BMAD-METHOD/src/modules/bmm/workflows/2-plan-workflows/tech-spec/instructions.md

# Tech-Spec Workflow - Context-Aware Technical Planning (Level 0-1)

<workflow>

<critical>The workflow execution engine is governed by: {project-root}/bmad/core/tasks/workflow.xml</critical>
<critical>You MUST have already loaded and processed: {installed_path}/workflow.yaml</critical>
<critical>Communicate all responses in {communication_language} and language MUST be tailored to {user_skill_level}</critical>
<critical>Generate all documents in {document_output_language}</critical>
<critical>This is for Level 0-1 projects - tech-spec with context-rich story generation</critical>
<critical>Level 0: tech-spec + single user story | Level 1: tech-spec + epic/stories</critical>
<critical>LIVING DOCUMENT: Write to tech-spec.md continuously as you discover - never wait until the end</critical>
<critical>CONTEXT IS KING: Gather ALL available context before generating specs</critical>
<critical>DOCUMENT OUTPUT: Technical, precise, definitive. Specific versions only. User skill level ({user_skill_level}) affects conversation style ONLY, not document content.</critical>
<critical>Input documents specified in workflow.yaml input_file_patterns - workflow engine handles fuzzy matching, whole vs sharded document discovery automatically</critical>

<step n="0" goal="Validate workflow readiness and detect project level" tag="workflow-status">
<action>Check if {output_folder}/bmm-workflow-status.yaml exists</action>

<check if="status file not found">
  <output>No workflow status file found. Tech-spec workflow can run standalone or as part of BMM workflow path.</output>
  <output>**Recommended:** Run `workflow-init` first for project context tracking and workflow sequencing.</output>
  <output>**Quick Start:** Continue in standalone mode - perfect for rapid prototyping and quick changes!</output>
  <ask>Continue in standalone mode or exit to run workflow-init? (continue/exit)</ask>
  <check if="continue">
    <action>Set standalone_mode = true</action>

    <output>Great! Let's quickly configure your project...</output>

    <ask>What level is this project?

**Level 0** - Single atomic change (bug fix, small isolated feature, single file change)
→ Generates: 1 tech-spec + 1 story
→ Example: "Fix login validation bug" or "Add email field to user form"

**Level 1** - Coherent feature (multiple related changes, small feature set)
→ Generates: 1 tech-spec + 1 epic + 2-3 stories
→ Example: "Add OAuth integration" or "Build user profile page"

Enter **0** or **1**:</ask>

    <action>Capture user response as project_level (0 or 1)</action>
    <action>Validate: If not 0 or 1, ask again</action>

    <ask>Is this a **greenfield** (new/empty codebase) or **brownfield** (existing codebase) project?

**Greenfield** - Starting fresh, no existing code
**Brownfield** - Adding to or modifying existing code

Enter **greenfield** or **brownfield**:</ask>

    <action>Capture user response as field_type (greenfield or brownfield)</action>
    <action>Validate: If not greenfield or brownfield, ask again</action>

    <output>Perfect! Running as:

- **Project Level:** {{project_level}}
- **Field Type:** {{field_type}}
- **Mode:** Standalone (no status file tracking)

Let's build your tech-spec!</output>
</check>
<check if="exit">
<action>Exit workflow</action>
</check>
</check>

<check if="status file found">
  <action>Load the FULL file: {output_folder}/bmm-workflow-status.yaml</action>
  <action>Parse workflow_status section</action>
  <action>Check status of "tech-spec" workflow</action>
  <action>Get project_level from YAML metadata</action>
  <action>Get field_type from YAML metadata (greenfield or brownfield)</action>
  <action>Find first non-completed workflow (next expected workflow)</action>

  <check if="project_level >= 2">
    <output>**Incorrect Workflow for Level {{project_level}}**

Tech-spec is for Level 0-1 projects. Level 2-4 should use PRD workflow.

