Skip to content
TrustList
CC
Application Development Software

Caveman Code

A Mac and browser app that runs Claude Agent, Codex, Cursor and its own Caveman agent side by side, with a supervising agent, pre-PR guides and cloud computers, free on a user's own keys or on credit plans from $20 a month.

  • Artificial Intelligence - AI
  • DevOps

TrustList summary

Not yet independently verified. Dated from the Y Combinator launch post of 13 November 2026; the maker's site was last updated on 12 November and gives no launch date. The cost comparisons with Codex and Claude Code are the maker's own benchmark runs, and the site names no city or country. We will update this when it can be confirmed, and remove this note.

Developers who keep three coding agents open in three terminals are the audience for Caveman Code. The app, from the Y Combinator-backed Caveman Labs, puts Claude Agent, Codex, Cursor and the company's own Caveman agent in one window, in tabs and splits, and adds a supervising agent it calls the Chief, which starts chats, answers their questions, reports back and carries on by itself when the developer does not reply in time.

Built around review

Before a pull request, the app writes a guide with a one-line summary, the decision the developer has to make, and each change set out as before, after and why. Every chat gets its own worktree, agents can message each other and share memory, and plugins from Claude Code, Codex and Cursor work inside it. In beta, agents can drive a Linux desktop and browser on cloud computers and keep working while the Mac's lid is closed. A terminal version is listed as coming soon.

Open-source parts and the cloud

The desktop app is proprietary. The Caveman command-line tool, which wraps Claude Code, Codex and Gemini CLI on the developer's own machine, is open source under Apache-2.0, as are the agent, SDK and skill; a memory package, cavemem, is MIT. Caveman Cloud runs and monitors agents, with spend, keys, policies and traces: free for one person with monthly allowances, $300 a month for a team with $300 of usage included after a seven-day trial, and an enterprise edition that installs in a customer's own AWS or GCP account with a Helm chart.

Credits rather than seats

The app is free when it runs on a user's own subscriptions and API keys. Paid plans buy monthly credits spent at each model's list price: Pro at $20 or $60, and Max at $100 (with $105 of credit) or $200 (with $220), with yearly billing 20% cheaper. A credit goes three times as far on GLM-5.3 and twice as far on DeepSeek Flash and GLM-5.3 Flash. No week, and no five-hour window, may use more than 35% of a month's credits, and top-ups of $5 to $500 last a year.

Key facts

Maker: Caveman Labs, Inc. Launched: Y Combinator launch post, 13 November 2026 Platforms: Mac app and browser Runs: Claude Agent, Codex, Cursor and the Caveman agent Pricing: free on own keys; credit plans $20, $60, $100 and $200 a month; Cloud Team $300 a month (site, read 13 November 2026) Backed by: Y Combinator

Categories & features

  • Artificial Intelligence - AI
  • DevOps

Request a demo or quote from Caveman Code

Protected by reCAPTCHA - Google Privacy Policy and Terms apply.

By sending, you agree we may share your request and contact details with the provider once you confirm your email.

Reviews

Write the first review of Caveman Code

Used it? Your experience helps other buyers decide.

Write a review

Questions & answers

No questions yet. Be the first to ask about Caveman Code.

Caveman Code alternatives

Similar listings buyers compare against this one.

Cloud sandboxes for AI agents, billed by the minute at low unit rates, for running code, agent harnesses, browsers and desktops, with a separate sandbox for each customer of the product built on it.

MIT-licensed runtime of single-request serverless actors, each with its own SQLite database and WebSocket server, written in TypeScript on a Rust runtime; run on Terse Cloud or self-hosted. Billed by the second for CPU and memory requested.

An AI coding agent for C, C++ and CUDA that works from the real build and call graph, takes bugs from Slack, Jira or the command line, and returns a pull request with build, test, sanitizer and performance results attached.

A Bolt.new agent that runs on open-source models and gives more usage in return for training data

Analytics tool showing how people use Model Context Protocol servers after installing them

Cloud Macs and Linux machines matching Bitrise CI, for developers and coding agents, started in seconds and archived later