MCP Integration¶
When you work with an AI assistant, you describe what needs to be built. The problem is your AI can't see those requirements unless you paste them every time.
Quadro solves this through the Model Context Protocol. Your AI can read your tasks and see what needs to be implemented. Write your specs as tasks. Your AI reads them and helps you build.
What is MCP?¶
MCP (Model Context Protocol) is a standard that lets AI assistants access tools and data. Quadro provides an MCP server that gives your AI direct access to your tasks.
Your AI can read task descriptions, see requirements, and help you implement them. When you create or update tasks through MCP, they appear as markdown files in your tasks/ directory.
Setup¶
Choose your AI assistant and follow the setup for your tool.
Claude Code¶
Claude Code uses a CLI to manage MCP servers. Add Quadro using the claude mcp add command:
The --scope user makes Quadro available across all your projects. Use --scope project to add it only to the current project.
Verify the setup:
You should see quadro in the list. The tools are now available in Claude Code.
Cursor¶
Cursor supports MCP through configuration files. You can use project-specific or global configuration:
- Project: Create
.cursor/mcp.jsonin your project directory - Global: Create
~/.cursor/mcp.jsonfor all projects
Add this configuration:
{
"mcpServers": {
"quadro": {
"command": "uvx",
"args": [
"--from",
"quadro",
"python",
"-m",
"quadro.mcp"
],
"env": {}
}
}
}
Restart Cursor. The Quadro tools will be available in your AI agent.
Windsurf¶
Windsurf by Codeium supports MCP through a configuration file:
- Config file:
~/.codeium/windsurf/mcp_config.json
Add this configuration:
{
"mcpServers": {
"quadro": {
"command": "uvx",
"args": [
"--from",
"quadro",
"python",
"-m",
"quadro.mcp"
],
"env": {}
}
}
}
Click the refresh button in Windsurf to apply the changes.
Cline¶
Cline is a VS Code extension that supports MCP. Install the extension first, then add Quadro through Cline's configuration interface:
- Open Cline settings in VS Code
- Navigate to MCP servers section
- Add a new server with these settings:
- Command:
uvx - Args:
--from,quadro,python,-m,quadro.mcp
Cline will display the Quadro tools once configured.
How to use it¶
You need to explicitly ask your AI assistant to use Quadro. The AI won't automatically know you want task management unless you tell it.
Creating tasks¶
When you're discussing code and want to capture a task:
The task file appears in your tasks/ directory.
Want to add it to a milestone?
Checking your board¶
Show me my Quadro tasks
List Quadro tasks in the MVP milestone
What tasks do I have in progress using Quadro?
Updating status¶
Moving tasks¶
Deleting tasks¶
Workflow examples¶
Planning a feature¶
Break down what needs to be built:
I need to build user authentication. Use Quadro to create these tasks in MVP: JWT token generation, login endpoint, password hashing.
Your AI creates the tasks. Later when you start implementing, your AI can read those task specs to help you code.
During implementation¶
Ask your AI to read a task and help you implement it:
Then mark progress:
Daily standup¶
Sprint planning¶
Use Quadro to create these in Sprint-2: refactor API handlers, add request middleware, improve error handling
Context switching¶
Bug tracking¶
Found a bug while coding:
Tips¶
Always mention "Quadro" or "using Quadro" when you want the AI to interact with your tasks. Otherwise, it might just discuss tasks without creating them.
You can request multiple operations:
Mix task management with code discussions:
Check your board periodically:
Keep things organized:
How it works¶
When you ask your AI to use Quadro, the AI calls the MCP server, which reads or modifies markdown files in your tasks/ directory. Changes appear in your filesystem immediately. You see them in git diff.
Your AI can read tasks to see what needs to be built. It can create tasks with titles, descriptions, and milestones. It can update status (TODO → PROGRESS → DONE). It can move tasks between milestones, delete tasks, and show milestone summaries.
Troubleshooting¶
AI can't find Quadro tools: Check the MCP configuration is correct, restart your AI assistant, verify uvx is in your PATH.
Permission errors: Make sure the tasks/ directory has read/write permissions for the user running the MCP server.
Next steps¶
Try creating a task with some requirements, then ask your AI to help you implement it. See if having specs your AI can read fits your workflow.