First-run setup

When you open Rundock for the first time, a guided sequence walks you through choosing a workspace and creating your first agent. This page sets expectations for what you will see and what each step is for. Nothing you do here is permanent. Workspaces, agents, and settings can all be changed later. The goal of first-run is a working team you can talk to, not a perfect one.

​

What you will see

​

Setup adapts to the runtimes you have

First-run detects which engines are on your machine. It always sets up Claude Code, which runs your team’s orchestrator and Doc himself. If you also have the Codex CLI installed, setup offers an optional step to sign in to Codex, so specialists can run on your ChatGPT plan. The step is skippable: you can sign in later by running codex login in a terminal, and Settings shows whether Codex is signed in. A machine without Codex sees the flow exactly as before.

​

1. Choose a workspace folder

Rundock opens on a screen that says Open a workspace to get started. A workspace is a folder on your machine that holds everything Rundock uses for one team: the agent files, the skill files, and any shared knowledge documents. The full concept is covered on the Workspaces page. You have two options, below any workspaces you have opened before:

• Open folder. Use this if you already have a folder you want Rundock to manage, or if you are joining a team that has shared a synced workspace folder with you (see the team workspace setup guide for that case).
• New workspace. Type a name and Rundock creates the folder at ~/Documents/Rundock/<name> and places the initial files in it.

For a first run, New workspace is the simplest choice. You can move the folder later if you change your mind.

​

2. Doc opens the conversation

A workspace with no team yet shows Welcome to Rundock and a Set up your team button. Press it and a conversation opens with Doc. Doc is Rundock’s built-in guide. It is a platform agent, not a specialist you build, and it is there to walk you through the first-run flow. You do not need to create or configure Doc. Rundock places Doc’s file in every workspace, at .claude/agents/rundock-guide.md, and brings it up to date each time the workspace opens, so do not edit it. Doc stays out of the org chart with your own agents.

​

3. Doc walks through first-agent creation

Once a workspace is selected, Doc helps you create your first agent. It will ask for:

• A name. Short and memorable. The name is used in conversation (“ask Quill to draft a post”) and on the org chart card.
• A role. A short description of what this agent is responsible for. Two to four words is typical.
• Instructions. What this agent does, what it should not do, and any voice or tone guidance. Doc will offer a starter template you can adjust.

Doc saves the agent as a markdown file in your workspace. You can edit the file later in any text editor or through Rundock’s interface.

​

4. The org chart appears

Once the first agent is saved, Rundock renders the org chart with that agent visible. You can talk to the agent directly, or you can talk to the orchestrator and it will route work to the right specialist. Right now, with one specialist, both routes go to the same place. This is the working state. From here you can create more agents, write skills, set up routines, or just start using the agent you have.

​

Where to next

Two paths from here. Pick the one that matches what you are doing.