01 : Quickstart

Quickstart

Create a project board, connect the assistant you already use, and verify its access before asking it to change anything.

Quick answer

Create an Agiflow workspace and project, choose a client-specific connection, approve the intended scope, and verify the project name and ID without changing board data.

Maintained by Agiflow. Published 29 May 2026. Last updated 14 Oct 2026.

Create your workspace and project

Sign in, create or select your workspace, then create a project. A workspace is the organization boundary; the project is the board for this piece of work. Keep its name handy for the access check.

Preview : your form will look like this

Create New Project

Initialize your workspace to organize your tasks and workflows.

Project Name

Description (optional)

Building an innovative solution with AI-powered automation and seamless user experience.

Project Color

Choose your assistant and connection path

Open Connect AI tools in Agiflow and choose the assistant you use. Follow its client-specific instructions rather than pasting one generic configuration into every client.

💡 Start with the smallest useful scope: Use a project connection for this walkthrough when available. The connection panel and consent screen supply the relevant context, so a manually copied Organization ID is not a prerequisite for the recommended OAuth path.

Connect and approve access

ChatGPT

Select Agiflow in Settings, Plugins, or use the Agiflow connection entry point. Install and connect when prompted. Availability can depend on your workspace settings.

Claude

Use the remote connector path and the versioned MCP App URL shown by Agiflow. Claude Code is a separate coding-agent connection.

Coding agents

For Cursor, Codex, Claude Code, or VS Code, follow the matching setup guide and use the generated endpoint. Configuration formats differ between clients.

💡 Review the consent screen: Sign in to the account that can access your project and approve only the intended scope. OAuth uses browser authorization; API-key setup is an advanced alternative, not an extra OAuth step. See the assistant connection reference for the correct path and troubleshooting.

Confirm it works

Run a read-only check with your connected assistant:

“Use the connected Agiflow tools to identify my authorized scope and return the name and ID of the project I selected. Do not create, edit, or delete anything.”

Compare the returned scope, project name, and ID with your board. A connection badge alone is not proof that the assistant can read the intended project.

Scope matches selection

Project name and ID match

No board changes

02 : If It Doesn’t Work

“I can’t find any projects”

Confirm the project exists and that the signed-in account can open it. Compare the approved scope with the intended project. A task-scoped connection does not need permission to list every project.

Sign-in window doesn’t appear

Check the client-specific setup and authentication method. An API-key connection does not open OAuth consent. For OAuth, use the client's connect or authenticate action and verify the generated endpoint.

Connection times out

Try restarting your AI assistant and connecting again. Some tools need a restart after adding a new server connection.

See connection troubleshooting before retrying. Do not broaden access or create duplicate projects just to make verification pass.

03 : Next Steps

Connection details

Step-by-step guides for ChatGPT, Claude, and Cursor.

View guide

The project loop

How planning, building, and reviewing works with your assistant.

View guide

Support

Need help getting your first board connected?