# Workflow Init - Project Setup Instructions

<critical>The workflow execution engine is governed by: {project-root}/bmad/core/tasks/workflow.xml</critical>
<critical>You MUST have already loaded and processed: workflow-init/workflow.yaml</critical>
<critical>Communicate in {communication_language} with {user_name}</critical>

<workflow>

<step n="1" goal="Quick scan and ask user about THEIR work">
<output>Welcome to BMad Method, {user_name}!</output>

<action>Quick scan for context (do NOT analyze in depth yet):</action>

- Check for codebase: src/, lib/, package.json, .git, etc.
- Check for BMM artifacts: PRD, epics, stories, tech-spec, architecture docs
- Store what was found but do NOT infer project details yet

<ask>What's your project called? {{#if project_name}}(Config shows: {{project_name}}){{/if}}</ask>
<action>Set project_name</action>
<template-output>project_name</template-output>

<check if="found artifacts OR found codebase">
<output>I found some existing work here. Let me understand what you're working on:</output>

<check if="found artifacts">
<output>
**Planning Documents Found:**
{{#each artifacts}}
- {{artifact_name}} ({{artifact_type}}, {{story_count}} stories, modified {{date}})
{{/each}}
</output>
</check>

<check if="found codebase">
<output>
**Codebase Found:**
- Source code in: {{source_dirs}}
- Tech stack: {{detected_tech_stack}}
{{#if git_history}}
- Git history: {{commit_count}} commits, last commit {{last_commit_date}}
{{/if}}
</output>
</check>

<ask>Looking at what I found, are these:

a) **Works in progress you're finishing** - continuing the work described in these documents
b) **Documents from a previous effort** - you're starting something NEW and different now
c) **The proposed work you're about to start** - these describe what you want to do
d) **None of these** - let me explain what I'm actually working on

Your choice [a/b/c/d]:</ask>

<check if="choice == a">
  <action>User is continuing old work - analyze artifacts to get details</action>
  <action>Set continuing_old_work = true</action>
  <action>Go to Step 2 (Analyze artifacts for details)</action>
</check>

<check if="choice == b">
  <action>User is doing NEW work - old artifacts are just context</action>
  <action>Set continuing_old_work = false</action>
  <action>Go to Step 3 (Ask about NEW work)</action>
</check>

<check if="choice == c">
  <action>Artifacts describe proposed work</action>
  <action>Set continuing_old_work = true</action>
  <action>Go to Step 2 (Analyze artifacts for details)</action>
</check>

<check if="choice == d">
  <action>User will explain their situation</action>
  <action>Go to Step 3 (Ask about their work)</action>
</check>
</check>

<check if="NOT found artifacts AND NOT found codebase">
  <output>I don't see any existing code or planning documents. Looks like we're starting fresh!</output>
  <action>Go to Step 3 (Ask about their work)</action>
</check>
</step>

<step n="2" goal="Analyze artifacts for continuing work" if="continuing_old_work == true">
<action>Analyze found artifacts in detail:</action>
<action>Extract project type from content (game vs software)</action>
<action>Count stories/epics to estimate level:
  - Level 0: 1 story
  - Level 1: 1-10 stories
  - Level 2: 5-15 stories
  - Level 3: 12-40 stories
  - Level 4: 40+ stories
</action>
<action>Detect field type from codebase presence (greenfield vs brownfield)</action>

<output>Based on the artifacts you're continuing, I'm suggesting **Level {{project_level}}** because I found {{story_count}} stories across {{epic_count}} epics.

Here's the complexity scale for reference:

**{{field_type}} Project Levels:**

- **Level 0** - Single atomic change (1 story) - bug fixes, typos, minor updates
- **Level 1** - Small feature (1-10 stories) - simple additions, isolated features
- **Level 2** - Medium feature set (5-15 stories) - dashboards, multiple related features
- **Level 3** - Complex integration (12-40 stories) - platform features, major integrations
- **Level 4** - Enterprise expansion (40+ stories) - multi-tenant, ecosystem changes

**My suggestion:** Level {{project_level}} {{field_type}} {{project_type}} project
</output>

<ask>Does this match what you're working on? (y/n or tell me what's different)</ask>

<check if="user confirms">
  <action>Use analyzed values</action>
  <action>Go to Step 4 (Load workflow path)</action>
</check>