**Correct workflow:** `create-prd` (PM agent)
</output>
<action>Exit and redirect to prd</action>
</check>

  <check if="tech-spec status is file path (already completed)">
    <output>⚠️ Tech-spec already completed: {{tech-spec status}}</output>
    <ask>Re-running will overwrite the existing tech-spec. Continue? (y/n)</ask>
    <check if="n">
      <output>Exiting. Use workflow-status to see your next step.</output>
      <action>Exit workflow</action>
    </check>
  </check>

  <check if="tech-spec is not the next expected workflow">
    <output>⚠️ Next expected workflow: {{next_workflow}}. Tech-spec is out of sequence.</output>
    <ask>Continue with tech-spec anyway? (y/n)</ask>
    <check if="n">
      <output>Exiting. Run {{next_workflow}} instead.</output>
      <action>Exit workflow</action>
    </check>
  </check>

<action>Set standalone_mode = false</action>
</check>
</step>

<step n="1" goal="Comprehensive context discovery - gather everything available">

<action>Welcome {user_name} warmly and explain what we're about to do:

"I'm going to gather all available context about your project before we dive into the technical spec. This includes:

- Any existing documentation (product briefs, research)
- Brownfield codebase analysis (if applicable)
- Your project's tech stack and dependencies
- Existing code patterns and structure

This ensures the tech-spec is grounded in reality and gives developers everything they need."
</action>

<action>**PHASE 1: Load Existing Documents**

Search for and load (using dual-strategy: whole first, then sharded):

1. **Product Brief:**
   - Search pattern: {output*folder}/\_brief*.md
   - Sharded: {output*folder}/\_brief*/index.md
   - If found: Load completely and extract key context

2. **Research Documents:**
   - Search pattern: {output*folder}/\_research*.md
   - Sharded: {output*folder}/\_research*/index.md
   - If found: Load completely and extract insights

3. **Document-Project Output (CRITICAL for brownfield):**
   - Always check: {output_folder}/docs/index.md
   - If found: This is the brownfield codebase map - load ALL shards!
   - Extract: File structure, key modules, existing patterns, naming conventions

Create a summary of what was found:

- List of loaded documents
- Key insights from each
- Brownfield vs greenfield determination
  </action>

<action>**PHASE 2: Detect Project Type from Setup Files**

Search for project setup files in {project-root}:

**Node.js/JavaScript:**

- package.json → Parse for framework, dependencies, scripts

**Python:**

- requirements.txt → Parse for packages
- pyproject.toml → Parse for modern Python projects
- Pipfile → Parse for pipenv projects

**Ruby:**

- Gemfile → Parse for gems and versions

**Java:**

- pom.xml → Parse for Maven dependencies
- build.gradle → Parse for Gradle dependencies

**Go:**

- go.mod → Parse for modules

**Rust:**

- Cargo.toml → Parse for crates

**PHP:**

- composer.json → Parse for packages

If setup file found, extract:

1. Framework name and EXACT version (e.g., "React 18.2.0", "Django 4.2.1")
2. All production dependencies with versions
3. Dev dependencies and tools (TypeScript, Jest, ESLint, pytest, etc.)
4. Available scripts (npm run test, npm run build, etc.)
5. Project type indicators (is it an API? Web app? CLI tool?)
6. **Test framework** (Jest, pytest, RSpec, JUnit, Mocha, etc.)

**Check for Outdated Dependencies:**
<check if="major framework version > 2 years old">
<action>Use WebSearch to find current recommended version</action>
<example>
If package.json shows "react": "16.14.0" (from 2020):
<WebSearch query="React latest stable version 2025 migration guide" />
Note both current version AND migration complexity in stack summary
</example>
</check>

**For Greenfield Projects:**
<check if="field_type == greenfield">
<action>Use WebSearch for current best practices AND starter templates</action>
<example>
<WebSearch query="{detected_framework} best practices {current_year}" />
<WebSearch query="{detected_framework} recommended packages {current_year}" />
<WebSearch query="{detected_framework} official starter template {current_year}" />
<WebSearch query="{project_type} {detected_framework} boilerplate {current_year}" />
</example>

**RECOMMEND STARTER TEMPLATES:**
Look for official or well-maintained starter templates:

- React: Create React App, Vite, Next.js starter
- Vue: create-vue, Nuxt starter
- Python: cookiecutter templates, FastAPI template
- Node.js: express-generator, NestJS CLI
- Ruby: Rails new, Sinatra template
- Go: go-blueprint, standard project layout

