Blog
Tracer coding agentsTracer open sourceTracer free

Is Tracer Free? Open-Source Workspace for Coding Agents

Tracer is a free, open-source desktop app for running Claude Code, Codex and Cursor side by side on your own subscriptions. Here is what you get.

Portrait of Luis Chavez-Mattos
Edited by Luis Chavez-Mattos, Director, Product··RSS

For Engineering

Overhead flat-lay of an open laptop with three blank side-by-side panels, surrounded by a mouse, notebook, folder, pencils, earbuds and a coiled cable in a neat grid.

Is Tracer free?

Yes. Tracer is a free, open-source desktop app for macOS, Linux and Windows that gives your coding agents one shared workspace. It does not sell you a plan to use them. You bring your own subscriptions, so Claude Code, Codex, Cursor, OpenCode and other agents run side by side under the accounts you already pay for. The app is free, but the agents inside it still cost whatever your own plans cost. Tracer removes the extra layer, not the underlying bills.

TL;DR

  • Tracer is free and open source, and it runs on Mac, Linux and Windows.
  • You bring your own subscriptions, so you do not buy another plan to use Claude Code, Codex, Cursor or OpenCode in it.
  • Work is organized into tasks that hold files, artifacts, history and context, so you can switch models mid-chat without losing the thread.
  • Agents can ask each other questions, request reviews and hand off work without you copying messages between chats.
  • Mobile apps for Android and iOS let you continue the same work from your phone, while the agents keep running on your computer.
  • Model routing can keep a chat alive when a provider hits a rate limit, but it is off by default.
  • Long-running automations are planned, not released. There is a wait list.

What is Tracer, and who is it for?

Tracer is a desktop workspace for people who use AI coding agents. A coding agent is an AI tool that reads your project, writes and changes code, and runs commands, usually while you watch and steer. Claude Code, Codex, Cursor and OpenCode are examples.

Most people end up with several of these. One might be strong at planning, another at debugging, and a third is the one you still have credit on at the end of the month. The usual result is a pile of separate windows and chats, with you acting as the messenger between them.

Tracer puts them in one place and lets you direct the work from there. If you are a developer, a technical lead, or someone who builds software with agents regularly, this is aimed at you. If you do not write or review code at work, you will not need it, though you may work with people who do. The section on what this means for you covers that.

How much does Tracer cost, and what do you still pay for?

The app costs nothing to use. What you keep paying for is access to the models.

ItemCost with Tracer
The Tracer desktop appFree and open source
Claude Code, Codex, Cursor, OpenCode and othersYour existing subscription or account with each
An extra "workspace" planNone required

The practical benefit is that you do not add a subscription to organize subscriptions you already have. If you pay for two or three agents today, you keep paying for those and gain one place to run them. If you pay for none, Tracer does not give you free model access. You still need accounts with the agents you want to use.

How does the workspace work?

Everything lives inside a task. A task holds its files, artifacts, history and working context. Inside it, chats, files, artifacts, git diffs, pull requests, terminals and browsers appear as panels. You can keep several open side by side, so you see the chat and the result at once.

Because the context belongs to the task and not to a single chat, you can change models in the middle of a conversation without losing the thread. That is useful when one model gets stuck and you want a second opinion from another.

Can the agents work together?

Yes. One agent can ask another a question, request a review, or hand its work to the next agent. You do not copy text from one chat window into another. A realistic example: one agent drafts a change, a second agent reviews it, and you read the review instead of relaying it.

What are skills?

Tracer includes built-in skills, which are reusable workflows for planning, debugging, reviewing and writing docs. Think of them as saved procedures. Instead of re-explaining how you want a review done each time, you run the workflow.

Can teammates use the same workspace?

Tracer has multiplayer workspaces. Teammates can open your tasks and comment on artifacts in the same shared space. For a team, that means a reviewer can look at what the agent produced and leave feedback where the work lives, not in a separate chat thread.

How does the in-app browser help?

Tracer has a browser panel inside the task, next to your chats, terminals and files. You and your agents can both operate it.

  • Local dev servers show up automatically. The start page lists the ones it detects, so if your app is running you click it to open it.
  • You can point at the problem. Select the exact component that looks wrong, annotate it, and send that crop to the agent as context. This replaces long written descriptions like "the second button in the header on the right."
  • The agent can check its own work. It can navigate the page, inspect it, click elements and take screenshots to confirm a change works.
  • It hands control back to you when needed. If the agent reaches a login page, a captcha or an MFA prompt, you finish that step and it continues.
  • You can pop it out. A floating picture-in-picture window keeps your app visible while you work in other panels.

One limit: the local browser has dev tools, viewport presets and annotation. When the browser streams from a remote host to another desktop, you get navigation and the address bar, but not those local-only tools.

Can you use Tracer from your phone or another computer?

