Quickstart
Go from zero to your first AI-created stories in under 5 minutes.
1. Create a Workspace
Sign up and create your first workspace. A workspace is where your projects, stories, and team members live.
2. Connect your AI (OAuth — recommended)
Add DolphinQ's MCP server to your AI client. No token needed — your client signs you in through your browser on first connect.
Run this in your project directory:
claude mcp add --transport http dolphinq https://mcp.dolphin-q.com/Or add it to .mcp.json in your project root to share it with your team:
{
"mcpServers": {
"dolphinq": {
"type": "http",
"url": "https://mcp.dolphin-q.com/"
}
}
}On first connect your browser opens. Sign in with your DolphinQ account, pick a workspace, and you're connected — you stay signed in for up to 90 days.
Other MCP clients: DolphinQ is a spec-compliant MCP server, so community clients (mcphost + Ollama, etc.) connect too. Output quality depends on the model — smaller local models may dump raw tool JSON or ignore formatting instructions. For the best experience, use Claude or Codex.
3. Alternative: token authentication
If your client can't do OAuth, open Workspace Settings → Integrations, generate an MCP token (it starts with dlpq_), and send it as an Authorization header. For Claude Code:
{
"mcpServers": {
"dolphinq": {
"type": "http",
"url": "https://mcp.dolphin-q.com/",
"headers": {
"Authorization": "Bearer dlpq_..."
}
}
}
}Keep tokens out of URLs — URLs end up in logs and shell history. Treat a token like a password, and don't commit it: keep a token-based .mcp.json out of git.
4. Verify the connection
Start your AI (run claude or codex, or open Claude Desktop) and ask it to call say_my_name. It should reply with your name, workspace, and role.
5. Plan your first project
Ask your AI to set up a project and break the work into stories with dependencies. It creates the project and components for you if they don't exist yet:
Create a project called "Checkout" in DolphinQ.
Plan the checkout flow for our e-commerce app —
payments, order tracking, email receipts —
as stories with dependencies.Your AI creates stories with titles, components, priorities, and dates, then wires the dependency links — up to 50 stories in a single call.
Tip: Point your AI at a Notion page, Google Doc, or spreadsheet and it will turn it into stories — as long as the matching connector is set up in your AI client.
6. Query your project
Ask your AI about project status at any time:
What's the status of the Checkout project?
Any overdue stories or blockers?Your AI returns completion, overdue stories, blocked work, and the critical path.