Agent Skills: Domain Storytelling Skill

Domain Storytelling Skill

Overview

Domain Storytelling is a collaborative modeling technique that captures business processes through pictographic stories. This skill guides AI-assisted domain storytelling sessions that can feed into event storming and bounded context discovery.

Key Principle: Stories are told from the perspective of domain experts, using their language and understanding.

When to Use This Skill

Keywords: domain storytelling, pictographic, actors, work objects, activities, AS-IS, TO-BE, business workflow, domain modeling, story collection, bounded context discovery, requirements gathering

Use this skill when:

• Gathering requirements from domain experts
• Understanding existing business workflows (AS-IS)
• Designing future state processes (TO-BE)
• Onboarding team members to a domain
• Preparing for event storming sessions
• Identifying bounded context candidates
• Building a ubiquitous language glossary

Story Types

AS-IS Stories

Document how things work today:

• Current state processes
• Existing pain points
• Workarounds and exceptions
• Real behavior (not idealized)

When to use: Understanding current state, identifying problems, baseline before changes.

TO-BE Stories

Document how things should work:

• Desired future state
• Improved processes
• New capabilities
• Idealized flow (but achievable)

When to use: Requirements gathering, designing solutions, communicating vision.

Pictographic Language

Domain Storytelling uses simple pictographic elements:

| Element | Symbol | Description | | --- | --- | --- | | Actor | 🧑 (stick figure) | Person or system that performs activities | | Work Object | 📄 (document) | Data, documents, or physical items exchanged | | Activity | ➡️ (arrow with verb) | Action performed by an actor | | Sequence | ① ② ③ | Numbered order of activities | | Annotation | 💬 (note) | Additional context or explanation |

Detailed notation guide: See references/pictographic-notation.md

AI-Assisted Story Collection Protocol

Phase 1: Story Collection

Goal: Gather the narrative from the user in their own words.

Prompts:

• "Tell me about a typical [process] from start to finish"
• "Walk me through what happens when [trigger event]"
• "Who is involved and what do they do?"

Capture:

• Who does what (actors and activities)
• What they work with (work objects)
• In what order (sequence)
• Any variations or exceptions

Phase 2: Story Refinement

Goal: Explore edge cases and variations.

Prompts:

• "What happens if [X] fails or is unavailable?"
• "Are there any special cases or exceptions?"
• "What's the most common path vs rare paths?"
• "What frustrates you about this process?"

Capture:

• Alternative flows
• Error handling
• Pain points
• Implicit knowledge

Phase 3: Actor Identification

Goal: Map all participants in the story.

Prompts:

• "Who else is involved that we haven't mentioned?"
• "Are there any systems or external parties?"
• "Who approves, reviews, or audits?"

Capture:

• Human actors (by role, not name)
• System actors (internal and external)
• Actor responsibilities

Phase 4: Work Object Cataloging

Goal: Identify all data and documents exchanged.

Prompts:

• "What information is passed between actors?"
• "What documents or forms are used?"
• "What data is created, updated, or referenced?"

Capture:

• Documents and forms
• Data entities
• Physical items (if applicable)
• Work object lifecycle

Phase 5: Boundary Discovery

Goal: Find bounded context candidates.

Analysis:

• Where does terminology change?
• Which actors work together closely?
• What work objects belong together?
• Where are the natural handoff points?

Output: Potential bounded context candidates for event storming.

Detailed boundary discovery: See references/boundary-discovery.md

Story Output Format

Text Representation

# Domain Story: [Story Name]

**Type:** AS-IS | TO-BE
**Domain:** [Domain Name]
**Date:** YYYY-MM-DD

## Narrative Summary

[2-3 sentence plain language summary]

## Story Sequence

① **Customer** submits **Order Form** to **Sales Rep**
② **Sales Rep** validates **Order Form** using **Product Catalog**
③ **Sales Rep** creates **Order** in **Order System**
④ **Order System** notifies **Warehouse** with **Pick List**
⑤ **Warehouse Staff** picks items using **Pick List**
...

