Dockup CLI · AI agent skill

The deploymentlayer forAI agents

Add Dockup as a skill and give your AI agent the power to provision infrastructure, connect databases, deploy services, and operate production end to end.

Quick installCLI + agent skill
01$ npm install -g dockup-cliCLI
02$ dockup skill installSkill
Agent-readyCodexClaude Code
Open Dockup
From repository to production
InfrastructureDatabasesServicesDomains
Agent deploymentProvision · Deploy · Operate
01Install once

One skill.Every deployment target.

Give Dockup to Codex or Claude Code once. The same agent can ship services, operate managed databases, and provision Linux or Windows compute without switching tools.

Agent workspaceReady
Codex / Claude CodeDeploy the stack.

The agent understands the request. Dockup turns it into infrastructure actions.

$ dockup skill installskill connected_
Deployment targets3 connected
01
Create · Build · Deploy

Application service

Ship a repository or container as a running production service.

dockup create api --repo … --deploy --wait
Available
02
Connect · Operate · Back up

Managed database

Manage production data, private connections, users and backups.

dockup db backup project/main-db --json
Available
03
Provision · Run · Access

Compute / VM

Create Linux boxes or Windows VMs for workloads beyond app containers.

dockup box create --image ubuntu-22.04 --json
Available
One interfaceStructured JSONReal exit codesAgent-safe control
02Execution loop

One request.A verified production stack.

Your agent discovers the available controls, prepares every workload, waits for deployment, and verifies the result before handing the workflow back.

agent-run / productionConnected
Agent request

“Ship the API, connect its database, provision the worker compute, and confirm everything is healthy.”

01
Discoverdockup schema --json
Capabilities and targets resolvedDone
02
Provisionservice + database + compute
Infrastructure preparedDone
03
Deploydockup deploy --wait --json
Build completed in 42sDone
04
Verifydockup status && dockup uptime
Production health confirmedLive
IntentInfrastructureDeploymentVerified production
03Production control

Deploy is notthe finish line.

When production changes, your agent can read runtime and build logs, diagnose the failure, restart or roll back the service, and confirm that it recovered.

Agent recovery traceIncident #0041 · Resolved
0114:02:11
Detectdockup uptime production/api --json
Health check failed · 503Alert
0214:02:14
Diagnosedockup logs production/api --json
Runtime crash identifiedDone
0314:02:19
Recoverdockup rollback dep_7f2a production/api --json
Stable release restoredDone
0414:02:46
Verifydockup status production/api --json
Running · Health passingPassed
$ dockup logs production/api --build --follow --jsonAgent control active_
Runtime logsBuild logsRestartRollbackExec / ShellUptime
04Config as code

See the changebefore production does.

Put service configuration in dockup.yaml. Your agent can inspect the diff, explain exactly what will change, and apply it only when the plan is right.

dockup.yamlSource of truth
  1. service:
  2. branch: main
  3. port: 3000
  4. resources:
  5. cpu: 2
  6. memory: 4096
  7. healthcheck:
  8. path: /health
  9. retries: 5
  10. domains:
  11. - api.example.com
Tracked with your codeValid YAML
dockup plan / production-apiRead only
$ dockup plan production/api --json4 results
~
resources.cpuUpdate
1→2
+
healthcheck.pathAdd
—→/health
+
domains[0]Add
—→api.example.com
=
service.portNo change
3000→3000
Plan complete3 changes · 0 destructive
dockup up --deploy
01Plan is read-only
02Additive by default
03Secrets stay out
04Prune is explicit
05Preview environments

Production shape.Without production power.

Every pull request or branch can get an isolated deployment, its own URL and environment, plus safe access to the data needed for a realistic review.

Change sourcePR #142
Branch

feature/checkout

Commit a84c12f is ready for a production-shaped review.

$ dockup preview branch feature/checkout
Ephemeral environmentLive
pr-api-feature-checkout.dockup.tech
Checkout previewReady for review.

Built from the branch and isolated from the production service.

Service
Running
Environment
Isolated
Private net
Connected
Production dataProtected
Managed database

production-db

Dockup creates a dedicated preview credential automatically.

Access modeRead only
Auto-provisioned userRead only
Preview controldockup preview
LogsEnvironmentRedeployMergeDelete
06Agent guardrails

Autonomous work.Production guardrails.

Dockup gives agents direct control without giving them permission to guess. Sensitive output stays hidden, destructive actions stop for confirmation, and every deploy passes security and health checks.

Agent actionAwaiting decision
Destructive intent detected

Remove production service.

The agent can request the action. Dockup decides whether the request is explicit enough to execute.

$ dockup destroy production/api --json
{"ok": false,"code": "needs_confirm"}
Dockup policy gateActive
01
IdentityToken context verified
Passed
02
SecretsValues redacted from output
Protected
03
IntentExplicit confirmation missing
Blocked
04
AuditDecision recorded
Recorded
Production decisionAction paused
Nothing changedConfirmation required.

The agent receives a structured error and can stop, ask for intent, or re-run with --yes.

Latest deploy securityA
Critical
0
High
0
Config
Pass
Health
Pass
Structured errorsSecret redactionExplicit confirmationCVE + config scanHealth gateAudit trail
07Usage-based pricing

Pay for what runs.Not how much you ship.

Pick a monthly credit level and use it across your production stack. Deployments stay unlimited on paid plans; actual resource usage draws from your balance.

PlanMonthlyIncluded balanceScopeAction
01Free
$0/ month
$2.50 launch credits1 workspace · 3 databases · 3 deploymentsStart free
02Hobby
$5/ month
$5 usage creditsUnlimited workspaces · databases · deploymentsChoose Hobby
03GoRecommended
$20/ month
$20 usage creditsUnlimited workspaces · databases · deploymentsChoose Go
04Pro
$35/ month
$35 usage creditsUnlimited workspaces · databases · deploymentsChoose Pro
ServicesManaged databasesLinux / Windows computeOne shared balance
08Start here

Give your agenta deployment layer.

Install Dockup once. Your agent gets one interface to deploy, operate, recover and protect the complete production stack.

Open Dockup
Works withCodexClaude CodeCI / Containers
dockup / setupOnline
01
$ npm install -g dockup-cliDockup CLI installed
Done
02
$ dockup skill installShared agent skill connected
Done
03
$ dockup whoami --jsonAuthentication and agent context verified
Ready
{"ok": true, "agent": "ready"}
One installEnd-to-end production control_