Benefits of starters:

- ✅ Modern best practices baked in
- ✅ Proper project structure
- ✅ Build tooling configured
- ✅ Testing framework set up
- ✅ Linting/formatting included
- ✅ Faster time to first feature

**Present recommendations to user:**
"I found these starter templates for {{framework}}:

1. {{official_template}} - Official, well-maintained
2. {{community_template}} - Popular community template

These provide {{benefits}}. Would you like to use one? (yes/no/show-me-more)"

<action>Capture user preference on starter template</action>
<action>If yes, include starter setup in implementation stack</action>
</check>

Store this as {{project_stack_summary}}
</action>

<action>**PHASE 3: Brownfield Codebase Reconnaissance** (if applicable)

<check if="field_type == brownfield OR document-project output found">

Analyze the existing project structure:

1. **Directory Structure:**
   - Identify main code directories (src/, lib/, app/, components/, services/)
   - Note organization patterns (feature-based, layer-based, domain-driven)
   - Identify test directories and patterns

2. **Code Patterns:**
   - Look for dominant patterns (class-based, functional, MVC, microservices)
   - Identify naming conventions (camelCase, snake_case, PascalCase)
   - Note file organization patterns

3. **Key Modules/Services:**
   - Identify major modules or services already in place
   - Note entry points (main.js, app.py, index.ts)
   - Document important utilities or shared code

4. **Testing Patterns & Standards (CRITICAL):**
   - Identify test framework in use (from package.json/requirements.txt)
   - Note test file naming patterns (.test.js, \_test.py, .spec.ts, Test.java)
   - Document test organization (tests/, **tests**, spec/, test/)
   - Look for test configuration files (jest.config.js, pytest.ini, .rspec)
   - Check for coverage requirements (in CI config, test scripts)
   - Identify mocking/stubbing libraries (jest.mock, unittest.mock, sinon)
   - Note assertion styles (expect, assert, should)

5. **Code Style & Conventions (MUST CONFORM):**
   - Check for linter config (.eslintrc, .pylintrc, rubocop.yml)
   - Check for formatter config (.prettierrc, .black, .editorconfig)
   - Identify code style:
     - Semicolons: yes/no (JavaScript/TypeScript)
     - Quotes: single/double
     - Indentation: spaces/tabs, size
     - Line length limits
   - Import/export patterns (named vs default, organization)
   - Error handling patterns (try/catch, Result types, error classes)
   - Logging patterns (console, winston, logging module, specific formats)
   - Documentation style (JSDoc, docstrings, YARD, JavaDoc)

Store this as {{existing_structure_summary}}

**CRITICAL: Confirm Conventions with User**
<ask>I've detected these conventions in your codebase:

**Code Style:**
{{detected_code_style}}

**Test Patterns:**
{{detected_test_patterns}}

**File Organization:**
{{detected_file_organization}}

Should I follow these existing conventions for the new code?

Enter **yes** to conform to existing patterns, or **no** if you want to establish new standards:</ask>

<action>Capture user response as conform_to_conventions (yes/no)</action>

<check if="conform_to_conventions == no">
  <ask>What conventions would you like to use instead? (Or should I suggest modern best practices?)</ask>
  <action>Capture new conventions or use WebSearch for current best practices</action>
</check>

<action>Store confirmed conventions as {{existing_conventions}}</action>

</check>

<check if="field_type == greenfield">
  <action>Note: Greenfield project - no existing code to analyze</action>
  <action>Set {{existing_structure_summary}} = "Greenfield project - new codebase"</action>
</check>

</action>

<action>**PHASE 4: Synthesize Context Summary**

Create {{loaded_documents_summary}} that includes:

- Documents found and loaded
- Brownfield vs greenfield status
- Tech stack detected (or "To be determined" if greenfield)
- Existing patterns identified (or "None - greenfield" if applicable)

Present this summary to {user_name} conversationally:

"Here's what I found about your project:

**Documents Available:**
[List what was found]

**Project Type:**
[Brownfield with X framework Y version OR Greenfield - new project]

**Existing Stack:**
[Framework and dependencies OR "To be determined"]

**Code Structure:**
[Existing patterns OR "New codebase"]

This gives me a solid foundation for creating a context-rich tech spec!"
</action>

<template-output>loaded_documents_summary</template-output>
<template-output>project_stack_summary</template-output>
<template-output>existing_structure_summary</template-output>

</step>

<step n="2" goal="Conversational discovery of the change/feature">

<action>Now engage in natural conversation to understand what needs to be built.

