Boxes.dev
Boxes.dev gives every Claude Code or Codex agent its own cloud Linux devbox — bring your own subscription, steer from Slack, Linear, or
If your parallel Claude Code or Codex agents keep stepping on each other's ports, env, and test databases, a full cloud computer per agent is a cleaner fix than another worktree. Boxes.dev is worth it once your dev environment lives in the cloud and you want runs to survive a closed laptop — Starter at $19/user/mo with 40 hours fits a side-project habit, Pro at $99/user/mo with 250 hours fits daily parallel work. The migration is agent-assisted but still an upfront project, and you keep paying
Verified 2d ago · liveness 74/100 · cite: rightaichoice.com/tools/boxes-dev
- Developers running multiple Claude Code or Codex agents in parallel who keep hitting port and resource collisions
- Small teams running agents every day who want work to continue overnight and on schedule
- Engineering orgs standardizing cloud coding agents with SSO, SCIM, RBAC, and audit logs
- Developers who want to review diffs, approve requests, or open a terminal from a phone
- Pure manual coding with no AI agents involved — you're paying for isolation you never use
- Single-agent local workflows where a plain worktree is cheaper and simpler
- Developers who rarely close the lid or need runs to survive disconnection
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Boxes.dev if you run a single coding agent locally and rarely close the lid — the one-time cloud migration and per-hour compute only pay for themselves once several agents are working in parallel.
Starter's 40 included hours go fast if you keep 4 awake devboxes running all month, and extra hours cost $0.75/hr as prepaid credit.
Trial ($0, 10 compute hours) fits a side-project developer testing parallelism; Starter ($19/user/mo, 40 hours) fits an individual with a few agents; Pro ($99/user/mo, 250 hours) fits daily multi-agent developers and small teams. Teams is custom-priced with volume hourly rates. Compare against a self-managed cloud VM (cheaper hourly but you write the setup script and own the isolation) or Devin and Cursor (agent harness included, but you give up your own Claude Code or Codex subscription).
In short
Boxes.dev — Boxes.dev gives every Claude Code or Codex agent its own cloud Linux devbox — bring your own subscription, steer from Slack, Linear, or. Best for Developers running multiple Claude Code or Codex agents in parallel who keep hitting port and resource collisions, Small teams running agents every day who want work to continue overnight and on schedule, Engineering orgs standardizing cloud coding agents with SSO, SCIM, RBAC, and audit logs. Free to start; paid plans from $19/user/mo.
What people actually say about Boxes.dev — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
9 mentions across 2 sources (Hacker News, Lemmy) · researched Jul 2, 2026.
Average across the 2 sources that answered — each source counts once, not each post.
- +Isolates each AI agent into its own cloud container, preventing conflicts.
- +Parallel agent execution without draining local machine resources.
- +Customizable templates for team onboarding and environment standardization.
- +Configurable CPU/RAM per box (default 4 vCPU, 8 GB).
- +Web-based terminal and IDE accessible from any browser.
- −Geographic access restrictions block some international users.
- −No option to self-host or bring your own cloud provider.
- −Still early stage with rough edges and potential unreliability.
- −Few alternative providers exist, but some users prefer open-source.
- −Lack of detailed pricing information creates buyer uncertainty.
