One staging, many branches
A shared staging environment holds one branch at a time. Agent PRs wait in line, or overwrite each other.
Preview environments
for every agent PR
Your agents write the code. Qovery gives every agent PR a live environment on your own Kubernetes. Free for your team and unlimited agents.
3 users, unlimited agents, 1,000 deployment minutes a month. No time limit, no credit card.
A coding agent fixes a bug and opens pull request 1302 against staging. Qovery creates a preview environment for the pull request on your cluster and posts its link in the PR comments. The environment is deleted when the PR is merged or closed.
Preview env
Trusted by 200+ companies to simplify their infrastructure burden
Agents open PRs fast. Review is now the bottleneck.
A coding agent can open several pull requests in an afternoon. The diff shows what changed. Only a running environment shows whether it works.
Testing locally does not keep up
Starting the whole stack on a laptop for each branch can take longer than the agent took to write the change.
Agents should not hold your admin keys
Cloud credentials or a shared kubeconfig give an agent far more access than one pull request needs.
Three steps to your first agent preview
Most of the setup is a one-time switch on an environment you already run
1. Connect your cluster
Sign up for free and install Qovery on your Kubernetes cluster, 1.31 or later. Create the environment previews copy, usually staging.
2. Install the skill or MCP
One command installs the Qovery skills for Claude Code, Cursor, Codex and other agents. The MCP server lets them read what is running.
3. Turn on preview environments
In the environment settings, enable previews and pick the services. From then on, every agent PR gets a live environment.
Every agent PR gets its own environment
Turn on preview environments once, on the environment you already have. Each pull request gets its own copy, whoever opened it: Claude Code, Cursor, Codex or a developer.
- Created when the PR opens, deleted on merge or close
- Services, databases, variables and secrets from your blueprint
- Redeployed on every new commit
- Preview link posted in the PR comments
Agents deploy through guardrails
Agents work through the Qovery skills, the MCP server or the CLI, on your cluster and in your cloud account. Your code, data and logs stay with you.
- Each agent has its own scoped API token
- MCP access is read-only until you enable writes
- Every action lands in the audit log, under its token
Qovery played a crucial role in halving our release cycle time and providing a more cost-effective solution!
Léo-Paul Goffic
CTPO at Elevo
Free, trial or demo?
Pick the one that matches what you already run
Free
$0, no time limit
You already run a Kubernetes cluster
- 3 users, plus unlimited AI agents
- 1 cluster of your own, 1.31 or later (EKS, GKE, AKS, k3s, on-premise)
- 1,000 deployment minutes a month, then deployments pause
- Preview environments and environment cloning
- No credit card. Community support and docs
Starter trial
14 days free
You want Qovery to manage the cluster
- 3 users, plus unlimited AI agents
- 1 cluster, yours or managed by Qovery in your cloud account
- 3,000 deployment minutes a month
- Standard support in Slack or Microsoft Teams
- No credit card for the trial
Demo
No cloud account
You want to look before you connect anything
- Live demo: the Console on sample data, in your browser
- qovery demo up: Qovery on a local k3s cluster
- Deploy an app and read its logs
- No advanced settings
Frequently asked questions
How do I get a preview environment for every pull request an AI agent opens?
Connect your Kubernetes cluster to Qovery and create the environment previews should copy, usually staging. In its settings, enable preview environments and select the services. Qovery then creates an environment for every pull request that targets the base branch, whoever opened it, redeploys it on every commit, posts the preview link in the PR comments and deletes it when the PR is merged or closed.
Is it really free?
Yes. The Free plan includes 3 users, unlimited AI agents, 1 Kubernetes cluster of your own, 1.31 or later (EKS, GKE, AKS, k3s or on-premise) and 1,000 deployment minutes a month. There is no time limit and no credit card. Support is community and docs. To try it first, run qovery demo up on your laptop.
Which coding agents does it work with?
Any agent that opens pull requests, because the preview is triggered by the pull request. To let an agent deploy, read logs and create previews itself, install the Qovery skills (Claude Code, Cursor, OpenAI Codex, Gemini CLI, OpenCode and other tools that support Agent Skills) or connect it to the Qovery MCP server at https://mcp.qovery.com/mcp.
Do I need GitHub Actions?
No. Preview environments are triggered by Qovery's Git integration, which watches pull requests on GitHub, GitLab or Bitbucket. If you would rather drive previews from CI, the Qovery CLI can clone an environment, deploy a commit and delete the environment from a GitHub Actions workflow.
What happens when the Free deployment minutes run out?
New deployments pause until next month, or until you upgrade to a paid plan. Free never bills overage.
How do agents stay inside guardrails?
Each agent authenticates with its own API token, scoped by the role it was issued under, or by a policy you write for an API Policy Token. The MCP server is read-only unless you enable writes on the connection and in the Console. Every action is recorded in the audit log against the token that made it. AI agents do not count as users.
What is the difference between Free, the Starter trial and the demo?
Free is a permanent plan on a Kubernetes cluster you already run, version 1.31 or later. The Starter trial is 14 days of the paid Starter plan, where Qovery can also create and manage the cluster in your cloud account. The demos need no cloud account: the live demo shows the Console on sample data, and qovery demo up installs Qovery on a local k3s cluster so you can deploy an app and read its logs. Not sure yet? Start with qovery demo up, then connect your cluster to Free.
Your agents write the code.
Qovery runs every PR.
Free for 3 users and unlimited agents, on your own Kubernetes. No time limit, no credit card.