Adapt questioning based on project_level:
</action>

<check if="project_level == 0">
  <action>**Level 0: Atomic Change Discovery**

Engage warmly and get specific details:

"Let's talk about this change. I need to understand it deeply so the tech-spec gives developers everything they need."

**Core Questions (adapt naturally, don't interrogate):**

1. "What problem are you solving?"
   - Listen for: Bug fix, missing feature, technical debt, improvement
   - Capture as {{change_type}}

2. "Where in the codebase should this live?"
   - If brownfield: "I see you have [existing modules]. Does this fit in any of those?"
   - If greenfield: "Let's figure out the right structure for this."
   - Capture affected areas

3. <check if="brownfield">
   "Are there existing patterns or similar code I should follow?"
   - Look for consistency requirements
   - Identify reference implementations
   </check>

4. "What's the expected behavior after this change?"
   - Get specific success criteria
   - Understand edge cases

5. "Any constraints or gotchas I should know about?"
   - Technical limitations
   - Dependencies on other systems
   - Performance requirements

**Discovery Goals:**

- Understand the WHY (problem)
- Understand the WHAT (solution)
- Understand the WHERE (location in code)
- Understand the HOW (approach and patterns)

Synthesize into clear problem statement and solution overview.
</action>
</check>

<check if="project_level == 1">
  <action>**Level 1: Feature Discovery**

Engage in deeper feature exploration:

"This is a Level 1 feature - coherent but focused. Let's explore what you're building."

**Core Questions (natural conversation):**

1. "What user need are you addressing?"
   - Get to the core value
   - Understand the user's pain point

2. "How should this integrate with existing code?"
   - If brownfield: "I saw [existing features]. How does this relate?"
   - Identify integration points
   - Note dependencies

3. <check if="brownfield AND similar features exist">
   "Can you point me to similar features I can reference for patterns?"
   - Get example implementations
   - Understand established patterns
   </check>

4. "What's IN scope vs OUT of scope for this feature?"
   - Define clear boundaries
   - Identify MVP vs future enhancements
   - Keep it focused (remind: Level 1 = 2-3 stories max)

5. "Are there dependencies on other systems or services?"
   - External APIs
   - Databases
   - Third-party libraries

6. "What does success look like?"
   - Measurable outcomes
   - User-facing impact
   - Technical validation

**Discovery Goals:**

- Feature purpose and value
- Integration strategy
- Scope boundaries
- Success criteria
- Dependencies

Synthesize into comprehensive feature description.
</action>
</check>

<template-output>problem_statement</template-output>
<template-output>solution_overview</template-output>
<template-output>change_type</template-output>
<template-output>scope_in</template-output>
<template-output>scope_out</template-output>

</step>

<step n="3" goal="Generate context-aware, definitive technical specification">

<critical>ALL TECHNICAL DECISIONS MUST BE DEFINITIVE - NO AMBIGUITY ALLOWED</critical>
<critical>Use existing stack info to make SPECIFIC decisions</critical>
<critical>Reference brownfield code to guide implementation</critical>

<action>Initialize tech-spec.md with the rich template</action>

<action>**Generate Context Section (already captured):**

These template variables are already populated from Step 1:

- {{loaded_documents_summary}}
- {{project_stack_summary}}
- {{existing_structure_summary}}

Just save them to the file.
</action>

<template-output file="tech-spec.md">loaded_documents_summary</template-output>
<template-output file="tech-spec.md">project_stack_summary</template-output>
<template-output file="tech-spec.md">existing_structure_summary</template-output>

<action>**Generate The Change Section:**

Already captured from Step 2:

- {{problem_statement}}
- {{solution_overview}}
- {{scope_in}}
- {{scope_out}}

Save to file.
</action>

<template-output file="tech-spec.md">problem_statement</template-output>
<template-output file="tech-spec.md">solution_overview</template-output>
<template-output file="tech-spec.md">scope_in</template-output>
<template-output file="tech-spec.md">scope_out</template-output>

<action>**Generate Implementation Details:**

Now make DEFINITIVE technical decisions using all the context gathered.

