Object Oriented Analysis and Design (OOAD)

Agile by Design — abd-ooad skill for turning raw material (specs, manuals, code, policies) into robust object-oriented domain models.

Quick Start

The core process is linear and methodical: Steps 0 → 1 → 2 (Scan → Extract → Refine).

Step 0: Domain Scan — Orientation, not extraction

• Read table of contents / section headings
• Identify 3–7 high-confidence anchors
• Flag suspected tensions/ambiguities
• Create Actor Registry to track concepts

Step 1: Extraction & Validation — Nouns, verbs, rules, states

• Read carefully for domain concepts
• Build candidate class list
• Validate against key scenarios

Step 2: Refinement — Relationships, responsibilities, invariants

• Add relationships and cardinality
• Encode domain rules into model
• Validate with walkthroughs

Where This Skill Fits

This skill handles model discovery, initial design, and diagram generation end-to-end. Diagram tooling is built in:

• scripts/drawio_cli.py — create, verify, relayout, fix-edge-styles, fix-shared-endpoints, fix-arrow-overlaps
• templates/ — domain model and domain realization templates
• content/parts/library/ — class diagram and sequence diagram layout rules
• runs/ — reference runs showing complete real-world OOAD output (format, notation, structure). Use as gold-standard examples before starting a new project run.

Reference Runs

runs/run-0/ contains the baseline reference for MM3E Heroes Handbook OOAD — a complete real session output:

• domain-scan-model.md / .drawio — Step 0 anchors + tensions in canonical format
• domain-scan-results.md — source map, anchor justification, tension table
• 1 - basics-checks-conditions/ — Phase 2 (domain-noun-verb.md) and Phase 3 (domain-raw-candidates.md) slice work
• progress/ — process and strategy checklists
• skill/ — snapshot of skill files (SKILL.md, process.md, templates) that governed this run
• README.md — reverse-engineered format conventions, notation rules, anchor/tension summary

Before starting a new project run: read runs/run-0/README.md and runs/run-0/domain-scan-model.md to align on the expected output format and notation.

Workspace Capability

When you invoke this skill, specify the project workspace where domain models will be created:

workspace: /path/to/project/directory

This workspace becomes the default location for:

• domain-scan-model.md — Step 0 (source map, anchors, tensions, actor registry)
• domain-scan-model.drawio — Step 0 diagram
• domain-noun-verb.md — Phase 2 per slice folder only (nouns / verbs / rules / candidate blocks). Do not append Phase 3 bucket tables to the end of this file.
• (Optional) domain-raw-candidates.md — Phase 3 integrated model: ### Class : << kind >> under anchor modules — templates/domain-raw-candidates-template.md. Tabular raw-candidates.md is deprecated (see templates/raw-candidates-template.md).
• Any subsequent step outputs

Authoring shapes for Phase 2/3 live under templates/ in this skill — do not paste those paths into slice artifacts.

Slice deliverables are domain content only. Do not write Source: / Slice: / OOAD phase: / registry Step / Upstream: / Pre-notation: / Integrated model: blocks in slice files — that is skill/process noise, not model output. In domain-raw-candidates.md, do not duplicate Phase 3 bucket tables or tack them onto domain-noun-verb.md — fold judgments into #### Note : with typed lines per templates/domain-raw-candidates-template.md.

Tensions in domain-raw-candidates.md: Cross-cutting ambiguities (e.g. registry T1 / T3) must appear under the #### Note : of each owning class as *[S# · Phase n]* Tension — … — [S# · Phase n] first, then Tension, same shape as Trace / Evidence (see templates/domain-raw-candidates-template.md). Do not use *Tension — [S# · Phase n]*. Do not add a closing tension index table at EOF.

If no workspace is provided, assume the current working directory.

How to Use This Skill

When you have domain material (specs, manual, code, policy docs) and need to extract an object-oriented model:

1. Provide the source material (text, code, spec excerpt, etc.)
2. Specify the workspace directory where outputs should be created
3. Ask Claude to "perform a Step 0 domain scan" or similar

Claude will then:

• Analyze the material using appropriate technique (spec, code, transcript, expert)
• Create domain-scan-model.md with source map, anchors, tensions, actor registry
• Create domain-scan-model.drawio with initial class diagram (anchors only)
• Output both files to your workspace

Continue with Step 1 (Extraction) and Step 2 (Refinement) for deeper modeling.

Key Output Artifacts

Slice artifacts: domain-noun-verb.md is required per slice (Phase 2). Phase 3 is an optional domain-raw-candidates.md — classes with << Entity >> / << ValueObject >> / << Process >> / << Policy >> / << Role >> / << Event >> under anchor modules. No separate bucket-table file and no Phase 3 tables appended under domain-noun-verb.md.

Always maintain both Markdown and diagram side-by-side where your project uses diagrams; keep them synchronized as you refine.

Complete OOAD Methodology

For the full 20-step walkthrough with detailed techniques, worked examples, ASCII diagram notation, and battle-tested patterns, see content/parts/process.md.

For Step 0 details by source type (specification, codebase, transcript, domain expert), see content/parts/domain-scan.md.

For diagram CLI usage and layout rules, see content/parts/library/using-diagram-cli.md.