Backed by
Multiplayer cloud workstations for coding agents and humans
Ditch local git worktrees for cloud workstations called "Rigs"
We call these cloud workstations Rigs. Each Rig is a VM loaded with your dev setup, your favorite coding agents, and your own AI subscriptions.
Loaded with your coding setup
Boot in seconds into a pre-configured VM with your repos, dev services, and preferred coding agents.
Steer agents and VMs from anywhere
Control agents and Rigs from the web, Slack, Linear, SSH, CLI, API, Codex, or Cursor.
Share with your team
Invite teammates into the same Rig to inspect previews, share agent sessions, and take over work.
Pre-configure with your setup
No more tedious environment setup. Every Rig is loaded with your git repo(s), development environment, and choice of coding agents.
- VM config per repo(s): Configure the right VM environment per repo (or multi-repo) setup. Take VM snapshots so you can boot in seconds from the right environment.
- Run your dev services: Each Rig can expose HTTPS URLs of the services you’re working on, so you can preview your work.
- Use your existing agents: Amika lets you use your favorite agent harness and existing subscriptions, for free.
Repository
github.com/acme/web
Development environment
Coding agents
Steer agents from anywhere
You can control agents and Rigs (VMs) from anywhere. Spin them up programmatically to run background jobs. Connect to a Rig with your favorite agent app, and use it like it was your desktop.
- Programmatic control: Control agents and Rigs programmatically with API or CLI. Write a job that auto-fixes Sentry issues, or even let agents create and run other Rigs and agents.
- Connect how you want: Connect to Rigs via SSH, open them up in built-in Pi Web and OpenCode browser apps, or connect with your favorite app like Codex or Cursor.
- Integrate with your work apps: Send agents messages from Slack; assign Linear issues to a Rig right from Linear; bind a Rig to a GitHub PR so agents can auto-fix CI.
- Slack@amika fix the flaky checkout test
- LinearAssigned: ENG-482
- GitHub@amika address review comments
- CLI$ amika start --task …
- APIPOST /v1/tasks
Share with your team
Rigs are multiplayer, so you and your teammates can share them. No need to copy-paste screenshots or ask each other to checkout a git branch and restart the dev servers.
- Share previews of the work: Send your teammate a preview URL of your agent’s changes. Your teammate can hop in and play around with the real application the agent changed.
- Take over an agent session: Teammates can hop into the same agent session to share the work, or they can reference a session from another Rig or agent.
acme-web.preview.amika.dev
Fixing the checkout flow · teammates can take over
Teammates in this Rig
- Youactive
- Priyaviewing preview
- Marcojoined agent
How can I use Amika?
- pick up ticket · ENG-482
- write code in sandbox
- tests + typecheck · pass
- open PR #847
Run your software factory
Background agents pick up tickets in a sandbox with every tool and service they need to implement and verify the work, then loop until checks pass and open a PR.
- Sentry · TypeError in checkout
- pull logs · GCP + AWS
- reproduce in sandbox
- open PR #481
Auto-fix production bugs
Background agents watch Sentry, pull matching logs from GCP or AWS, reproduce the issue in a sandbox, and open a GitHub PR with the fix.
Standardized cloud dev environment for humans and agents
One person sets up the whole stack, dependencies, services, and tools, plus agent hooks, skills, and MCP servers, and the whole team (humans and agents) boots the same environment.
An agent across all your tools
Start an agent in Slack, dive deeper on your computer, then share results for a teammate to inspect on their phone.
- GitHub CI · failed
- agent fixes test failure
- tests + typecheck · pass
- PR #847 · updated
Agentic CI
Connect your VMs to GitHub so when CI fails or code review comments come in, your agents can auto-fix the issues or you can take over as a human-in-the-loop.
Run agents on the machine in your closet
Message agents and sandboxes running on hardware you own, even the desktop in your closet, from anywhere.
Amika versus others
Unlike agent harnesses or sandbox providers, Amika gives you direct control over both your agents and the VMs they run on. Run any agent harness on Amika's cloud, your own cloud, or your own computers, then steer and share those Rigs from anywhere.
Compare Amika to othersEach VM Rig and agent boots from a git repo whose config defines your whole environment: the base image, services, seeded data, and which agents run where.
- Reproducible. Every VM Rig boots from the same config, so nothing drifts.
- Reviewable. Changes go through code review like anything else in your repo.
- Versioned. The config lives and moves alongside the code it works on.
An agent is three composable pieces: a VM to run in, a model paired with a harness, and your software and source code.
- Any harness, any model. Claude Code, Codex, OpenCode, or one you built, with open source or frontier-lab models.
- Reuse environments. Run many agents on top of one shared environment.
- Use and extend your tools. Agents run your existing software tools or write their own, committing the changes back to your repo.
The Amika control plane spins up VM Rigs wherever you want them to run.
- Any cloud. Provision with any cloud VM provider or Kubernetes cluster.
- Your own machines. Slice a machine you own into VM Rigs with libkrun micro-VMs or Docker containers.
- Move later. Start hosted and migrate to your own infrastructure without rewriting anything.
Send shell and exec commands to a VM and natural-language messages to its agents — all over one channel.
- Run anything. Execute programs or workflows directly on the VM.
- Multiplayer. You and your teammates can talk to the same agent at once.
- Agent-to-agent. An agent in a VM Rig can message other VM Rigs and agents to coordinate.
Every VM Rig is reachable like any other host on your network.
- SSH in. Connect to any VM Rig directly over SSH.
- On your VPN. Bring a VM Rig onto your VPN so it's addressable from anywhere.
- Live previews. Expose ports and services so a preview of what the agent is running is one URL away.
Amika snapshots the full VM state, installed tools, running services, and warm caches, so every agent starts from the same point and nothing rebuilds from scratch.
- Resume in seconds. VM Rigs and their agents come back almost instantly.
- Preserved exactly. Your environment returns just as you left it.
- No cold starts. Skip reinstalling tools and rewarming caches on every boot.
You decide exactly what an agent can see and where it can reach.
- Hidden secrets. Keep secrets out of the agent's reach while its programs still use them.
- Network control. Approve inbound and outbound requests and data.
FAQ