**Source Tree Changes - BE SPECIFIC:**

Bad (NEVER do this):

- "Update some files in the services folder"
- "Add tests somewhere"

Good (ALWAYS do this):

- "src/services/UserService.ts - MODIFY - Add validateEmail() method at line 45"
- "src/routes/api/users.ts - MODIFY - Add POST /users/validate endpoint"
- "tests/services/UserService.test.ts - CREATE - Test suite for email validation"

Include:

- Exact file paths
- Action: CREATE, MODIFY, DELETE
- Specific what changes (methods, classes, endpoints, components)

**Use brownfield context:**

- If modifying existing files, reference current structure
- Follow existing naming patterns
- Place new code logically based on current organization
  </action>

<template-output file="tech-spec.md">source_tree_changes</template-output>

<action>**Technical Approach - BE DEFINITIVE:**

Bad (ambiguous):

- "Use a logging library like winston or pino"
- "Use Python 2 or 3"
- "Set up some kind of validation"

Good (definitive):

- "Use winston v3.8.2 (already in package.json) for logging"
- "Implement using Python 3.11 as specified in pyproject.toml"
- "Use Joi v17.9.0 for request validation following pattern in UserController.ts"

**Use detected stack:**

- Reference exact versions from package.json/requirements.txt
- Specify frameworks already in use
- Make decisions based on what's already there

**For greenfield:**

- Make definitive choices and justify them
- Specify exact versions
- No "or" statements allowed
  </action>

<template-output file="tech-spec.md">technical_approach</template-output>

<action>**Existing Patterns to Follow:**

<check if="brownfield">
Document patterns from the existing codebase:
- Class structure patterns
- Function naming conventions
- Error handling approach
- Testing patterns
- Documentation style

Example:
"Follow the service pattern established in UserService.ts:

- Export class with constructor injection
- Use async/await for all asynchronous operations
- Throw ServiceError with error codes
- Include JSDoc comments for all public methods"
  </check>

<check if="greenfield">
"Greenfield project - establishing new patterns:
- [Define the patterns to establish]"
</check>

</action>

<template-output file="tech-spec.md">existing_patterns</template-output>

<action>**Integration Points:**

Identify how this change connects:

- Internal modules it depends on
- External APIs or services
- Database interactions
- Event emitters/listeners
- State management

Be specific about interfaces and contracts.
</action>

<template-output file="tech-spec.md">integration_points</template-output>

<action>**Development Context:**

**Relevant Existing Code:**
<check if="brownfield">
Reference specific files or code sections developers should review:

- "See UserService.ts lines 120-150 for similar validation pattern"
- "Reference AuthMiddleware.ts for authentication approach"
- "Follow error handling in PaymentService.ts"
  </check>

**Framework/Libraries:**
List with EXACT versions from detected stack:

- Express 4.18.2 (web framework)
- winston 3.8.2 (logging)
- Joi 17.9.0 (validation)
- TypeScript 5.1.6 (language)

**Internal Modules:**
List internal dependencies:

- @/services/UserService
- @/middleware/auth
- @/utils/validation

**Configuration Changes:**
Any config files to update:

- Update .env with new SMTP settings
- Add validation schema to config/schemas.ts
- Update package.json scripts if needed
  </action>

<template-output file="tech-spec.md">existing_code_references</template-output>
<template-output file="tech-spec.md">framework_dependencies</template-output>
<template-output file="tech-spec.md">internal_dependencies</template-output>
<template-output file="tech-spec.md">configuration_changes</template-output>

<check if="field_type == brownfield">
  <template-output file="tech-spec.md">existing_conventions</template-output>
</check>

<check if="field_type == greenfield">
  <action>Set {{existing_conventions}} = "Greenfield project - establishing new conventions per modern best practices"</action>
  <template-output file="tech-spec.md">existing_conventions</template-output>
</check>

<action>**Implementation Stack:**

Comprehensive stack with versions:

- Runtime: Node.js 20.x
- Framework: Express 4.18.2
- Language: TypeScript 5.1.6
- Testing: Jest 29.5.0
- Linting: ESLint 8.42.0
- Validation: Joi 17.9.0

All from detected project setup!
</action>

<template-output file="tech-spec.md">implementation_stack</template-output>

<action>**Technical Details:**

Deep technical specifics:

- Algorithms to implement
- Data structures to use
- Performance considerations
- Security considerations
- Error scenarios and handling
- Edge cases

Be thorough - developers need details!
</action>

<template-output file="tech-spec.md">technical_details</template-output>

<action>**Development Setup:**

What does a developer need to run this locally?

Based on detected stack and scripts:

```
1. Clone repo (if not already)
2. npm install (installs all deps from package.json)
3. cp .env.example .env (configure environment)
4. npm run dev (starts development server)
5. npm test (runs test suite)
```

Or for Python:

```
1. python -m venv venv
2. source venv/bin/activate
3. pip install -r requirements.txt
4. python manage.py runserver
```

Use the actual scripts from package.json/setup files!
</action>

<template-output file="tech-spec.md">development_setup</template-output>

<action>**Implementation Guide:**

**Setup Steps:**
Pre-implementation checklist:

- Create feature branch
- Verify dev environment running
- Review existing code references
- Set up test data if needed

**Implementation Steps:**
Step-by-step breakdown:

For Level 0:

1. [Step 1 with specific file and action]
2. [Step 2 with specific file and action]
3. [Write tests]
4. [Verify acceptance criteria]

For Level 1:
Organize by story/phase:

1. Phase 1: [Foundation work]
2. Phase 2: [Core implementation]
3. Phase 3: [Testing and validation]

**Testing Strategy:**

- Unit tests for [specific functions]
- Integration tests for [specific flows]
- Manual testing checklist
- Performance testing if applicable

**Acceptance Criteria:**
Specific, measurable, testable criteria:

1. Given [scenario], when [action], then [outcome]
2. [Metric] meets [threshold]
3. [Feature] works in [environment]
   </action>

<template-output file="tech-spec.md">setup_steps</template-output>
<template-output file="tech-spec.md">implementation_steps</template-output>
<template-output file="tech-spec.md">testing_strategy</template-output>
<template-output file="tech-spec.md">acceptance_criteria</template-output>

<action>**Developer Resources:**

**File Paths Reference:**
Complete list of all files involved:

- /src/services/UserService.ts
- /src/routes/api/users.ts
- /tests/services/UserService.test.ts
- /src/types/user.ts

**Key Code Locations:**
Important functions, classes, modules:

- UserService class (src/services/UserService.ts:15)
- validateUser function (src/utils/validation.ts:42)
- User type definition (src/types/user.ts:8)

**Testing Locations:**
Where tests go:

- Unit: tests/services/
- Integration: tests/integration/
- E2E: tests/e2e/

**Documentation to Update:**
Docs that need updating:

- README.md - Add new endpoint documentation
- API.md - Document /users/validate endpoint
- CHANGELOG.md - Note the new feature
  </action>

<template-output file="tech-spec.md">file_paths_complete</template-output>
<template-output file="tech-spec.md">key_code_locations</template-output>
<template-output file="tech-spec.md">testing_locations</template-output>
<template-output file="tech-spec.md">documentation_updates</template-output>

<action>**UX/UI Considerations:**

<check if="change affects user interface OR user experience">
  **Determine if this change has UI/UX impact:**
  - Does it change what users see?
  - Does it change how users interact?
  - Does it affect user workflows?

