Pick your path

Same docs, three audiences. The fastest read for you depends on which one you are.

I'm a developer with a laptop

You write code. Sometimes an agent does. You want to step away from the keyboard while it works.

Start here:

1. Developer quickstart · 5 minutes, pair a machine, approve a decision from your phone.
2. Add a repo · tell the daemon which directories to watch.
3. Mobile experience · the killer feature: decide from anywhere.
4. Daily driver loop · what an average day looks like once it's set up.

Then, per agent:

• Claude Code, Codex, Aider, Gemini.

I'm rolling Viewport out to my team

You're not just running agents. Your team is. You want a shared inbox, audit log, and approval routing.

Start here:

1. Create an organization.
2. Invite teammates.
3. Set up teams · org-local groups for approval routing.
4. Share resources · workflows, plans, vaults.
5. Approval policies · who decides what, and when it can auto-approve.

Plus:

• Slack integration, Webhook delivery, Audit log.

I want to self-host

You can't or won't run on hosted infrastructure. The relay is shippable today; the control plane is on the roadmap.

Start here:

1. Self-host overview · what you can deploy today, what's still hosted.
2. Deploy the relay · one Docker command, plus DNS and TLS.
3. Backplane modes · multi-relay topologies (server vs Redis).
4. Monitor and operate · /state, /metrics, /logs, alerting recipes.
5. Security posture · threat model, what the relay can and cannot see.
6. On-prem roadmap · what's coming, in what order, on what timeline.

Common to all paths

Whoever you are, you'll hit these eventually:

• Concepts · sessions, plans, inbox, vault, workflows. The mental model.
• CLI reference · every vpd subcommand with real flags.
• Workflow YAML · the spec for .viewport/workflows/*.yaml.
• API · REST + WebSocket protocol.
• Troubleshooting · the top failure modes and recovery.