the agent is tuned for Grok and the open-source repository takes no outside contributions.
Performance4/5
A fast Rust client with parallel subagents and frontier Grok models behind it; results still track the chosen model, and nothing runs off hardware you keep on.
Portability4/5
Apache-2.0 licensed, runs on any machine or server, and can point at other providers or a local model, though hosted use and plan pricing assume a SpaceXAI account.
Tools in the Grok Build Dev Stack Stack
Development Tools
Add-ons (optional — add any, or none)
Server
Remote Access
Session Persistence
Terminal
Code Review
Model Inference
CI/CD
Containerization
Grok Build Dev Stack Pricing
Grok Build is free and open source; the cost is the model. SuperGrok at $30/month covers Grok models in the terminal, with Plus at $100 and Heavy at $300 for higher limits, or the API bills per token, from $1 per million input tokens on Grok Build 0.1. Claude and GPT bill per token on their own keys, and an Ollama model costs only hardware. GitHub or GitLab are free for individuals, and an optional server runs about €5.49–$24/month.
Apache-2.0 agent; hosted Grok use needs a SuperGrok plan or a SpaceXAI API key.
SuperGrok $30/mo; SuperGrok Plus $100/mo and Heavy $300/mo raise usage limits.
Grok Build 0.1 $1/M input, $2/M output; Grok 4.7 $2/$6 (double at 200K+ token prompts); Claude and GPT on their own API keys.
Free tier covers individual devs; GitHub Team is $4/mo, GitLab Premium is $29/mo for more seats.
Ollama is free; the cost is the hardware that runs the model.
Tailscale's personal tier is free.
4 GB covers the agent plus builds: about €5.49 at Hetzner, $6.49 intro at Hostinger ($11.99 renewal), $24 at DigitalOcean (1-Click Grok Build Droplet); or a Raspberry Pi as a one-time cost.