If YES, document:

**UI Components Affected:**

- List specific components (buttons, forms, modals, pages)
- Note which need creation vs modification

**UX Flow Changes:**

- Current flow vs new flow
- User journey impact
- Navigation changes

**Visual/Interaction Patterns:**

- Follow existing design system? (check for design tokens, component library)
- New patterns needed?
- Responsive design considerations (mobile, tablet, desktop)

**Accessibility:**

- Keyboard navigation requirements
- Screen reader compatibility
- ARIA labels needed
- Color contrast standards

**User Feedback:**

- Loading states
- Error messages
- Success confirmations
- Progress indicators
  </check>

<check if="no UI/UX impact">
  "No UI/UX impact - backend/API/infrastructure change only"
</check>
</action>

<template-output file="tech-spec.md">ux_ui_considerations</template-output>

<action>**Testing Approach:**

Comprehensive testing strategy using {{test_framework_info}}:

**CONFORM TO EXISTING TEST STANDARDS:**
<check if="conform_to_conventions == yes">

- Follow existing test file naming: {{detected_test_patterns.file_naming}}
- Use existing test organization: {{detected_test_patterns.organization}}
- Match existing assertion style: {{detected_test_patterns.assertion_style}}
- Meet existing coverage requirements: {{detected_test_patterns.coverage}}
  </check>

**Test Strategy:**

- Test framework: {{detected_test_framework}} (from project dependencies)
- Unit tests for [specific functions/methods]
- Integration tests for [specific flows/APIs]
- E2E tests if UI changes
- Mock/stub strategies (use existing patterns: {{detected_test_patterns.mocking}})
- Performance benchmarks if applicable
- Accessibility tests if UI changes

**Coverage:**

- Unit test coverage: [target %]
- Integration coverage: [critical paths]
- Ensure all acceptance criteria have corresponding tests
  </action>

<template-output file="tech-spec.md">test_framework_info</template-output>
<template-output file="tech-spec.md">testing_approach</template-output>

<action>**Deployment Strategy:**

**Deployment Steps:**
How to deploy this change:

1. Merge to main branch
2. Run CI/CD pipeline
3. Deploy to staging
4. Verify in staging
5. Deploy to production
6. Monitor for issues

**Rollback Plan:**
How to undo if problems:

1. Revert commit [hash]
2. Redeploy previous version
3. Verify rollback successful

**Monitoring:**
What to watch after deployment:

- Error rates in [logging service]
- Response times for [endpoint]
- User feedback on [feature]
  </action>

<template-output file="tech-spec.md">deployment_steps</template-output>
<template-output file="tech-spec.md">rollback_plan</template-output>
<template-output file="tech-spec.md">monitoring_approach</template-output>

<invoke-task halt="true">{project-root}/bmad/core/tasks/adv-elicit.xml</invoke-task>

</step>

<step n="4" goal="Auto-validate cohesion, completeness, and quality">

<critical>Always run validation - this is NOT optional!</critical>

<action>Tech-spec generation complete! Now running automatic validation...</action>

<action>Load {installed_path}/checklist.md</action>
<action>Review tech-spec.md against ALL checklist criteria:

**Section 1: Output Files Exist**

- Verify tech-spec.md created
- Check for unfilled template variables

**Section 2: Context Gathering**

- Validate all available documents were loaded
- Confirm stack detection worked
- Verify brownfield analysis (if applicable)

**Section 3: Tech-Spec Definitiveness**

- Scan for "or" statements (FAIL if found)
- Verify all versions are specific
- Check stack alignment

**Section 4: Context-Rich Content**

- Verify all new template sections populated
- Check existing code references (brownfield)
- Validate framework dependencies listed

**Section 5-6: Story Quality (deferred to Step 5)**

**Section 7: Workflow Status (if applicable)**

**Section 8: Implementation Readiness**

- Can developer start immediately?
- Is tech-spec comprehensive enough?
  </action>

<action>Generate validation report with specific scores:

- Context Gathering: [Comprehensive/Partial/Insufficient]
- Definitiveness: [All definitive/Some ambiguity/Major issues]
- Brownfield Integration: [N/A/Excellent/Partial/Missing]
- Stack Alignment: [Perfect/Good/Partial/None]
- Implementation Readiness: [Yes/No]
  </action>

<check if="validation issues found">
  <output>⚠️ **Validation Issues Detected:**

