Provides guided design frameworks for system architecture, API definitions, and data models prior to coding.

Install

mkdir -p .claude/skills/design-matteocervelli && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/15447" && unzip -o skill.zip -d .claude/skills/design-matteocervelli && rm skill.zip

Installs to .claude/skills/design-matteocervelli

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Design component architecture, REST API contracts, and data models (Pydantic schemas) before implementation. Use when planning how to build a feature, defining endpoints, or modeling data. Trigger on "design the architecture", "design the API", "data model", "how should I structure this feature".
297 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Intermediate

Key capabilities

  • Design component architecture.
  • Design REST API contracts.
  • Design data models using Pydantic schemas.
  • Auto-detect context from project files.
  • Provide progressive disclosure of design guidance.

How it works

The skill determines the design mode (architecture, API, or data) from arguments or project context, then provides guidance and writes design documents using templates. It follows a sequence for full design workflows: architecture, then data models, then API contracts.

Inputs & outputs

You give it
User request for architecture, API, or data model design, or project context
You get back
Design documents in `docs/architecture/` (e.g., `architecture-{feature}.md`, `api-{feature}.md`, `data-{feature}.md`)

When to use design

  • Design API architecture
  • Model data schemas
  • Structure new feature
  • Design REST endpoints

About this skill

Design — Architecture, API & Data Models

Unified design guidance for feature implementation. Auto-detects context from project files or accepts explicit mode.

Usage

/design                    # Auto-detect from project, show summary
/design architecture       # Component architecture, SOLID, patterns
/design api                # REST API design, auth, rate limiting
/design data               # Pydantic schemas, validators, relationships
/design full               # All three in sequence (architecture → data → API)

Workflow

Step 1: Determine Mode

Parse $ARGUMENTS for explicit mode. If no arguments provided, detect context:

DETECTED=$(bash "$HOME/.claude/skills/design/lib/design-detector.sh")

Map detection results to recommended mode:

  • pydantic,fastapi → suggest api + data
  • pydantic only → suggest data
  • fastapi or flask or django → suggest api
  • Generic Python/Node → suggest architecture
  • Unknown → show all summaries, let user choose

Step 2: Show Relevant Guidance

Based on detected or explicit mode, use progressive disclosure:

Level 1 — Summary (default, ~15 lines): Read and present the summary template for the chosen domain:

  • templates/architecture/summary.md
  • templates/api/summary.md
  • templates/data/summary.md

Level 2 — Patterns (on request for more detail, ~50 lines): Read the patterns template with decision guidance and examples:

  • templates/architecture/patterns.md
  • templates/api/patterns.md
  • templates/data/patterns.md

Level 3 — Full Reference (on request for complete SOP): Read the deep reference documents on demand:

  • reference/architecture-patterns.md + reference/component-design-guide.md
  • reference/api-design-guide.md + reference/function-design-patterns.md
  • reference/data-model-guide.md + reference/pydantic-patterns.md

Step 3: Handle Multiple Domains

If context suggests multiple domains (e.g., new feature with API + data):

  • Show summary for each detected domain
  • Let user choose which to dive deeper into
  • Or use /design full for the complete sequence

Step 4: No Context Detected

If detection returns "unknown" and no explicit mode:

  • Show all three summaries as overview
  • List available modes for user to choose

/design full Sequence

When running the full design workflow:

  1. Architecture first: Component boundaries, layers, patterns, DI
  2. Data models second: Pydantic schemas based on architecture components
  3. API contracts third: Endpoints and schemas using data models

Output uses templates/architecture-doc.md as the final document template.

Output

Write documents to docs/architecture/ in the project root:

SubcommandOutput Path
/design architecturedocs/architecture/architecture-{feature}.md
/design apidocs/architecture/api-{feature}.md
/design datadocs/architecture/data-{feature}.md
/design fulldocs/architecture/{feature}.md (combined)

Use templates/architecture-doc.md as the document template with sections:

  • Overview, Architecture Pattern, Component Design, Data Model, API Specification
  • Data Flows, Module Structure, Error Handling, Configuration
  • Testing Strategy, Security Considerations, Performance Considerations

Create docs/architecture/ if it doesn't exist.

Integration

Invoked by: User directly (/design), /story prp workflow (design phase) Invokes: Nothing directly — produces architecture docs consumed by prp-generator Connected skills:

  • /story create → requirements → /design → architecture doc → /story prp → PRP
  • /implementation consumes design output

Progressive Disclosure Rules

  • Never load reference/ docs unless user explicitly asks for full SOP or L3 detail
  • Default to L1 summary — enough to orient and decide
  • L2 patterns when user asks "how", "what pattern", "show me examples"
  • L3 full when user asks for "complete guide", "full reference", or specific deep topic

When not to use it

  • When the user needs implementation rather than design.
  • When the user does not want design documents written to `docs/architecture/`.
  • When the user explicitly asks for full reference documents without progressive disclosure.

Limitations

  • Does not directly invoke other tools for implementation.
  • Requires user input for mode selection if context is unknown.
  • Output documents are written to a specific `docs/architecture/` path.

How it compares

This skill offers structured, context-aware design guidance and document generation, which is more systematic than ad-hoc design processes.

Compared to similar skills

design side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
design (this skill)02moReviewIntermediate
backend-architect104moNo flagsAdvanced
python-pro02moNo flagsIntermediate
fastapi-templates5202moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry