PROTOCOL: INCEPTION-WORKSHOP (EXECUTION LEVEL 3)

<state_machine> CURRENT_STEP -> REQUIRED_TOOL -> REQUIRED_OUTPUT -> MANDATORY_HALT Step 1: Vision & Boundaries -> read_file(templates/1-.md, references/1-.md) -> docs/inception/1-product-vision-and-boundaries.md -> 🛑 HALT TURN in facilitate mode Step 2: Tradeoffs -> read_file(templates/2-.md, references/2.2-.md) -> docs/inception/2-tradeoffs.md (7 X marks Golden Rule) -> 🛑 HALT TURN in facilitate mode Step 3: Personas -> read_file(templates/3-.md, references/3-.md) -> docs/inception/3-personas/ -> 🛑 HALT TURN in facilitate mode Step 4: Empathy Map -> read_file(templates/4-.md, references/4-.md) -> docs/inception/4-empathy-map.md -> 🛑 HALT TURN in facilitate mode Step 5: User Journey -> read_file(templates/5-.md, references/5-.md) -> docs/inception/5-user-journeys/ -> 🛑 HALT TURN in facilitate mode Step 6: Brainstorming -> read_file(templates/6-.md, references/6-.md) -> docs/inception/6-brainstorming.md -> 🛑 HALT TURN in facilitate mode Step 7: Features/Sequencing -> read_file(templates/7-.md, references/7-.md) -> docs/inception/7-features-and-sequencing.md -> 🛑 HALT TURN in facilitate mode Step 8: MVP Canvas -> read_file(templates/8-.md, references/8-.md) -> docs/inception/8-mvp-canvas.md -> 🛑 HALT TURN (Present workshop output) </state_machine>

<execution_invariants>

1. WORKFLOW IMMUTABILITY: Facilitate Lean Inception workshops through the 8 structured steps according to the scenario run-order table (Greenfield vs. Brownfield) without skipping steps or inventing custom layouts (workflow structure = immutable | step reasoning = flexible).
2. TEMPLATE LOCK: Before generating any step artifact (1-product-vision-and-boundaries.md through 8-mvp-canvas.md), you MUST call read_file on the corresponding template in templates/ and validator in references/. Never reconstruct an inception template from memory.
3. STRIP INSTRUCTIONAL META-TEXT: Output documents must preserve 100% of structural headings, tables, and diagrams from the template while stripping all AI instructions, format guidelines, and validation checklists.
4. ESCAPE HATCH: If the situation does not fit the contract, do not silently improvise and do not force-fit: declare the deviation in one line and ask. </execution_invariants>

<output_contract>

• TARGET_DIR: docs/inception/ (honouring .sdlc/config.json overrides)
• SOURCE_TEMPLATES: templates/1-product-vision-and-boundaries.md through templates/8-mvp-canvas-definition.md (MUST call read_file before writing each step)
• VALIDATOR_RUBRICS: references/1-product-vision-boundary-validator.md through references/8-mvp-canvas-definition-validator.md
• ESCAPE_HATCH: Never silently omit or invent template sections or tables. If a section genuinely does not apply to the product context, preserve the heading and write "N/A — [1-line rationale]", or halt and ask the user before deviating. </output_contract>

Critical Quality Rules:

• Product Name Consistency: Use the exact product name consistently throughout all documents. Include it in titles, introductions, and key sections. After establishing context, use pronouns or "the system" naturally. Focus on clarity, not forced repetition.
• Boundary Definitions: MUST include at least 4 items in EACH of the 4 boundary columns (IS, IS NOT, DOES, DOES NOT) - this is non-negotiable.
• Tradeoff Golden Rule: Exactly ONE check (X) per priority column. If you have 7 priorities, you must have exactly 7 X marks total.
• Template Instruction Filtering: STRICT RULE - When generating output documents from templates, ALL instructional text must be removed. Instructions are for the AI agent only, not for the final document. See "Template Usage Guidelines" below for detailed filtering rules.

Template Usage Guidelines

CRITICAL: Template instructions must NEVER appear in output documents.

Templates serve two purposes:

1. Structure - Define the document format (tables, sections, diagrams)
2. Guidance - Tell AI agents how to fill the template

Only the structure should appear in output documents. Guidance must be stripped.