Tracer lets you use a laptop, desktop, dev box or VM as a remote host. The host holds the repo, the running agents, the terminals, the browser sessions and your credentials. Code and execution stay there. Nothing is copied to your phone. Your phone or second machine is a window into the host.

You register the host first, then connect from another device and continue the same tasks, chats, artifacts, terminals and browsers. Access is tied to your Tracer account, and you can remove registered hosts and devices, which helps if you lose one.

The mobile apps for Android and iOS go beyond a status dashboard. From the phone you can start a task (with an image attached if useful), answer an agent's question, approve a permission request, and send follow-up instructions. Notifications tell you when an agent needs you or finishes, and tapping one opens that task. Panels open full screen, and the terminal has an on-screen key bar with keys a phone keyboard lacks, such as escape, tab, control and the arrows.

Know the limits before you rely on it:

  • You need an active remote host, because the agents do not run on the phone.
  • The phone cannot act as a host or restart one.
  • Voice dictation is not in the mobile app yet.

What does model routing do when you hit a rate limit?

A rate limit is a cap a provider puts on how much you can use a model in a given window. Hitting one mid-task is annoying. Model routing lets you build a fallback plan in settings. The plan can:

  1. Try another account on the same provider.
  2. Switch to an equivalent model on another provider.
  3. Wait for your usage limit to reset.
  4. Notify you if nothing else works.

You choose the order, decide which models may replace one another, cancel a pending switch, and decide whether the chat returns to the original provider when it is available again.

Details worth knowing: automatic routing is off by default, so you must turn it on. It applies to chat agents on the selected host, not terminal agents. A switch starts a new agent session that uses the chat history, so the new model has the context. The wait option only works when the provider gives a reset time that falls within the maximum wait you set. It is most useful if you hold several subscriptions.

How do you customize Tracer?

Open settings, then appearance. You can create a theme, which copies your current one and lets you edit colors with a live preview, or import one. Imports accept community themes, Tracer theme files and VS Code theme files, so a VS Code theme you already like can come with you. You can set separate light and dark themes, use the built-in ones, or let Tracer follow your device's appearance. Themes can be exported to share.

What is coming next for Tracer?

The team is working toward long-running automations: agents that handle ongoing work and come back to you only when your input matters. The stated vision includes monitoring tools such as Sentry, PostHog and Datadog, workflows that review and improve themselves, cloud execution so a laptop does not need to stay open, team sharing with access controls, and sandboxes for the work. None of this is released. It is a stated direction, with a wait list for people who want in. Treat it as a plan, not a feature.

What does this mean for you?

If you already pay for more than one coding agent, try Tracer. It costs nothing, and the question it answers is whether one workspace beats several windows. Start with one real task, hand it between two agents, and see whether the shared context saves you time.

If you hit rate limits often, set up model routing deliberately. Turn it on, pick your fallback order, and decide which models you accept as substitutes. Remember the new model starts a fresh session fed by chat history.

If you lead a team, look at multiplayer workspaces. Reviewers commenting on artifacts inside the task is a different workflow from pasting screenshots into a chat.

If you are not a developer, the takeaway is the pattern. Agents are moving from single chats to coordinated workspaces where they hand work to each other and ask for approval. Expect the people around you to start working that way, and expect questions about permissions and review.

Is there a way to get a business tool without managing coding agents at all?

Tracer is for people who want to work directly with coding agents. If your goal is a finished internal tool, such as a request form, a tracker or a small dashboard, you may not want to run agents yourself. Remy takes a plain-English description of the app, writes a spec, builds the full-stack application with its own database and sign-in, and deploys it to a live URL. It suits a team replacing a spreadsheet or a subscription with software it owns. Remy costs $99 a month ($79 billed annually), with AI usage billed at cost on top.

Frequently Asked Questions

5 questions

Is Tracer really free, or is there a catch?

The desktop app is free and open source. The catch is that you still need your own accounts with the agents you want to run, such as Claude Code, Codex, Cursor or OpenCode. Tracer does not provide model access and does not require another plan.

Does Tracer work on Windows, Mac and Linux?

Yes. It is a desktop app for macOS, Linux and Windows. There are also mobile apps for Android and iOS, which connect to a host machine where the agents run.

Does my code go to my phone when I use the mobile app?

No. The host holds the repo, running agents, terminals, browser sessions and credentials, and execution stays there. Your phone is a window into that host, which is why the host machine must be active.

Can I switch models in the middle of a chat?

Yes. A task holds its files, artifacts, history and working context, so changing models does not lose the thread. Automatic switching on rate limits is a separate setting called model routing, and it is off until you enable it.

Are the automations available now?

No. Long-running automations are described as the direction Tracer is building toward. They are not out yet, and there is a wait list for anyone interested.

How we source, write, and correct our articles. Editorial standards

Related articles

Straight to your inbox.
New pieces as they publish. No noise.