Wallfacer
NewWallfacer raises $1.23M to build persistent dev environments for AI coding agents→

Watch the agent build. Ship from anywhere.

Wallfacer runs Claude Code or Codex inside a real cloud dev environment with your repo, services, Docker, Xcode, and the iOS Simulator. Chat, diffs, and live preview all live in one task you can open from any device.

Start free
Image

02 · Lifecycle

Attach the issue. Steer the agent. Open the PR.

Wallfacer turns the engineering loop into a single task. Pin the issue, watch the agent work in a real environment, validate the live preview, and publish the PR — all from one screen, on any device.

Image

01

Attach context

Attach a GitHub issue or PR and Wallfacer pulls in the full thread, comments and all, then keeps it current as the discussion continues. Reference it anytime from inside the task instead of bouncing between Wallfacer, your terminal, your IDE, and a stack of GitHub tabs. No issue handy? Just type the brief into the box.

02

The agent works

The agent opens the repo, edits files, runs commands, and asks questions. You see every action above and on the next page.

03

Validate live

Live preview, iOS Simulator, screenshots. Confirm the work before you ship — see the next section.

Image

04

Publish the PR

Generate the title and body, edit, ship to GitHub. The task keeps the chat, diff, and PR in one thread.

One task owns the whole thread.

Real work is not one clean prompt. A Wallfacer task holds every attachment, every session, every correction, every diff, and every PR. Pause it, hand it off, resume it, or come back days later.

Attachments

#128 Email validator

#131 Reset timeout

Sessions

First pass, 47 msgs

Follow-up, 23 msgs

Final pass, live

Diff

auth.ts updated

retry.test.ts added

Pull requests

#412 merged

#418 open

Start free

Or read the docs to see how it works.

03 · Real previews

See it actually running. Web, mobile, and native.

Code is not done because a file changed. Wallfacer exposes any port your app listens on as a private preview URL. macOS VMs run an actual iOS Simulator on Apple Silicon, so Flutter, Expo, native Swift, and React Native preview the same way as web servers and HTTP APIs.

Image

Any port, any preview

Web servers, APIs, dashboards, Storybook. If your app listens on a port, Wallfacer gives you a private URL while it runs.

Real iOS Simulator

macOS VMs run an actual Simulator on Apple Silicon. Flutter, Expo, native Swift, and React Native render the same as on your laptop.

Live, not after-the-fact

The preview updates while the agent works. Hot reload, refresh, restart, the same loop you'd have running it locally.

One task view

Chat, diff, logs, every preview, and PR publishing stay attached to the same task.

04 · Run anything

Real toolchains. Real services. Real environments.

Define what your project needs once: repos, services, tools, ports, credentials, simulator, and instructions. Wallfacer boots a real Linux or macOS VM with the same stack a developer would use, so the agent can clone the repo, install dependencies, run services, and ship work that actually compiles.

agent-env.yml

version: 1
platform:
  os: darwin/arm64

sources:
  - repo: acme-corp/mobile-app
    workspace: /Users/admin/workspace/mobile-app

setup:
  - name: Install dependencies
    run: npm install

services:
  - name: api
    run: npm run serve
    ports:
      - port: 8080
        name: http
    healthcheck:
      url: http://localhost:8080/health
      interval: 2s
      timeout: 30s

simulators:
  - platform: ios

agent:
  instructions:
    - .claude/CLAUDE.md

Running in Wallfacer VM

API Server
api--vm-abc123.wf.devhealthy
mobile-app
cloned
iOS Simulator
iPhone 16 Prorunning
MobileApp
Debug buildinstalled

Full toolchains

Node, Docker, databases, Xcode, GitHub CLI, package managers, native builds, test suites.

Multi-repo context

Clone your main repo and reference repos into the same VM so the agent sees shared code.

Same credentials

Sessions run with the assignee's GitHub and API context, not a generic bot account.

And it's fast

Configure it once. Boot every task in seconds.

Wallfacer snapshots the configured environment after the first boot. Every new task starts inside that warm stack in a couple of seconds, so you can run many sessions without rebuilding the world each time.

Manifest reference →Environment docs →Snapshot startup →

05 · Automation

Kick off a task from anywhere. Finish it from your phone.

A GitHub Action, Slack command, webhook, cron job, or internal tool can create a Wallfacer task, attach context, choose an environment, and assign it to a teammate. The app becomes the handoff point: watch the agent work, redirect it, review the diff, and publish the PR.