## Actors

| Actor | Type | Responsibilities |
| --- | --- | --- |
| Customer | Human (External) | Initiates orders |
| Sales Rep | Human (Internal) | Validates and enters orders |
| Order System | System (Internal) | Order management |
| Warehouse | System (Internal) | Inventory and fulfillment |

## Work Objects

| Work Object | Type | Used By | Description |
| --- | --- | --- | --- |
| Order Form | Document | Customer, Sales Rep | Customer order request |
| Product Catalog | Reference | Sales Rep | Available products |
| Order | Data | Order System | Validated order record |
| Pick List | Document | Warehouse Staff | Items to pick |

## Annotations

- [Note 1]: Exception handling for out-of-stock items
- [Note 2]: Peak season requires additional staff

Mermaid Diagram (Optional)

sequenceDiagram
    participant C as Customer
    participant SR as Sales Rep
    participant OS as Order System
    participant W as Warehouse

    C->>SR: ① submits Order Form
    SR->>SR: ② validates using Product Catalog
    SR->>OS: ③ creates Order
    OS->>W: ④ notifies with Pick List
    W->>W: ⑤ picks items

Story templates: See references/story-templates.md

Integration with Event Storming

Domain stories naturally feed into event storming:

| Story Element | Event Storming Element | | --- | --- | | Activity | Command or Event | | Actor | Actor (yellow sticky) | | Work Object | Aggregate or Read Model | | Sequence | Timeline ordering | | Boundary | Bounded context candidate |

Workflow:

Domain Storytelling (understand "what happens")
    ↓
Event Storming (design "how it happens")
    ↓
Bounded Contexts → Modular Architecture

To proceed to event storming: Invoke the enterprise-architecture:event-storming skill with collected stories as input.

Facilitation Modes

Interactive Mode (Recommended)

The skill guides an interactive conversation with the user:

1. Ask open-ended questions
2. Capture responses as story elements
3. Reflect back for validation
4. Iterate until story is complete

Quick Mode

Rapid story capture from user-provided narrative:

1. User provides full narrative
2. Skill extracts actors, work objects, activities
3. Skill structures into story format
4. User validates and refines

Document Mode

Extract stories from existing documentation:

1. Read existing process documents
2. Extract story elements
3. Structure into story format
4. Validate with user

Glossary Building

As stories are collected, build a domain glossary:

| Term | Definition | Context | Aliases | | --- | --- | --- | --- | | Order | A validated customer request for products | Sales | Purchase Order, PO | | Pick List | List of items to retrieve from warehouse | Warehouse | Picking Ticket |

Glossary purpose:

• Establishes ubiquitous language
• Identifies term collisions (same word, different meanings)
• Documents domain knowledge
• Supports bounded context discovery

Best Practices

DO

• Use domain expert's language, not technical jargon
• Capture stories at the right granularity (not too detailed)
• Include exceptions and variations
• Number activities sequentially
• Document annotations for implicit knowledge
• Build glossary as you go

DON'T

• Impose technical terminology
• Skip edge cases and exceptions
• Assume you understand without asking
• Mix AS-IS and TO-BE in same story
• Forget to validate with domain expert

References

• references/pictographic-notation.md - Detailed notation guide with examples
• references/story-templates.md - YAML headers, output formats
• references/boundary-discovery.md - Finding bounded contexts from stories

Related Skills

• event-storming - Design "how it happens" after understanding "what happens"
• modular-architecture - Implement bounded contexts discovered from stories
• adr-management - Document significant decisions discovered during storytelling

Last Updated: 2025-12-22

User-Facing Interface

When invoked directly by the user, this skill creates domain stories.

Execution Workflow

1. Parse Arguments - Extract journey name and type (as-is or to-be). If no name provided, ask the user what business process to capture. Default type is as-is.
2. Spawn Story Facilitator - Launch the story-facilitator agent to guide the interactive session:
   • Collect the story (actors, work objects, activities, sequence)
   • Refine with edge cases and variations
   • Identify actors (human roles and systems)
   • Catalog work objects (documents, data, physical items)
   • Structure with numbered activities and Mermaid sequence diagram
3. Discover Boundaries - Identify bounded context candidates from terminology changes, actor groupings, and natural handoff points.
4. Save Results - Save to docs/domain-stories/[journey-name]-[type]-[date].md (or custom --dir).
5. Suggest Follow-Ups - Recommend TO-BE stories, related processes, event storming for domain model design.

Version History

• v1.0.0 (2025-12-26): Initial release