<check if="user corrects">
  <action>Update values based on user corrections</action>
  <ask>Updated to: Level {{project_level}} {{field_type}} {{project_type}}. Correct? (y/n)</ask>
  <action>Go to Step 4 (Load workflow path)</action>
</check>

<template-output>project_name</template-output>
<template-output>project_type</template-output>
<template-output>project_level</template-output>
<template-output>field_type</template-output>
</step>

<step n="3" goal="Ask user about THEIR work">
<ask>Tell me about what you're working on. What's the goal?</ask>

<action>Analyze user's description using keyword detection:

- Level 0 keywords: "fix", "bug", "typo", "small change", "update", "patch", "one file"
- Level 1 keywords: "simple", "basic", "small feature", "add", "minor", "single feature"
- Level 2 keywords: "dashboard", "several features", "admin panel", "medium", "feature set"
- Level 3 keywords: "platform", "integration", "complex", "system", "architecture"
- Level 4 keywords: "enterprise", "multi-tenant", "multiple products", "ecosystem", "phased"
  </action>

<action>Make initial determination:

- project_type (game or software)
- project_level (0-4) - tentative based on keywords
- field_type (greenfield or brownfield)
- confidence (high/medium/low) - based on clarity of description
  </action>

<check if="confidence == low OR description is ambiguous">
  <output>Thanks! Let me ask a few clarifying questions to make sure I route you correctly:</output>

<ask>1. Roughly how many distinct features or changes do you think this involves?

- Just one thing (e.g., fix a bug, add one button, update one API)
- A small feature (2-5 related changes)
- Several features (5-15 related things)
- A major addition (15-40 things to do)
- A large initiative (40+ changes across many areas)
  </ask>

<action>Adjust project_level based on response</action>

<ask>2. How much of the existing codebase will this touch?

- Single file or small area
- One module or component
- Multiple modules (2-4 areas)
- Many modules with integration needs
- System-wide changes
  </ask>

<action>Validate and adjust project_level based on scope</action>

  <check if="project_type unclear">
    <ask>3. Is this a game or a software application?</ask>
    <action>Set project_type based on response</action>
  </check>
</check>

<check if="found codebase BUT field_type still unclear">
  <ask>I see you have existing code here. Are you:

1. **Adding to or modifying** the existing codebase (brownfield)
2. **Starting fresh** - the existing code is just a scaffold/template (greenfield)
3. **Something else** - let me clarify

Your choice [1/2/3]:</ask>

  <check if="choice == 1">
    <action>Set field_type = "brownfield"</action>
  </check>

  <check if="choice == 2">
    <action>Set field_type = "greenfield"</action>
    <output>Got it - treating as greenfield despite the scaffold.</output>
  </check>

  <check if="choice == 3">
    <ask>Please explain your situation:</ask>
    <action>Analyze explanation and set field_type accordingly</action>
  </check>
</check>

<action>Build reasoning for suggestion</action>
<action>Store detected_indicators (keywords, scope indicators, complexity signals)</action>

<output>Based on what you've described, I'm suggesting **Level {{project_level}}** because:

{{reasoning}} (detected: {{detected_indicators}})

Here's the complexity scale for reference:

**{{field_type}} Project Levels:**

- **Level 0** - Single atomic change (1 story) - bug fixes, typos, minor updates, single file changes
- **Level 1** - Small feature (1-10 stories) - simple additions, isolated features, one module
- **Level 2** - Medium feature set (5-15 stories) - dashboards, multiple related features, several modules
- **Level 3** - Complex integration (12-40 stories) - platform features, major integrations, architectural changes
- **Level 4** - Enterprise expansion (40+ stories) - multi-tenant, ecosystem changes, system-wide initiatives

**My suggestion:** Level {{project_level}} {{field_type}} {{project_type}} project
</output>

<ask>Does this match what you're working on? (y/n or tell me what's different)</ask>

<check if="user confirms">
  <action>Use determined values</action>
  <action>Go to Step 4 (Load workflow path)</action>
</check>

<check if="user corrects">
  <action>Update values based on corrections</action>
  <output>Updated to: Level {{project_level}} {{field_type}} {{project_type}}</output>
  <ask>Does that look right now? (y/n)</ask>
  <action>If yes, go to Step 4. If no, ask what needs adjustment and repeat.</action>
</check>

<template-output>project_name</template-output>
<template-output>project_type</template-output>
<template-output>project_level</template-output>
<template-output>field_type</template-output>
</step>

<step n="4" goal="Load appropriate workflow path">
<action>Determine path file based on selections:</action>

