Your First Workflow

This guide walks through a complete Monday Morning workflow: creating a feature, writing a spec, tracking tasks, and marking work complete. You can do this manually with a text editor or through AI agent commands.

The Workflow

1. Create a feature (optional grouping)
2. Create a spec (define the work)
3. Add tasks (break it down)
4. Implement and track progress
5. Complete the spec

1. Create a Feature

Features group related specs. They are optional — you can create specs without features.

Create a feature directory and feature.json:

mkdir -p .mm/features/onboarding

.mm/features/onboarding/feature.json

{
  "id": "onboarding",
  "name": "User Onboarding",
  "description": "First-run experience for new users",
  "status": "planning",
  "priority": "high",
  "progress": {
    "total_specs": 0,
    "completed_specs": 0,
    "percentage": 0
  },
  "linked_specs": [],
  "created": "2026-03-31T10:00:00Z",
  "updated": "2026-03-31T10:00:00Z"
}

2. Create a Spec

Specs are the core unit of work. Each spec lives in a dated folder under .mm/specs/.

mkdir -p .mm/specs/2026-03-31-welcome-wizard

Write the specification:

.mm/specs/2026-03-31-welcome-wizard/spec.md

# Specification: Welcome Wizard

## Goal
Build a 3-step welcome wizard that runs on first login.

## Requirements
- Step 1: Collect user name and role
- Step 2: Choose notification preferences
- Step 3: Import existing project or create new one

## Out of Scope
- Team onboarding (separate spec)
- Email verification (handled by auth spec)

Create the implementation file to track tasks:

.mm/specs/2026-03-31-welcome-wizard/implementation.md

# Welcome Wizard - Implementation

## Completed

## In Progress

## Backlog
- [ ] T1: Wizard layout and navigation component [M]
  - [ ] T1-1: Step indicator bar
  - [ ] T1-2: Next/back navigation
- [ ] T2: Step 1 — name and role form [S]
- [ ] T3: Step 2 — notification preferences [S]
- [ ] T4: Step 3 — project import or create [L]
- [ ] T5: Persist wizard state and mark onboarding complete [S]

3. Link the Spec to Your Feature

Update the feature to reference the new spec:

.mm/features/onboarding/feature.json

{
  "id": "onboarding",
  "name": "User Onboarding",
  "description": "First-run experience for new users",
  "status": "in-progress",
  "priority": "high",
  "linked_specs": [
    "../specs/2026-03-31-welcome-wizard"
  ],
  "progress": {
    "total_specs": 1,
    "completed_specs": 0,
    "percentage": 0
  },
  "created": "2026-03-31T10:00:00Z",
  "updated": "2026-03-31T10:00:00Z"
}

Tip

The linked_specs path is relative to the .mm/features/ directory, so it uses ../specs/ to reference the specs folder.

4. Track Progress

As you work, move tasks between sections and check them off.

When you start working on a task, move it to In Progress:

.mm/specs/2026-03-31-welcome-wizard/implementation.md

# Welcome Wizard - Implementation

## Completed

## In Progress
- [ ] T1: Wizard layout and navigation component [M]
  - [x] T1-1: Step indicator bar
  - [ ] T1-2: Next/back navigation

## Backlog
- [ ] T2: Step 1 — name and role form [S]
- [ ] T3: Step 2 — notification preferences [S]
- [ ] T4: Step 3 — project import or create [L]
- [ ] T5: Persist wizard state and mark onboarding complete [S]

When a task is done, move it to Completed and mark it with [x]:

## Completed
- [x] T1: Wizard layout and navigation component [M]
  - [x] T1-1: Step indicator bar
  - [x] T1-2: Next/back navigation

Progress is calculated from top-level checkboxes only. In this example, completing T1 means 1 of 5 tasks done (20%).

5. Using AI Agents

Instead of editing files manually, you can use AI agent commands to manage your workflow. With the MCP server configured, your AI assistant has access to tools like:

Tool	What it does
mm_create_spec	Create a new spec with folder structure
mm_update_tasks	Move tasks between sections
mm_complete_task	Mark a task as done with rollup
mm_get_project_status	See overall project progress
mm_create_issue	Log a bug or problem
mm_create_note	Capture a decision or meeting note

For example, asking your AI assistant to “create a spec for the welcome wizard” will call mm_create_spec and generate the folder, spec.md, and implementation.md for you.

6. Complete the Spec

When all tasks are checked off, the spec is complete. If the spec is linked to a feature, the feature’s progress updates automatically.

Final implementation.md

# Welcome Wizard - Implementation

## Completed
- [x] T1: Wizard layout and navigation component [M]
- [x] T2: Step 1 — name and role form [S]
- [x] T3: Step 2 — notification preferences [S]
- [x] T4: Step 3 — project import or create [L]
- [x] T5: Persist wizard state and mark onboarding complete [S]

## In Progress

## Backlog

The desktop app dashboard and mm_get_project_status tool will reflect 100% completion for this spec.

Summary

The Monday Morning workflow is:

1. Define — Create specs that describe what you are building and why.
2. Break down — Add tasks with size estimates to the implementation file.
3. Track — Move tasks through In Progress to Completed as you work.
4. Recover — Next session, AI agents read .mm/ and know exactly where you left off.

All of this is stored in plain files in your repo. No external service required.

Next Steps

• MCP Tools Reference — See every tool available to AI agents.
• Integrations — Connect GitHub, Grain, Notion, and more.