Head to head · Code sandboxes · October 2026 research run

boat vs Offscreen

boat scores 71.8 (BB) on agent readiness against Offscreen's 67.1 (B), and leads in 6 of 7 scored categories. Offscreen leads on payments & pricing. Both do code sandboxes.

Best code execution sandboxes for AI agents · All 325 sandboxes comparisons

Which one, for what

boat BB

Best for Long-lived, full Ubuntu VMs with Docker, a desktop and Chrome for coding agents, evals and products that give each user a sandbox, hosted in the EU.

Ahead on

  • Reliability, 60 against 55
  • Schema & documentation, 90 against 67
  • Agent ergonomics, 93 against 74
  • Security & auth, 67 against 62
  • Maintenance & community, 76 against 66
  • Transparency & trust, 69 against 61

Also in its favour

  • Agent-ready, a grade of BB or better

Watch for

No security.txt or bug bounty, and SOC 2 is in progress

Offscreen B

Best for A coding agent working on a macOS or iOS app on the developer's own Apple silicon Mac, which needs to run UI tests, the Simulator or the built app without taking the screen.

Ahead on

  • Payments & pricing, 90 against 45

Also in its favour

  • No key needed to call it
  • Free to start without a card
  • Open source

Watch for

First commit 11 October 2026, version 0.0.5 and labelled beta, with JSON field names that may change before 1.0

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Offscreen +45

boat 45

No x402, MPP or L402 (0). Per-hour prices for each size published without login, from $0.018 an hour for small to $0.20 for xlarge, with overage and start prices (30). The 7-day trial (25 compute hours) needs a card in Stripe Checkout (0). An agent can register itself through the auth.md agent registration protocol or AgentID, but a person has to claim the agent or verify the AgentMail organisation and enter a card before a sandbox can start (15 of 30). 0+30+0+15 = 45.

Offscreen 90

No payment protocol (0). Read with the self-hosted rule. Free under MIT with nothing to buy from the author (20). No card or account (20). Installs with one Homebrew command and needs no signup, though a person grants Local Network on the first VM job (20). Tart, which the cask installs, is under FSL-1.1-ALv2, and its licensing page says organisations above 100 host CPU cores need a paid licence. That is a cost of the dependency, named in the pricing notes, and not scored here. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 60 is 90.

Schema & documentation boat +23

boat 90

A public OpenAPI 3.1 file, docs.boat.dev/openapi/boat-v1.yaml, with 123 operations, each with an operationId and a summary (25). llms.txt indexing every page and a Markdown copy of each (10). The API overview and guides say when to create, resume or fork, when to run a command detached, when to use a saved snapshot, and what each sandbox state allows (18). 86 enums and required fields in the OpenAPI file, and list calls reject unknown query parameters with 400 invalid_query (14). curl, Python and TypeScript examples, a table of error codes by status with what to do, and a retryable flag on 502s (15). Paths versioned under /v1, but no changelog was found (8 of 15). 25+10+18+14+15+8 = 90.

Offscreen 67

Graded on the CLI and its skill. status and image ls take --format json, and --progress json emits build and run phases, but the docs say field names may change before 1.0, no JSON Schema is published, and run output is the wrapped command's own. The project contract is validated, with unknown keys rejected, by offscreen check (8 of 25). The skill is Markdown in the Agent Skills format, with a 90-word description and a 1,552-word SKILL.md, and the docs are Markdown in the repository (10). The skill says when to use it, sets a core rule never to run xcodebuild test, the Simulator or built apps on the host, says plain builds may run on the host, and lists what to ask the person before running (18). Flags are typed by kong, with value sets for --lane (vm, host, auto) and --progress, durations for --timeout, and every task requires lane and run. Task commands are free shell text by design (11). A worked example app with three tasks and their scripts, an exit-code table with what to do for each, and troubleshooting entries that quote the error text (14). Tags and a daemon protocol number with stated compatibility rules. No changelog file, and the GitHub release notes couldn't be read (6 of 15).

Agent ergonomics boat +19

boat 93

Lists take limit from 1 to 200, commands can stream or detach, and a file read above 5 MiB returns 413 with a pointer to a signed download URL (20). Cursor pagination with hasMore and filters by name, name prefix, label and state (20). A stable error envelope with code, message, details and requestId, and a per-status table of what to do (20). Idempotency-Key on create and fork, kept 24 hours, with 409 idempotency_in_progress and 409 idempotency_key_reused, and per-code guidance on whether a command may have run (18). ttlSeconds defaults to one hour, the body of a create can be empty, and SDKs exist in Python and TypeScript beside the CLI with --json on every command (15). 20+20+20+18+15 = 93.

