Cloudnova Agents · Cloudnova Deployments
Connect Cloudnova to your cloud account. Agents take issues from your backlog through your development lifecycle, with your team approving every step. Deployments give every branch and every release an environment of its own. Your code, your credentials and your AI subscriptions stay with you.
delegated to @cloudnovaenvironment acme-142in your cloud
Scoped to the billing module, one question asked and answered
Sara moved it to Todo
Six-step plan posted on the issue for review
Tom approved the plan
Writing src/billing/export.ts
$ pnpm test billing
✓ 48 passed (3.1s)
$ pnpm dev
Starts when you move it to review
Products
Both products stand on the same foundation: environments built from your repository, running in your cloud account, managed from one Cloudnova control plane.
Your backlog, worked by agents.
Delegate an issue from your tracker and an agent takes it through your development lifecycle, in an environment of its own. After every phase it stops, until someone on your team says go.
An environment for every branch and every release.
The environments your agents work in, for the rest of your delivery: a preview for every pull request, a deploy on every merge, all in your own cloud.
Cloudnova Agents
Agents follow the lifecycle your team already works by. Each phase ends on the issue, with a report and a person deciding what happens next. The phases, their names and what an agent does in each are yours to define.
Reads the issue and the code around it, asks what is missing and sizes the work.
You move it to the backlog
Writes an implementation plan onto the issue and parks it for review.
You approve the plan
Branches, writes the code, runs your tests and your stack, and opens the pull request.
You move it to review
Reviews the pull request against the plan and your conventions, and gives a verdict.
You merge
Nothing starts until somebody hands an issue to the agent. An issue nobody delegated is left alone.
Where an issue sits on your board decides which agent runs next. You approve by moving the issue, exactly as you would for a colleague.
Reply on the issue and the same agent picks it up in the same environment, with everything it did and said before.
Agents run on the AI subscriptions or keys your team already holds, from the provider you choose. Which model does which phase is configuration.
Environments
Every issue gets an environment of its own, with your repository checked out and your whole stack running. The agent tests its work the way your developers do, and you can look at what it built. The same environments power Cloudnova Deployments.
Resumes in seconds on the next reply
Services
Workspace
Between turns an environment is paused with everything in it still running. The next reply picks it up in seconds, dev server and database included.
Services an agent starts show up on their own. Open one from the session at a stable address, behind your own sign-in.
The conversation and the work in progress are saved outside the environment. Rebuilding one costs a moment of start-up, never the work.
Every night your repository is refreshed, dependencies are installed and the database is prepared, so work starts on today's code.
Agents working in parallel never share files, processes or ports. What happens in one environment stays in it.
Your cloud
Most platforms ask you to move your code onto theirs. Cloudnova works the other way around: you connect our control plane to your own cloud account, and everything that touches your code runs there.
Network
The runner in your cloud connects out to the control plane. Nothing needs a way into your network.
Data
Your repository, images, secrets and keys live in your own cloud account, under your own policies.
AI
Agents run on the AI subscriptions or API keys you hold. Usage is tracked per seat, before and after every turn.
People
Every organization gets its own address. Members sign in with the Microsoft or Google work account you already manage.
Tools
The tools and MCP servers your repository declares get their credentials at the start of a turn, never baked into an image.
Audit
What was asked, who asked it, what the agent did and what it used, in one timeline per session.
Your repository
How your agents work and what their environment looks like are decisions about your code, so they live in your code: versioned, reviewed and yours.
Which agents exist, what each one does and which column of your board starts it are declared in your repository. Four phases, two or seven: it is your process.
Cloudnova builds the image, runs the step that turns a fresh clone into a working tree, and keeps it ready for the next issue.
A change to how agents work is a pull request. Cloudnova picks it up from your main branch; nothing hides in a settings screen.
# .cloudnova/agents.yaml
version: 1
# name them after your own process
agentTypes:
scope:
skill: scope-issue
build:
skill: build-feature
security:
skill: security-review
# which column of your board runs which agent
triggerRules:
- event: delegated
statusType: triage
agentType: scope
- event: delegated
statusType: started
agentType: build
- event: delegated
statusType: completed
agentType: securityCloudnova started as a platform that ran cloud environments for other people's software. The environments are still the hard part. Now agents work in them, and your deployments will run on them.
Cloudnova opens to new organizations soon. Already set up? Sign in to your organization.