- • Potential overage charges for CPU/RAM usage beyond default allocation
Viability Score
How well maintained and how widely used is Boxes.dev? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Cloud devboxes: isolated Linux machines with their own filesystem, ports, services, and database state
- Runs Claude Code and Codex agents on remote boxes using your existing subscription or API key
- Your coding agent migrates your local dev environment via a setup plan you review and approve
- Snapshot and restore of filesystem and memory for repeatable agent environments
- New devboxes start from a Template box snapshot in seconds
- Parallel agent execution across separate devboxes with no port juggling
- Agents keep running in the cloud after you close your laptop
- Idle devboxes sleep automatically; sleeping devboxes are free, compute billed by the second
- Start agents by mentioning @boxes.dev in any Slack channel or thread
- Assign a Linear issue to a Boxes.dev agent; it reports back and attaches PRs
- Scheduled (cron) agent runs for nightly QA and recurring commit reviews
- Webhook and REST API triggers at api.boxes.dev/v1/threads
- Persistent threads that keep work context alive across sessions and resume later
- Localhost port forwarding from the devbox to your machine
- Public preview URLs for any devbox port to share a running app before merge
About Boxes.dev
Boxes.dev runs Claude Code and Codex agents on cloud Linux machines instead of your local checkout. Each agent gets its own devbox with a separate filesystem, port forwarding to localhost, services, and database state, so you can run three tasks on three fresh boxes in parallel without port collisions or resource fights. You bring your existing Codex or Claude Code subscription or API key — Boxes.dev doesn't sell tokens and doesn't insert its own harness between you and the agent. Getting your environment to the cloud doesn't require writing a setup script. You paste one prompt into Claude Code or Codex in your project folder; your agent reads env files, local databases, and tools outside the repo, then drafts a setup plan you review. A second agent executes the approved plan and confirms the app runs remotely. Snapshot filesystem and memory, and new devboxes boot from that snapshot in seconds. Start work by mentioning @boxes.dev in any Slack channel or thread, assigning a Linear issue, running Codex or Claude on a cron schedule, or POSTing to api.boxes.dev/v1/threads. Threads persist, so you can pick one up later in the workspace from macOS desktop, iOS/Android, web, or CLI. Review diffs, open a web terminal, address PR feedback, and share a live session link with teammates who can watch the transcript and screenshots without pulling a branch. Every devbox port can get a public preview URL. Standard boxes are 4 vCPU / 8 GB RAM / 100 GB disk, with larger machines available. Idle boxes sleep automatically and sleeping boxes are free; compute is billed by the second. Compared with Devin or Cursor's cloud agents, the difference is that Boxes.dev keeps your existing harness and subscription rather than swapping in its own, and the whole product is built around isolation per body of work. It launched publicly on Hacker News on 2026-06-04.
Behind the Verdict
The pitch that actually matters here isn't "cloud IDE" — it's isolation. A git worktree gives you another checkout on the same machine, sharing processes, ports, and database state. A devbox gives you another computer entirely, created from a Template box snapshot, so one agent can rewrite a schema or break a dependency without touching the next agent's world. If you've ever had two agents fight over port 3000 at 1am, you already understand the value. Where it earns its keep: overnight and unattended work. Agents run in the cloud, so closing the laptop or going to bed doesn't stop the thread. Add cron runs for nightly QA or recurring commit reviews and the marginal cost stays bounded because sleeping devboxes are free and compute bills by the second — but note the tiers cap how many boxes can be awake at once: 2 on Trial, 4 on Starter, 10 on Pro. The Slack and Linear triggers are the sleeper feature for teams. Mention @boxes.dev in a thread and the agent gets the context and reports back in-thread; assign a Linear issue and it replies there with PRs attached. That's a workflow change, not a novelty — it moves agent kickoff out of your terminal and into the systems your team already reads. When to pass: if you run one agent with the lid open, the isolation overhead buys you nothing. Pure manual coding doesn't need it either, and anyone allergic to terminals, VMs, and CLIs will bounce off the migration flow. Cost is another honesty check — Boxes.dev is the infrastructure, not the intelligence. You still pay OpenAI or Anthropic for the Codex or Claude subscription or API key that drives the agent, and every hour past your tier allotment costs $0.75/hr on Starter or $0.60/hr on Pro. Closest alternatives: Cursor's cloud agents and Devin both run agents remotely, but
Researching Boxes.dev? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Boxes.dev actually fits — and what changes day-one when you adopt it.
They paste the Boxes.dev setup prompt into Codex in their project folder, approve the generated setup plan, and snapshot the resulting devbox. They then open three boxes — one for a flaky auth test, one for a dependency bump, one for a UI refactor — and mention @boxes.dev in a Slack channel to kick off the first.
Outcome: The three agents never contend for ports or the local database, localhost port forwarding follows whichever fork they want to inspect, and the snapshot means every new box starts in seconds instead of re-running setup.
On Pro they schedule Codex and Claude on cron to run end-to-end nightly QA and review past commits, with agent responses posted to a Slack channel. In the morning, each engineer opens the thread on the desktop app or phone and reads the diffs.
Outcome: Recurring checks happen without anyone staying late, results land where the team already reads them, and per-second billing plus free sleeping boxes keeps the overnight run cheap.