{{list_of_issues}}

I can fix these automatically. Shall I proceed? (yes/no)</output>

<ask>Fix validation issues? (yes/no)</ask>

  <check if="yes">
    <action>Fix each issue and re-validate</action>
    <output>✅ Issues fixed! Re-validation passed.</output>
  </check>

  <check if="no">
    <output>⚠️ Proceeding with warnings. Issues should be addressed manually.</output>
  </check>
</check>

<check if="validation passes">
  <output>✅ **Validation Passed!**

**Scores:**

- Context Gathering: {{context_score}}
- Definitiveness: {{definitiveness_score}}
- Brownfield Integration: {{brownfield_score}}
- Stack Alignment: {{stack_score}}
- Implementation Readiness: ✅ Ready

Tech-spec is high quality and ready for story generation!</output>
</check>

</step>

<step n="5" goal="Generate context-rich user stories">

<action>Now generate stories that reference the rich tech-spec context</action>

<check if="project_level == 0">
  <action>Invoke {installed_path}/instructions-level0-story.md to generate single user story</action>
  <action>Story will leverage tech-spec.md as primary context</action>
  <action>Developers can skip story-context workflow since tech-spec is comprehensive</action>
</check>

<check if="project_level == 1">
  <action>Invoke {installed_path}/instructions-level1-stories.md to generate epic and stories</action>
  <action>Stories will reference tech-spec.md for all technical details</action>
  <action>Epic provides organization, tech-spec provides implementation context</action>
</check>

</step>

<step n="6" goal="Finalize and guide next steps">

<output>**✅ Tech-Spec Complete, {user_name}!**

**Deliverables Created:**

<check if="project_level == 0">
- ✅ **tech-spec.md** - Context-rich technical specification
  - Includes: brownfield analysis, framework details, existing patterns
- ✅ **story-{slug}.md** - Implementation-ready user story
  - References tech-spec as primary context
</check>

<check if="project_level == 1">
- ✅ **tech-spec.md** - Context-rich technical specification
- ✅ **epics.md** - Epic and story organization
- ✅ **story-{epic-slug}-1.md** - First story
- ✅ **story-{epic-slug}-2.md** - Second story
{{#if story_3}}
- ✅ **story-{epic-slug}-3.md** - Third story
{{/if}}
</check>

**What Makes This Tech-Spec Special:**

The tech-spec is comprehensive enough to serve as the primary context document:

- ✨ Brownfield codebase analysis (if applicable)
- ✨ Exact framework and library versions from your project
- ✨ Existing patterns and code references
- ✨ Specific file paths and integration points
- ✨ Complete developer resources

**Next Steps:**

<check if="project_level == 0">
**For Single Story (Level 0):**

**Option A - With Story Context (for complex changes):**

1. Ask SM agent to run `create-story-context` for the story
   - This generates additional XML context if needed
2. Then ask DEV agent to run `dev-story` to implement

**Option B - Direct to Dev (most Level 0):**

1. Ask DEV agent to run `dev-story` directly
   - Tech-spec provides all the context needed!
   - Story is ready to implement

💡 **Tip:** Most Level 0 changes don't need separate story context since tech-spec is comprehensive!
</check>

<check if="project_level == 1">
**For Multiple Stories (Level 1):**

**Recommended: Story-by-Story Approach**

For the **first story** ({{first_story_name}}):

**Option A - With Story Context (recommended for first story):**

1. Ask SM agent to run `create-story-context` for story 1
   - Generates focused context for this specific story
2. Then ask DEV agent to run `dev-story` to implement story 1

**Option B - Direct to Dev:**

1. Ask DEV agent to run `dev-story` for story 1
   - Tech-spec has most context needed

After completing story 1, repeat for stories 2 and 3.

**Alternative: Sprint Planning Approach**

- If managing multiple stories as a sprint, ask SM agent to run `sprint-planning`
- This organizes all stories for coordinated implementation
  </check>

**Your Tech-Spec:**

- 📄 Saved to: `{output_folder}/tech-spec.md`
- Contains: All context, decisions, patterns, and implementation guidance
- Ready for: Direct development or story context generation

The tech-spec is your single source of truth! 🚀
</output>

</step>

</workflow>