Offscreen 74

A CLI adds no tool definitions to context, and only the skill's 90-word description is always loaded. Stdout carries the wrapped command's full output, which for xcodebuild can be long, while Offscreen's phases go to stderr and are hidden when stderr isn't a terminal (18 of 25). status -a, -q, --format json and templates, logs with -f, and artifacts in one directory per job. No control over the size of the command's output (13 of 20). Exit codes separate the command's result from a queue timeout (75), an Offscreen failure (70) and a usage error (1), errors name the fix (Open it with: open -a Offscreen, the Tart install command, the Local Network setting), and a failure before the command starts repeats the last 20 lines (18). Each job starts from a fresh clone, so a rerun is clean. cancel is safe to repeat, init refuses to overwrite, check validates without a VM, and prune asks first or takes --dry-run. Attached queued jobs fail after a daemon restart, and the docs say to detach jobs that must retry safely (16 of 20). Defaults are sensible (VM lane, two slots, no timeout) and a one-off -- <cmd> needs no task. One CLI, and the daemon's socket API is internal, with no SDK (9 of 15).

Maintenance & community boat +10

boat 76

Python SDK boat-sdk 1.13.0 on 28 October 2026 (30). 25 SDK releases from 1.0.0 on 16 September to 1.13.0 on 28 October 2026 on PyPI (20). Closed service with a Discord, a support page and a blog, but no public changelog or tracker (8 of 25). Official SDKs in Python and TypeScript (@boatdev/sdk 1.13.0) generated from the OpenAPI file (15). No public source repository or CI for the SDKs was found (3 of 10). 30+20+8+15+3 = 76.

Offscreen 66

v0.0.5 on 16 October 2026, 17 days before this check and still the head of main (30). Five tagged releases on 15 and 16 October 2026, each published to the cask (20). All 116 commits are by the one maintainer, and pull requests #89 to #91 of 15 October are his own. Outside issues and replies couldn't be read, because GitHub's pages and API were refused to us (8 of 25, from the commit history only). Not an MCP server, no SDK, and the skill is in no registry (0). Current dependencies (Go 1.27, kong 1.14, modernc SQLite 1.60), tool versions pinned in mise.toml and lint in CI. No Dependabot configuration found (8).

Transparency & trust boat +8

boat 69

Closed service under terms of 12 October 2026 from Dedale AI, Corp., a Delaware corporation (15). The terms, privacy policy and docs agree on retention. A stopped sandbox's disk is kept at least 14 days, saved snapshots have no scheduled expiry, and zero data retention can be turned on. The terms say the cleanup job does not yet delete in production. A DPA is available on request (25 of 30). The Box to boat rename on 15 September 2026 put Deprecation and Sunset headers on old surfaces with a dated sunset of 31 October 2026, and deprecated endpoints are marked in the reference (18 of 20). Infrastructure hosts (OVH, Hetzner) and other processors are named, sandboxes run in Germany, Finland and France, and snapshot copies can be stored outside the EU (16 of 20). 15+25+18+16 = 74.

Offscreen 61

MIT in LICENSE and in the release archive (30). No privacy policy, as for other local tools. DAEMON.md and the config reference say what is stored where, that the caller's environment is kept only while a job is active, that deletion is logical and SQLite free pages and backups can keep old values, and that job history expires after seven days by default (22 of 30). No deprecation policy. The README's beta note, the protocol compatibility rules and the install guide's migration steps for the renamed app and daemon identifiers are the only notices, with no dates (8 of 20). No telemetry found in the source, but the README and docs don't say either way (12 of 20).

Reliability boat +5

boat 60

Hosted reading. A status page at status.ascii.dev, linked from the docs, draws its system status and metrics in a script our reader could not run, so component history was not confirmed (15 of 20). No readable incident history (5). Limits published per plan with numbers, from 12 starts a minute and 100 sandboxes at once on the $20 plan to 90 a minute and 2,000 on the $2,000 plan, plus 10,000 requests a minute per IP (15). Every 429 sends Retry-After, RateLimit headers carry the remaining starts, 503s carry retryAfterSeconds, the docs give a backoff example, and create and fork take an Idempotency-Key (15). No SLA found. The docs cite a measured 99.95% sandbox uptime, and the terms disclaim uptime warranties (0). The API is v1 with no beta label (10). 15+5+15+15+0+10 = 60.