They put Boxes.dev on the Teams plan with custom seat and hour pools, SSO, SCIM, RBAC, audit logs, and spend controls, and wire the REST endpoint at api.boxes.dev/v1/threads into their CI so an agent starts on a fresh devbox per pipeline event.
Outcome: Agent runs are tied to identity and budget, isolation is per task rather than per laptop, and the security review is covered by a custom MSA instead of ad-hoc exceptions.
Use Cases
- Run several Codex or Claude Code agents in parallel against the same repo without file or port conflicts
- Put two agents to work overnight on a long refactor and review the diffs in the morning
- Trigger nightly QA or recurring commit reviews on a cron schedule and post results to Slack
- Assign a Linear issue to a Boxes.dev agent and get the PR attached back in the issue
- Kick off an agent with a single API call from CI, a webhook, or a custom tool
- Keep localhost port forwarding pointed at whichever agent's build you want to inspect
- Roll back a botched refactor by restoring the devbox snapshot
- Review diffs and approve agent requests from an iPhone or Android phone
Models Under the Hood
as of 2026-10-11
Limitations
- Boxes.dev only pays off if your dev environment lives in the cloud; that migration is a real one-time project, and it's pointless for a single local agent.
- You supply the Codex or Claude Code subscription or API — Boxes.dev doesn't sell tokens.
- Compute is metered: Starter at $19/user/mo allows a template box plus 4 awake devboxes with 40 hours included, Pro at $99/user/mo allows 10 awake devboxes with 250 hours; extra hours cost $0.75/hr on Starter and $0.60/hr on Pro, sold as prepaid credit.
- Some capabilities are tier-gated: scheduling, localhost port forwarding, and Slack and Linear triggers sit on Pro and above, and team controls (SSO, SCIM, RBAC, audit logs, spend controls) are custom-priced.
- It is not a production host, and it is not affiliated with or endorsed by Anthropic or OpenAI.
as of 2026-10-09
Verification history
We have re-verified Boxes.dev 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Boxes.dev tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Trial
$0
Ideal for
A developer with a side project who wants to try running a couple of Codex or Claude Code agents in the cloud before paying anything.
What this tier adds
Free entry point: 10 hours of compute, a template box plus 2 awake devboxes, up to 2 projects, unlimited free sleeping devboxes.
Starter
$19/user/mo
Ideal for
An individual developer on a side project with a few agents occasionally running in parallel.
What this tier adds
Adds paid compute at $19/user/mo with 40 hours included, 4 awake devboxes, up to 5 projects, extra hours at $0.75/hr, and email support.
Pro
$99/user/mo
Ideal for
Developers and small teams running multiple agents in parallel every day, including from Slack and Linear.
What this tier adds
Raises to $99/user/mo with 250 hours, 10 awake devboxes, up to 10 projects, extra hours at $0.60/hr, plus Slack and Linear triggers, scheduled and webhook runs, localhost port forwarding across forks, and priority support.
Teams
Custom volume
Ideal for
Engineering organizations rolling out cloud coding agents across a team that needs SSO, RBAC, and audit logs before adopting.
What this tier adds
Custom seat and hour pools at volume hourly rates, with SSO, SCIM, RBAC, audit logs and spend controls, dedicated resources and deployment, and a custom MSA and security review.
Where the pricing makes sense
The company stage and team size where Boxes.dev's pricing actually pencils out — and where peers do it cheaper.
Trial ($0, 10 compute hours) fits a side-project developer testing parallelism; Starter ($19/user/mo, 40 hours) fits an individual with a few agents; Pro ($99/user/mo, 250 hours) fits daily multi-agent developers and small teams. Teams is custom-priced with volume hourly rates. Compare against a self-managed cloud VM (cheaper hourly but you write the setup script and own the isolation) or Devin and Cursor (agent harness included, but you give up your own Claude Code or Codex subscription).
Setup time & first value
How long it actually takes to get something useful out of Boxes.dev — broken out by persona, not the marketing-page minute.