<check if="project_type == game">
  <action>Load {path_files}/game-design.yaml</action>
  <action>Set workflow_path_file = "game-design.yaml"</action>
</check>

<check if="project_type == software">
  <!-- field_type will be "greenfield" or "brownfield", project_level will be 0-4 -->
  <action>Build filename: {field_type}-level-{project_level}.yaml</action>
  <action>Load {path_files}/{field_type}-level-{project_level}.yaml</action>
  <action>Set workflow_path_file = constructed filename</action>
</check>

<action>Parse workflow path file to extract phases and workflows</action>
<template-output>workflow_path_file</template-output>
</step>

<step n="5" goal="Build workflow status YAML structure">
<action>Parse the loaded workflow path file and extract all workflows</action>

<action>For each phase in the path file:

- Extract phase number and name
- Extract all workflows in that phase
- For each workflow, determine its status type:
  - required: true → status = "required"
  - recommended: true → status = "recommended"
  - conditional: "if_has_ui" → status = "conditional"
  - optional: true → status = "optional"
  - Default if not specified → status = "required"
    </action>

<action>Build the workflow_items list in this format:

For each phase:

1. Add comment header: `  # Phase {n}: {Phase Name}`
2. For each workflow in phase:
   - Add entry: `  {workflow-id}: {status}`
3. Add blank line between phases

Example structure:

```
  # Phase 1: Analysis
  brainstorm-project: optional
  research: optional
  product-brief: recommended

  # Phase 2: Planning
  prd: required
  validate-prd: optional
  create-design: conditional
```

</action>

<action>Scan for existing workflow output files to auto-detect completion:

For each workflow in the list, check common output locations:

- {output_folder}/brainstorm-\*.md for brainstorm-project
- {output_folder}/research-\*.md for research
- {output_folder}/product-brief.md for product-brief
- {output_folder}/prd.md for prd
- {output_folder}/ux-design.md for create-design
- {output_folder}/architecture.md for create-architecture
- {output_folder}/tech-spec.md for tech-spec
- {output_folder}/sprint-status.yaml for sprint-planning

CRITICAL: If file exists, replace status with ONLY the file path - nothing else.
Example: product-brief: docs/product-brief.md
NOT: product-brief: "completed - docs/product-brief.md" or any other text.
</action>

<template-output>workflow_items</template-output>
</step>

<step n="6" goal="Create workflow status file">
<action>Set generated date to current date</action>
<template-output>generated</template-output>

<action>Prepare all template variables for workflow-status-template.yaml:

- generated: {current_date}
- project_name: {project_name}
- project_type: {project_type}
- project_level: {project_level}
- field_type: {field_type}
- workflow_path_file: {workflow_path_file}
- workflow_items: {workflow_items from step 5}
  </action>

<action>Display a preview of what will be created:

Show the first workflow in each phase and total count:

"Ready to create workflow status tracking:

- Phase 1 ({phase_1_workflow_count} workflows): Starting with {first_workflow_phase_1}
- Phase 2 ({phase_2_workflow_count} workflows): Starting with {first_workflow_phase_2}
- Phase 3 ({phase_3_workflow_count} workflows): Starting with {first_workflow_phase_3}
- Phase 4 (Implementation tracked separately in sprint-status.yaml)

{{#if detected_completed_workflows}}
Found existing work:
{{#each detected_files}}

- {{workflow_name}}: {{file_path}}
  {{/each}}
  {{/if}}"
  </action>

<ask>Ready to create your workflow status file? (y/n)</ask>

<check if="answer == y">
  <action>Generate YAML from workflow-status-template.yaml with all variables</action>
  <action>Save status file to {output_folder}/bmm-workflow-status.yaml</action>

<action>Identify the first non-completed workflow in the list</action>
<action>Look up that workflow's agent and command from the path file</action>

<output>✅ Workflow status file created at {output_folder}/bmm-workflow-status.yaml

**Next Steps:**

{{#if detected_completed_workflows}}
You have {{detected_count}} workflow(s) already completed. Great progress!
{{/if}}

**Next Workflow:** {{next_workflow_name}}

**Agent:** {{next_agent}}

**Command:** /bmad:bmm:workflows:{{next_workflow_id}}

{{#if next_agent !== 'pm'}}
It is recommended to start a new chat and load the {{next_agent}} agent before running the next workflow.
{{/if}}
</output>
</check>
</step>

</workflow>