What to Remove (All Modes)

What to Keep

Validation Checklist

Before finalizing any output document, verify:

• No "Instructions" sections remain
• No "How to use" guidance remains
• No column/field definition lists remain
• No validation checklists for the AI remain
• Document contains only actual content, not meta-instructions

Failure to follow this rule results in documents that mix instructions with content, making them unusable as actual deliverables.

Natural Language Activation

This skill can be triggered using conversational phrases instead of command-line flags. The system will automatically detect your intent and activate the appropriate mode.

Natural Language Command Patterns

Examples

# Natural language (system auto-detects mode)
"Start a Lean Inception workshop for SessioFlow"
→ pi skill inception-workshop --mode facilitate --step 1

"Let's work on the tradeoffs using AI debate"
→ pi skill inception-workshop --mode tradeoff-generator

"Generate all 8 inception steps automatically"
→ pi skill inception-workshop --mode batch --context "SessioFlow"

"Validate the user journey mapping"
→ pi skill inception-workshop --mode validate --step 5 --file docs/inception/5-user-journeys/

"Create flow specifications from the journey"
→ pi skill inception-workshop --mode generate-flows --from-step 6

Step Recognition

The skill recognizes these step names in natural language:

Command-Line Interface

Start Interactive Facilitation

Template -> Fill -> Validate workflow:

pi skill inception-workshop --mode facilitate --step [N]

Options:

• --mode facilitate - Interactive template->fill->validate (default)
• --mode batch - Generate all steps automatically
• --mode tradeoff-generator - Use AI debate to generate tradeoffs
• --step [N] - Start from specific step (1-8), default: 1

Examples

# Start from Step 1 (default)
pi skill inception-workshop --mode facilitate

# Start from Step 5 (Journey Mapping)
pi skill inception-workshop --mode facilitate --step 5

# Generate all 8 steps automatically
pi skill inception-workshop --mode batch --context "SessioFlow: Call-for-Papers platform"

# Use AI debate to generate tradeoffs (Step 2 only)
pi skill inception-workshop --mode tradeoff-generator

# Validate a completed step
pi skill inception-workshop --mode validate --step 1 --file docs/inception/1-product-vision-and-boundaries.md

Workflow

Facilitate Mode (Template -> Fill -> Validate)

CRITICAL RULE: Template Instructions Must NOT Leak to Output

Templates contain instructional guidelines for AI agents. When generating output documents:

MUST REMOVE:

• "Instructions" sections (entire section)
• "Format" guidelines and field definitions
• "How to use" instructions
• Column explanations and usage notes
• Validation checklists meant for the AI
• Meta-commentary about the template itself

MUST KEEP:

• Content structure (tables, sections, placeholders)
• Actual data entries (filled content)
• Diagrams and visual elements
• "Next Step" references
• Notes & Observations sections (for actual observations, not instructions)

Implementation:

1. Read the template file from templates/
2. Create the output file in docs/inception/
3. Strip all instructional text before writing output
4. Prompt the user to fill the document
5. When ready, validate against the validator criteria
6. Provide feedback and scoring
7. Proceed to next step if score ≥ 8

Implementation: Use read/write/edit tools to manage files, provide interactive guidance to the user.

Tradeoff Generator Mode (AI Debate)

Special mode for Step 2 only - The agent simulates stakeholder debate to generate tradeoff analysis:

Agent Implementation:

1. Read Step 1 (Product Vision) from docs/inception/1-product-vision-and-boundaries.md
2. Simulate 4 stakeholder perspectives (PO, UX Advocate, Tech Lead, Agile Coach)
3. Generate individual priority perspectives based on the vision
4. Synthesize consensus tradeoff board with STRICT 1-7 ranking
5. VERIFY: Count X marks - must be exactly 7 (one per column). If not, fix immediately.
6. Validate against validator criteria
7. Output to docs/inception/2-tradeoffs.md

Note: The agent should generate the debate content internally, presenting each stakeholder viewpoint and then synthesizing consensus.