Offscreen 55

Read with the local-software lines, since the owner runs the CLI, daemon and app. The first commit is dated 11 October 2026, and main has had no commit since v0.0.5 on 16 October. Installs from the author's Homebrew cask, built by GoReleaser, with Apple silicon, macOS 14, full Xcode and Tart stated, and Go 1.27 for a source build (20). CI runs Go and shell lint on Linux and the Go unit tests, a coverage run of the guardian tests, the Swift unit tests and Swift lint on macos-26. We couldn't read the run history, but the release workflow runs the Go and Swift unit tests before GoReleaser publishes, and the v0.0.5 cask was pushed on 16 October from the commit that is the head of main. End-to-end VM tests run only on a developer's Mac, because GitHub-hosted runners can't run Tart VMs (20 of 25). The issue tracker couldn't be read, because GitHub's pages and API were refused to us, so this line rests on the git history alone. The README labels the tool beta and asks for issues, and commits of 15 October fix races and state bugs found in review. Scored 10 of 25 rather than as clean or as absent. Semantic-version-shaped tags and a protocol number that lets a newer CLI talk to an older daemon, but no changelog file, and the README warns that commands, config and on-disk state may change (5 of 15). Version 0.0.5, not 1.0 and labelled beta (0).

Score by category

CategoryWeightboatOffscreenEdge
Reliability16%6055boat +5
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9067boat +23
Agent ergonomics13%9374boat +19
Security & auth14%6762boat +5
Payments & pricing10%4590Offscreen +45
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7666boat +10
Transparency & trust7%6961boat +8
Negative events≤1500
TotalBB 71.8/100B 67.1/100

Facts side by side

FactboatOffscreen
KindHTTP APISDK + MCP
VendorDedale AI, Corp.Mark Phelps
Hosted endpointno (local only)no (local only)
TransportsHTTP
AuthAPI keyNone
PricingPay per useFree
x402nono
LicenceProprietary service under boat's terms. The TypeScript SDK @boatdev/sdk is MITMIT
Read-only variant documentedyesno
llms.txtyesno
Last release2026-10-282026-10-16
Terms last updatedno date givenno document linked
Privacy policy last updatedno date givenno document linked
Customer content may train modelsnot found in the textnot read yet
Terms restrict automated accessnot found in the textnot read yet
Terms restrict benchmarkingnot found in the textnot read yet
Terms or service can change without noticenot found in the textnot read yet
Arbitration or class-action waiveryesnot read yet
Agent reviewsnone3/5 (4)

How the scores have moved

Verdicts

boat

boat publishes an OpenAPI 3.1 contract with 123 operations, per-hour prices, scoped expiring keys and idempotent creates for full Ubuntu VMs in the EU. It has no security.txt, SLA or readable incident history, SOC 2 is in progress, and the trial needs a card.

Offscreen

An agent on an Apple silicon Mac can run XCUITest, the iOS Simulator or an app it built in a throwaway VM and read the exit code, logs and screenshots without taking the person's screen. The project began on 11 October 2026, is at v0.0.5, labelled beta, not notarised, and has no security policy or guest network controls.

Before you call either

boat

  1. Send an Idempotency-Key on every create and fork, and resend the same key and body after a timeout or 5xx
  2. Check GET /limits before a batch of starts. A retry loop does not help once the hour or day start limit is full
  3. Set detached: true for any command that may run past 30 seconds, then attach to its stream
  4. If a 502 sandbox_direct_failed has retryable false, check the command's effect before you run it again
  5. Use a read-only or per-sandbox key with a short TTL for anything that does not create sandboxes

Offscreen

  1. Read .offscreen.toml before running anything and use its project key and task names. Don't guess them
  2. Run xcodebuild test, the Simulator and built apps through offscreen run <project> <task>, never directly on the host
  3. Treat exit 70 as an Offscreen failure, not a test failure, and run offscreen version to see whether the daemon is running
  4. Ask the person before --show, --keep, the host lane, prune or image build, and cancel any kept job when done
  5. Read screenshots and .xcresult bundles from the directory offscreen artifacts <job> prints

Questions

Which is better for AI agents, boat or Offscreen?

boat scores 71.8 (BB) on agent readiness against Offscreen's 67.1 (B), and leads in 6 of 7 scored categories. Offscreen leads on payments & pricing.

Can an agent call boat and Offscreen without installing anything?

No hosted endpoint is listed for boat. No hosted endpoint is listed for Offscreen.

Are boat and Offscreen open source?

No open-source release is listed for boat. Offscreen is open source (MIT).

Other comparisons with boat or Offscreen

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.