Build and review from your phone

Your dev
environment.
Pocket-sized.

Ask for a change. Try the app. Point to what needs fixing—all from your phone. Your agent runs on your own Mac, Linux computer, or server.

10 free sessions every week. No credit card required.
Bring your machine and coding agent. Connect them after signup.

BRING YOUR AGENT

Claude Code Codex Pi OpenCode

A Bivy session with the Chat / Terminal toggle, agent response, and message composer. — Preview · Try the running app.
Actual Bivy UI · this website in an example session

Bivy’s live fullscreen preview displaying the landing page being built. — Preview · Try the running app.
Actual Bivy UI · this website in an example session

YOUR WORK DOESN’T STOP HEREPhone ↔ Browser ↔ Terminal↓

01 / FROM REQUEST TO REVIEW

See what your
agent built.

An agent says the change is done. Open the app and try it. If something’s off, point to it and ask for a fix in the same session.

INSIDE BIVY / THIS WEBSITE

A website change, a live preview, and a specific request for a fix. Here’s what the feedback loop looks like in Bivy.

01 / ASK

Start with a change.

Tell your agent what you want to build. It works in your repo, with your existing tools and local services.

Bivy’s chat with an agent response and a composer for the next request. — Chat with your agent in the same session.

02 / TRY

Open what it built.

Try the running app on your phone. Tap through the page and see what needs another pass.

The Bivy website running inside Bivy’s live app preview. — This website, open in the live preview.

03 / REFINE

Draw it. Add a note.

Circle what needs changing on the preview. Add a note, then send the marked-up screenshot and your instructions to the agent.

Bivy’s Draw tool: a pink pen circle around the pricing heading, with a written note asking the agent to make Every week stand out, ready to add to chat. — Actual Bivy UI · example pen annotation and note, ready to add to chat.

You decide when it’s ready. Try the app again, inspect the diff, and review the checks before you merge.

Explore previews and visual feedback ↗

02 / WHEN YOU LEAVE YOUR DESK

Pick it up
on your phone.

You’ve started a task, but you need to head out. Keep the machine running and open Bivy on your phone. Answer the agent’s question or review its changes without restarting the conversation.

⌘AT YOUR DESK

Give your agent a task.

Use the repo and agent you normally work with.

↗ON YOUR PHONE

Unblock the next step.

Answer a question, approve an action, or open the terminal.

✓BEFORE YOU MERGE

Check the work yourself.

Try the app, inspect the diff, and check the test results. Review the branch and pull request before you merge.

03 / BEYOND ONE MACHINE

Your agents. Your machines.
One workspace.

Claude Code on your Mac. Codex on a server. Keep your sessions together and choose the setup that fits each task.

01 / CREDENTIALS

Less signing in.

Reuse supported keys and OAuth logins across compatible agents and machines with end-to-end encrypted sync. Keep work and personal credentials separate.

See what syncs and which logins stay local ↗

02 / MACHINES

The right setup for the task.

Your laptop for a quick fix. An always-on server for a longer job. Choose a machine and agent, then check in from any device. The machine running the work must stay awake and connected.

03 / HANDOFFS

Carry the work forward.

Fork or move a session with its code and conversation context. History transfer depends on the agents involved; review the new session before retiring the original.

How handoffs work ↗

WHEN AN AGENT HITS ITS LIMIT

Hit a limit? Keep the task moving.

A provider limit doesn’t have to end the task. Choose how to continue.

01 / KEEP GOING

Switch to another agent.

Fork the session to an agent with available usage and carry the work forward.

02 / PICK IT UP LATER

Retry when the limit resets.

Ask Bivy to retry automatically when your provider’s usage becomes available again.

Automatic retry requires a reset time from the provider. Switching agents doesn’t reset the original provider’s limit.

Actual Bivy UI · two ways to continue at a usage limit.

04 / AUTOMATE THE HANDOFF

Let an issue
start the task.

Pick it up on your phone when it needs you. A GitHub issue, failed CI check, Slack or Linear message, schedule, or signed webhook can start your agent on the machine you choose. Open the same session to add context, try the app, and review the result.

GitHubSlackLinearSchedulesWebhooks

OPEN THE SESSION ANYTIMEThe agent starts on your machine.

Read its progress and send a follow-up.

ChangesChecksPull request ↗

05 / BUILT AROUND YOUR SETUP

Keep your tools.
And your options.

No new dev environment to set up.

Connect the Mac, Linux computer, or server where you already work. Bivy uses your existing agent subscriptions, API keys, or local models.

The relay can’t read your session.

Interactive traffic is end-to-end encrypted between your machine and paired devices. Your model provider still receives what the agent sends it.

Use Cloud or host it yourself.

Bivy is open source. Cloud handles the app and relay for you; self-hosting puts those on your infrastructure too. Slack commands and generic webhook instructions arrive at the routing service as plaintext.

Read the security model ↗

Pricing

10 free sessions.
Every week.

You get 10 new sessions each week on Cloud, including automations. Upgrade when you need more. Reopening a session doesn’t use another slot.

You supply the machine and pay your model provider as usual. Local sessions are always free.

Managed hosting

Cloud

Free

10 new sessions every week.

Need more? $15/mo for unlimited sessions with support.

Phone and browser access to your sessions
Live previews, terminal, and automations
Multiple machines, agents, and credential sync
App and relay hosting managed for you
All integrations included on the free plan too

Start with 10 free sessions a week. No credit card required.

Open source

Self-hosted

Run the app server and relay yourself.

AGPL-3.0 open source
Every feature, unlimited
Your infrastructure and operations
Community support

Self-host guide

FAQ

Before you
get started.

What is Bivy?

Bivy is an open-source remote development environment built around coding agents. From your phone or browser, work with an agent, use the terminal, preview running apps, send visual feedback, and review changes on machines you control. Bivy Cloud hosts the app and relay.

Does my computer need to stay on?

Yes. The machine running your agent and dev server must stay awake and connected. You can use an always-on Mac, Linux computer, or server to keep working when your laptop is closed.

Do I have to use Bivy Cloud?

No. Bivy Cloud hosts the app and relay for you. Bivy Core is open source under AGPL-3.0, so you can self-host the app server and relay if you prefer.

Which agents are supported?

Pi, Claude Code, Codex, and OpenCode are Supported. More agents are available as Beta or Experimental adapters; the runtime matrix has the details.

Can Bivy Cloud read my code or prompts?

Not for interactive remote sessions: prompts, transcripts, diffs, files, and model keys are end-to-end encrypted between your machine and devices. Slack commands and generic webhook instructions are plaintext in the hosted routing service.

Does Bivy sandbox every agent?

No. Some agents have native sandboxing or structured approvals; others run largely as you. Bivy does not ship a universal OS-level jail.

Try it

Try your next task
from your phone.

Create an account or sign in to Bivy. Choose Connect a Machine and copy your personal connection command. Run it on your Mac, Linux computer, or server, then pick a repo and start a session.

10 free sessions every week. No credit card required.

Your dev environment. Pocket-sized.

See what your agent built.