Batch Mode

CRITICAL RULE: Template Instructions Must NOT Leak to Output

All templates contain instructional guidelines for AI agents. When generating output documents in batch mode:

MUST REMOVE:

• "Instructions" sections (entire section)
• "Format" guidelines and field definitions
• "How to use" instructions
• Column explanations and usage notes
• Validation checklists meant for the AI
• Meta-commentary about the template itself

MUST KEEP:

• Content structure (tables, sections, placeholders)
• Actual data entries (filled content)
• Diagrams and visual elements
• "Next Step" references
• Notes & Observations sections (for actual observations, not instructions)

Agent Implementation:

1. Parse the context/product description - EXTRACT the exact product name
2. Sequentially generate all 8 steps using templates
3. Strip all instructional text before writing each output file
4. Product Name Guidelines:
   • Include the product name in titles and document headers
   • Use the product name in introductions and key sections
   • After establishing context, use "the system", "the platform", or pronouns naturally
   • Maintain consistency - don't switch between different names
   • Prioritize natural readability over forced repetition
5. Create output files in docs/inception/
6. Validate each step and report scores

# All 8 steps generated automatically
pi skill inception-workshop --mode batch --context "SessioFlow: Call-for-Papers platform"

Step Mapping

Step Ordering by Scenario

The numbered step sequence (1–8) is the canonical reference — it keeps file names, cross-links, and tool output stable. Do not renumber steps. However, the facilitation order should adapt to whether you are discovering a new product or documenting an existing one.

Greenfield (discovering a new product from scratch)

Run steps in sequential order — no swaps needed:

Brownfield (documenting or extending a live product)

Run order diverges from step numbers — Brainstorming (Step 6) runs before User Journey (Step 5):

Key differences from greenfield: Tradeoffs and Empathy Map drop in priority because the product already exists. Brainstorming (Step 6) moves before User Journey (Step 5) because you need to know what is already built before you can map how it connects.

When running incremental additions (e.g. adding a new persona to an existing product)

Steps 5 and 6 are tightly coupled — bounce between them freely. The distinction between the two scenarios above dissolves for small increments. Write the feature row in Step 5 first (it is faster), then let the journey in Step 6 validate and flesh it out.

Tradeoff Generator Mode (Step 2 Special)

What It Does

The Tradeoff Generator creates a consensus tradeoff board based on the product vision:

Key Components:

1. Individual Perspectives - Capture different stakeholder viewpoints
2. Final Consensus Board - STRICT 1-7 ranking (Golden Rule: one check per column)
3. Consensus Reasoning - Explain trade-offs and compromises made

Process

1. Analyze Vision - Read Step 1 output to understand goals, users, and constraints
2. Create Perspectives - Define PO, UX, and Tech Lead viewpoints
3. Build Consensus - Create strict 1-7 ranking with compromises
4. VERIFY GOLDEN RULE - Count X marks: must be EXACTLY 7 (one per priority column)
5. Validate - Check against validator criteria
6. Generate Output - Create filled tradeoff template

Output Structure

# Step 2: Tradeoffs

## 1. Individual Perspectives

| Role | Initial Priority #1 | Reasoning |
| :--- | :---: | :--- |
| Product Owner | Cost | Vision targets free-tier infrastructure |
| UX Advocate | Usability | Core differentiator vs manual tools |
| Tech Lead | Simplicity | Essential for self-hosting |

## 2. Final Consensus Trade-off Board

| Priority Rank | 1 (Most) | 2 | 3 | 4 | 5 | 6 | 7 (Least) |
| :--- | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
| **Usability** | **X** | | | | | | |
| **Cost** | | **X** | | | | | |
| **Simplicity** | | | **X** | | | | |

⚠️ **GOLDEN RULE CHECK**: Count all X marks above. There must be EXACTLY 7 (one per column).

## 3. Consensus Reasoning

The team agreed on Usability as #1 because...

Key trade-offs made:
- Cost > Scalability (MVP focus from vision)
- Simplicity > Flexibility (Non-technical users)

Usage

# Generate tradeoffs with AI debate simulation
pi skill inception-workshop --mode tradeoff-generator

Requirements:

• Step 1 (Product Vision) must be completed first
• Reads: docs/inception/1-product-vision-and-boundaries.md
• Outputs: docs/inception/2-tradeoffs.md

Note: This mode uses the agent to simulate stakeholder debate perspectives and create consensus tradeoffs directly.