Individual developer: roughly an afternoon the first time — paste the setup prompt into Codex or Claude Code in your project folder, review the setup plan with the agent, approve it, then snapshot; afterward a new devbox starts from that snapshot in seconds. Small team: about a day to migrate the shared environment and connect Slack and Linear on Pro. Engineering org: days to a couple of weeks,
Switching to or from Boxes.dev
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From a local Claude Code or Codex worktree: paste the Boxes.dev setup prompt in your project folder, have the agent draft a setup plan covering env files, local databases, and tools outside the repo, approve it, then
- →From a self-managed cloud VM: reuse the installed packages from your setup script, but let a Boxes.dev agent build and verify the remote box, then snapshot it so later devboxes start in seconds.
- →From Cursor or Devin: keep your existing Claude Code or Codex subscription instead of the bundled harness, and move the environment rather than rewriting prompts.
- →From GitHub repo only: point Boxes.dev at the repo, let the setup plan cover the tools your app needs, and confirm the app runs remotely before snapshotting.
- →From CI-only agents: replace the pipeline step with a POST to api.boxes.dev/v1/threads so each event gets a fresh devbox and a persistent thread you can resume.
- ↗To a self-managed cloud VM: export the packages and env config from your snapshot plan and rebuild them as a setup script you own, accepting that you also take back port and resource contention between agents.
- ↗To Cursor or Devin: move the prompts and repo across, but you give up bringing your own Claude Code or Codex subscription and inherit their harness.
- ↗To local worktrees: acceptable only if you drop back to one agent at a time, since worktrees isolate files but not ports, databases, or background services.
- ↗To GitHub Actions or another CI runner: fine for scheduled and triggered runs, not for persistent threads you review diffs in over days.
Integrations
Resources & Guides
- Guideboxes.dev
Guides · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
How To Run Claude Code In The Cloud · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
How To Run Codex In The Cloud · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
Run Coding Agents In Parallel · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
Keep Coding Agents Running After You Close Your Laptop · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
Run Claude Code And Codex From Your Phone · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
Cloud Development Environments For Ai Coding Agents · Boxes.dev
In-depth how-to from boxes.dev
- Guideboxes.dev
Best Cloud Coding Agents In 2026 · Boxes.dev
In-depth how-to from boxes.dev
Tutorials & Learning
YouTube returned 6 videos for “Boxes.dev”, and we withheld 6: 6 did not mention Boxes.dev. We are showing none, because we could not prove any of them are about Boxes.dev.
Official links
Tools that pair well with Boxes.dev
Common stack mates teams adopt alongside Boxes.dev, with the specific reason each pairing earns its keep.
Avibe
Local-first Agent OS that runs official Claude Code, Codex, and OpenCode CLIs behind one shared Workbench, Harness, and run graph.
Fly.io
Fly.io runs AI agent sandboxes on hardware-isolated Linux VMs called Sprites that checkpoint automatically and bill by the second.
Hostinger
Agentic web hosting where AI Builder generates your site and Hostinger Agent manages it — included on every plan.
Featured Head-to-Head Comparisons
Boxes Dev vs Spider Cloud
If your core need is fetching and structuring web data for AI agents or RAG, Spider Cloud is the clear winner with its Rust engine, AI extraction, and open-source flexibility. If you instead need isolated cloud environments to run multiple AI coding agents (like Claude Code) concurrently, Boxes.dev is purpose-built for that. These tools are complementary, not directly competitive.
Boxes Dev vs Temporal Ai
Temporal AI is the right choice for teams needing battle-tested orchestration with reliability guarantees, while Boxes.dev is purpose-built for developers who need isolated, parallel environments for AI coding agents. Choose Temporal if your workflows require durable execution, retries, and human-in-the-loop; choose Boxes.dev if your primary need is running multiple LLM-based coding agents in safe, sandboxed containers.
Boxes Dev vs Presto Voice
These tools serve completely different markets. Presto Voice is purpose-built for QSR drive-thru automation with proven revenue lift, while Boxes.dev is a developer tool for running AI coding agents in isolated cloud containers. Choose Presto if you operate multi-location fast-food drive-thrus and want to increase order accuracy and upsell revenue. Choose Boxes.dev if you are a developer needing parallel, sandboxed environments for multiple AI coding agents.
Alternatives to Boxes.dev
View all
Frequently Asked Questions
Categories
Used Boxes.dev? Help shape our editorial sentiment research.