Wallfacer app

Anything can start the work

GitHub Actions

Start a task when an issue is labeled, a check fails, or a workflow runs.

Issue automation

Turn triage labels, bug reports, or backlog grooming into assigned agent work.

Slack or Linear

Kick off a task from the tools your team already watches.

Internal tools

Wire Wallfacer into admin panels, support queues, cron jobs, or customer workflows.

Tasks API

Create tasks, attach context, pick an environment, and assign an owner.

Wallfacer iOS app showing a live coding agent session
GitHub Actions example →Tasks API reference →

06 · Build on it

Everything in the app. Available by API.

The API is not a separate product bolted on later. It is the same control plane the Wallfacer app uses: environments, snapshots, VMs, tasks, attachments, sessions, messages, commands, logs, and simulator hooks. Build agent coding into your own product without operating the infrastructure yourself.

Tasks API

Agent sessions on demand

Create tasks, attach issues or docs, start sessions, send follow-ups, and read transcripts.

IaaS

VMs from environments

Create an environment once, then boot Linux or macOS VMs from it when you need compute.

Your product

Embedded coding agents

Put customer-facing agent workflows in your own UI while Wallfacer runs the environments.

Internal tools

Repros, CI, back office

Spawn a repro from a bug report, wire Slack commands, or kick off cleanup tasks from cron.

One control plane for both sides of the stack.

Versioned, rate-limited, idempotent where it matters, and documented in OpenAPI. Use the app when a human is steering. Use the API when software should start the work.

API overview →Quickstart →Tasks API →

Core endpoints

POST
/environmentsDefine a repo, manifest, platform, and snapshot lifecycle
POST
/vmsBoot a VM from environment_id
POST
/environments/{id}/snapshotsRegenerate the environment snapshot
POST
/tasksCreate the durable unit of work
POST
/attachmentsAttach an issue, PR, doc, or blob
POST
/sessionsStart the coding agent inside the environment
POST
/messagesSend a follow-up message

07 · Foundations

Real VMs. Real environments. Real hardware.

Not a toy runner. Not a wrapper around someone else's agent session. Wallfacer schedules, connects, proxies, observes, and destroys isolated Linux and macOS VMs across a server fleet designed for coding agents.

Isolated by design.

Every VM gets its own SSH keypair, every task is scoped to an account member, and network isolation blocks VM-to-VM traffic. The control plane keeps logs and lifecycle state visible without exposing the VM internals directly.

Tenant isolation

Per-VM credentials, private networking, and host-side controls keep sessions separated.

Team-scoped access

Accounts, roles, invitations, API tokens, and assignee-owned tasks keep automation accountable.

Enterprise deployments

Run Wallfacer on your own hardware or in your own cloud. Custom server fleets, private networking, dedicated macOS hosts, and control over where your code runs.

Talk to us

08 · Pricing

Pay by the minute. Cheap on purpose.

Wallfacer should be affordable for any engineer, not just the ones with corporate cards. You pay by the minute, only while a VM is up.

Tux, the Linux mascotLinux
$0.0024/ minute

Ubuntu VMs with Docker, Node, Go, Python, the works. Boots from snapshot in seconds.

An hour a day for a month is about $4.30. Yes, total.

Apple logomacOS
$0.024/ minute

Real macOS on real Apple Silicon. Xcode, the iOS Simulator, and your project. Ten times the price because Apple hardware is ten times the hassle.

Same hour a day, same month: about $43.

AI tokens

Pass-through model pricing.

Tokens are billed at the model provider's published rate, with no Wallfacer markup. The rate card in our docs stays in sync with the provider.

See current rates →

Bring your own keys

Use your own API key.

Connect an Anthropic or OpenAI API key on your account and token charges are billed directly by the provider. You only pay Wallfacer for VM time. Enabling BYOK requires a credit card on file for identity verification.

On the house

$200 in free tokens. $1.44 in free infrastructure.

$200 covers a lot of Claude Code or Codex work at provider rates. $1.44 buys 10 hours of Linux or 1 hour of macOS at ours. Together, that's enough to build a real project before you spend a cent of your own money. Connect a GitHub account at signup and the credit is yours. One per human.

Questions you were probably going to ask.

Click to expand.

Built for engineers who already use coding agents.

Give the agent a real environment. Control the workflow from the app. Automate it from the API. Ship from wherever you are.

Start free