Output Structure

docs/inception/
├── 1-product-vision-and-boundaries.md    # Step 1 output
├── 2-tradeoffs.md                        # Step 2 output
├── 3-personas/                           # Step 3 output folder
│   ├── README.md                         # Overview & Persona validation checklist
│   ├── 01-[name].md                      # Individual persona file
│   └── 02-[name].md                      # Individual persona file
├── 4-empathy-map.md                      # Step 4 output
├── 5-user-journeys/                      # Step 5 output folder
│   ├── README.md                         # Composite View & Feature Coverage check
│   ├── journey-01-[name].md              # Individual journey file
│   └── journey-02-[name].md              # Individual journey file
├── 6-brainstorming.md                    # Step 6 output
├── 7-features-and-sequencing.md          # Step 7 output
└── 8-mvp-canvas.md                       # Step 8 output

Validation Scoring

Each step is validated against criteria from the validator file:

Integration with Flow Documentation

After completing Step 5 (User Journey) and Step 7 (Features & Sequencing), use the create-flow-documentation skill to generate flow specifications:

pi skill inception-workshop --mode generate-flows --from-step 5

This will:

1. Read the completed user journeys from docs/inception/5-user-journeys/
2. Extract MVP features from Step 7 (docs/inception/7-features-and-sequencing.md)
3. Generate flow specifications using skills/specification/create-flow-documentation/templates/flows.md
4. Output to docs/product/bounded-contexts/{context}/flows/

Configuration

Compatibility

Requires:

• Standard Pi toolset (read, write, edit, bash)
• No external dependencies

Bundled Assets

This skill includes all required templates and validators as bundled assets:

Templates: (in templates/ directory)

• templates/1-product-vision-and-boundaries.md
• templates/2-tradeoffs.md
• templates/3-personas.md
• templates/4-empathy-map.md
• templates/5-user-journey-mapping.md
• templates/6-brainstorming.md
• templates/7-features-and-sequencing.md
• templates/8-mvp-canvas-definition.md

Validators: (in references/ directory)

• references/1-product-vision-boundary-validator.md
• references/2.2-tradeoff-validator.md
• references/3-personas-validator.md
• references/4-empathy-map-validator.md
• references/5-user-journey-validator.md
• references/6-brainstorming-validator.md
• references/7-features-and-sequencing-validator.md
• references/8-mvp-canvas-definition-validator.md

Tradeoff Generator:

• references/2.1-tradeoff-generator.md - AI debate simulation prompt

Additional References:

• references/0-inception-workshop.md - Complete workshop guide
• references/automate-inception.md - Automation patterns
• references/facilitate-step.md - Step facilitation details
• references/run-inception-chain.md - Chain execution guide

Output Directory

• docs/inception/ - Created automatically if not exists

Examples

Interactive Facilitation

> pi skill inception-workshop --mode facilitate

Skill: Inception Workshop Facilitator

Step 1: Product Vision & Boundaries

I've created the template at:
  docs/inception/1-product-vision-and-boundaries.md

