Catenary Review — The Spatial IDE for AI Coding Agents
CodingFreemiumA local-first spatial IDE that orchestrates several AI coding agents with visual wires, parallel worktrees, and a Monaco editor, all without an account or telemetry.
What Is Catenary?
It is a spatial IDE for AI coding agents. Rather than one editor with one assistant, it puts several agents, terminals, Monaco editors, browsers, and isolated git worktrees on a shared canvas so you can watch them work at once. Visual wires let you push context from one panel to an agent, or delegate work between agents.
The tool is described as local-first from the first commit. There is no account, no telemetry, and the app lists exactly the two hosts it ever contacts, so your code and keys stay on your machine.
How It Works
The tool orchestrates the agent CLIs you already run. It wires in Claude Code, Codex, Cursor, Grok, OpenCode, and Pi, and those installed CLIs talk to Anthropic, OpenAI, or Google with the key you bring, or to nothing at all when pointed at a local Ollama or LM Studio endpoint. The app itself does not host or run the model.
A New Task button creates an isolated branch, its own working directory, and an agent without touching the code in front of you, and each task gets a colour that follows it through the sidebar, dock tabs, and canvas territory.
- Orchestrates Claude Code, Codex, Cursor, Grok, OpenCode, and Pi
- Agent CLIs do the model work with your keys
- 1-Click worktrees create isolated branches and agents
- Each task keeps a colour identity across the UI
- Run multiple projects with parallel branches per project
Maestro Mode
Maestro Mode is a human-approved team setup where one recruiting agent briefs and wires a team of up to five helpers, adjustable to ten. It is off by default, and every action opens an editable approval card before anything reaches the canvas. The team organizes itself while you stay the one who approves.
Larger plans include built-in squads such as Bug Whisperer, Git Wizard, Refactor Gremlin, and Code Reviewer, and Pro unlocks your own squads and higher helper counts.
Privacy and Local-First Design
The privacy story is its strongest feature. The analytics endpoint ships empty, the crash-reporter DSN is removed from the distribution, and the entire outbound traffic list is two hosts: a GitHub Releases version check every 15 minutes and a license endpoint that only answers when you paste a key or claim a trial.
The Pro key is verified on device by an Ed25519 signature and works offline, and the app never proxies source code, terminal output, project names, prompts, or API keys through itself. What an agent sends to your chosen model provider is between that provider and you.
- No telemetry and no analytics endpoint in published builds
- Outbound traffic limited to a version check and license activation
- Source code, prompts, and keys never leave your machine
- Pro key verified offline by Ed25519 signature
- Agent model traffic goes direct to your provider or local model
Who Should Use It?
Catenary is for developers already running agent CLIs who are tired of juggling several windows and losing sight of parallel work. It shines when you want two or three agents shipping experiments in isolated worktrees at once, or when you want a standing way to hand context from one agent to another.
If you only run one assistant in one editor, a simpler tool may be enough. The setup also assumes you are comfortable installing a not-yet-code-signed desktop app and pointing it at your own agents.
Top Alternatives
Terminal-focused agents like [Claude Code](/tools/claude-code) and editors such as [Cursor](/tools/cursor), [Windsurf](/tools/windsurf), and [Codeium](/tools/codeium) are the tools this app actually orchestrates. [GitHub Copilot](/tools/github-copilot) and [Tabnine](/tools/tabnine) cover single-editor assistance.
The difference is orchestration: where those tools help inside one editor, this is a workspace that coordinates several agents and worktrees at the same time. If you rely on one agent rather than a team of them, you may not need the canvas at all.
Pricing
Free plan with no card or account. Pro is a US$19 lifetime license (standard price US$29) with a 7-day free trial, covering unlimited projects and up to two computers.
Free
Canvas, terminals, editor, browser, and git with unlimited wires and two parallel tasks. No account, no card, no telemetry.
- Canvas, terminals, editor, browser, git
- Unlimited connection wires
- One project, two parallel tasks
- Maestro Mode with 2 helpers
Pro
Launch price lifetime license (standard US$29). Unlimited projects and parallel tasks, all squads, and up to two computers.
- Unlimited projects and parallel tasks
- Maestro Mode with 2-10 helpers
- All built-in squads and your own squads
- Up to two personal computers, 1.x updates included
Best For
Recommended use cases and scenarios where Catenary shines.
Pros and Cons
The strengths of Catenary are its local-first privacy posture, parallel worktrees and agents on one canvas, visual context wires, and a lifetime license rather than a subscription. For teams running several agents the spatial layout genuinely reduces context switching.
The trade-offs are that installers are not code-signed yet, the agent CLIs rather than the app do the model work, and the canvas is overkill if you run a single assistant. It is a smaller, single-creator project with a younger ecosystem.
Pros
- Local-first architecture with no account, no telemetry, no cloud proxy
- Runs multiple agents and parallel worktrees on one canvas
- Visual wires move context between agents and panels
- Monaco editor with git-aware side-by-side diffs
- Lifetime Pro license instead of a subscription
Cons
- Installers are not code-signed yet, so first run triggers a system warning
- The agent CLIs, not Catenary, carry out the actual model work
- Heavier setup for developers not already using agent CLIs
- Single-creator project with a smaller ecosystem
Frequently Asked Questions
Common questions about Catenary, answered.
What is Catenary?
It is a local-first spatial IDE that orchestrates several AI coding agents with visual wires, parallel worktrees, and a Monaco editor, without an account or telemetry.
How much does Catenary cost?
Free is unlimited in time with no card. Pro is a US$19 lifetime launch license (US$29 later) with a 7-day free trial, covering unlimited projects and up to two computers.
Does Catenary run the AI models itself?
No. It orchestrates the agent CLIs you install. Those CLIs talk to Anthropic, OpenAI, or Google with your key, or to a local model on Ollama or LM Studio.
Which agents does Catenary support?
The tool wires in Claude Code, Codex, Cursor, Grok, OpenCode, and Pi, and lets you point any of them at the model provider you choose.
Is Catenary private?
Yes. It is local-first with no telemetry, and its entire outbound traffic is a version check and license activation. Code, prompts, and keys never leave your machine.
Does Catenary work on my operating system?
Yes. There are downloads for Windows, macOS (Apple Silicon and Intel), and Linux, with the installer not yet code-signed on first run.
Reviews & Ratings
4.8
Based on 71 reviews
Loading reviews...
Priya Sharma
Fast, intuitive, and the results speak for themselves. Easily worth the subscription.
James Okafor
Great value for the price. The learning curve is small and the payoff is big.
Hannah Lee
Reliable and polished. I only wish the advanced features were on lower tiers.
Similar Tools
More Coding tools you might like
Claude Code
Anthropic's agentic coding tool that lives in your terminal — plan, build, test, and ship software by describing tasks in plain English.
GitHub Copilot
AI coding assistant that suggests code completions and entire functions in VS Code, JetBrains, and Neovim.
Amazon Q Developer
Amazon Q Developer is AWS's AI coding assistant for code completion, agents, and security scans across your IDE.