Template Structure:
```markdown
# Step 1: Product Vision & Boundaries

## Elevator Pitch
- **Problem:** [Fill this]
- **Solution:** [Fill this]
...

Please fill this document and commit when ready. Let me know when you want validation.

[User fills and commits]

Ready for validation

Skill: Validating Step 1...

Validation Results: ✅ Elevator Pitch: Complete (5/5 components) ✅ Is/Is Not: Clear boundaries (4+ items per column) ⚠️ Metrics: Could be more specific

Score: 8.5/10

Step 1 is complete! Ready to proceed to Step 2: Tradeoffs? (yes/no)

### Tradeoff Generator (Step 2 Special)

pi skill inception-workshop --mode tradeoff-generator

Skill: Tradeoff Generator (AI Debate Simulation)

Reading: docs/inception/1-product-vision-and-boundaries.md Template: templates/2-tradeoffs.md Validator: references/2.2-tradeoff-validator.md

Simulating stakeholder debate...

🎭 Product Owner: "Based on the vision's goal of 'quick MVP launch', I prioritize Time to Market and Cost over Scalability."

🎭 User Advocate: "The 'Who' is 'busy event organizers' - they need Simplicity and Usability above all."

🎭 Tech Lead: "The 'Is/Is Not' says this is NOT a complex enterprise system, so I can accept lower Flexibility for faster delivery."

🎭 AgileCoach: "Remember the Golden Rule - only one check per column. Let's find consensus."

Generating consensus trade-off board...

✅ GOLDEN RULE VERIFIED: Exactly 7 X marks (one per column) ✅ Tradeoff analysis complete!

Generated: docs/inception/2-tradeoffs.md

Key Trade-offs Made:

• Cost > Scalability (MVP focus from vision)
• Simplicity > Flexibility (Non-technical users)
• Time to Market > Perfect Architecture (Quick launch goal)

Ready to proceed to Step 3: Personas? (yes/no)

### Batch Generation

pi skill inception-workshop --mode batch --context "SessioFlow: Call-for-Papers platform for event organizers"

Skill: Inception Workshop (Batch Mode)

Generating all 8 steps...

Step 1: Product Vision... ✅ Step 2: Tradeoffs... ✅ Step 3: Personas... ✅ Step 4: Empathy Map... ✅ Step 5: User Journey... ✅ Step 6: Brainstorming... ✅ Step 7: Features & Sequencing... ✅ Step 8: MVP Canvas... ✅

All steps completed! Files created in docs/inception/

Validation Summary:

• Step 1: 9/10
• Step 2: 8.5/10
• Step 3: 9/10
• Step 4: 8/10
• Step 5: 8.5/10
• Step 6: 9/10
• Step 7: 8.5/10
• Step 8: 9/10

Average Score: 8.7/10

### Generate Flows from Journey

pi skill inception-workshop --mode generate-flows --from-step 6

Skill: Generating Flow Specifications

Reading: docs/inception/5-user-journeys/ Reading: docs/inception/7-features-and-sequencing.md

Generating flows for MVP features:

• Journey 01: Setup Event... ✅
• Journey 02: Submit Proposal... ✅
• Journey 03: Review Sessions... ✅
• Journey 04: Acceptance & Logistics... ✅

Files created:

• docs/product/bounded-contexts/event/flows/journey-01-setup-event.md
• docs/product/bounded-contexts/submission/flows/journey-02-submit-proposal.md
• docs/product/bounded-contexts/review/flows/journey-03-review-sessions.md
• docs/product/bounded-contexts/scheduling/flows/journey-04-acceptance-logistics.md

All flows include:

• Sequence diagram (with error paths)
• Flowchart (decision points)
• State diagram (entity lifecycle)
• Step-by-step walkthrough
• Edge cases & validation rules

## Related Skills

- **Flow Generator** - Create detailed flow specifications from user stories
- **Business Rule Extractor** - Identify business rules and invariants from requirements
- **DDD Model Builder** - Generate domain model from inception outputs

---

<self_validation_gate>
Before ending your turn, verify every item below with concrete evidence (not just a checkmark). If any item cannot be cited with real evidence from this turn, DO NOT produce your final response — fix the violation first:
- [ ] **Template & Validator Read Evidence:** Exact `templates/*.md` and `references/*-validator.md` paths passed to `read_file` in this turn: `<paths>`
- [ ] **Output & Schema Evidence:** Exact `docs/inception/*.md` path(s) written with 100% of template tables/diagrams preserved and 0 leaked AI instructions: `<paths>`
- [ ] **Quality Rules Evidence:** Step 1 boundary counts per column (`IS: N, IS NOT: N, DOES: N, DOES NOT: N` — each `≥4`); Step 2 `X` mark count (`7/7`, one per column).
</self_validation_gate>

---

**Version:** 3.6.0 (Level 3 Execution Protocol)  
**Last Updated:** 2026-09-08  
**Changes from v3.4:**
- Swapped Step 5 and Step 6: User Journey is now Step 5, Brainstorming is now Step 6. Templates, validators, and output folders renamed accordingly (`5-user-journey-mapping.md`, `5-user-journeys/`, `6-brainstorming.md`). Greenfield facilitation now runs in sequential step order with no swaps. Brownfield is the scenario that diverges (Step 6 before Step 5).
- Added "Step Ordering by Scenario" section documenting greenfield (sequential), brownfield (Step 6 before Step 5), and incremental additions (bounce freely).
**Changes from v3.3:**
- Step 5: no tech features rule — features must describe user-facing outcomes, not implementation mechanics (Kafka events, crons, webhooks are excluded). Borderline tech features must be reframed as user outcomes before inclusion
**Changes from v3.2:**
- Step 5 template quadrant charts now use `classDef` + `:::className` for per-dot colors and HTML `<span>` swatches in legend tables so colors match the chart exactly
- Color scheme: Core=`#2ecc71`, Supporting=`#3498db`, Differentiating=`#e67e22`, Nice-to-Have=`#95a5a6`
- Quadrant placement rule: fixed 6-point grid (0.08 / 0.20 / 0.35 / 0.62 / 0.75 / 0.88) on both axes; overlaps are intentional; offset by ±0.04 only when two features share the exact same slot
- Legend format: shields.io color-only badge (no text) — `![](https://img.shields.io/badge/_%20_-hexcolor)` — renders a plain color swatch on GitHub and VS Code preview. Use the same hex as the `classDef`