design

Design software features and architectures, then produce an implementation-ready plan handoff.

7|5|Updated Jul 22, 2025
One-click install
npx skills add https://github.com/nebius/nebius-ps-services --skill design-nebius
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: design
Source: https://github.com/nebius/nebius-ps-services/tree/main/skills/design
Command: npx skills add https://github.com/nebius/nebius-ps-services --skill design-nebius

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a feature idea or architecture question into a concrete, evidence-backed design and implementation plan is slow and error-prone, especially when technology choices, AI subsystem behavior, and existing code constraints must all be reconciled before coding starts. ## Core Features & Use Cases - Seven-phase design workflow: Understand requirements, inspect the existing system (brownfield) or confirm greenfield context, research unfamiliar technologies, delegate stack and AI-subsystem decisions, design the solution, review it against system-design rules, and produce a /plan handoff. - Scoped specialist delegation: Routes technology due diligence to research, application-stack choices to app-stack, AI subsystem behavior and policy to ai-agent-design, AI component selection to ai-stack, and checklist review to system-design-rules without reopening fixed decisions. - AI capability classification: Classifies every AI-enabled capability as deterministic code, a direct model call, a deterministic workflow, or an agent, following the principle of deterministic where possible and agentic where necessary. - Use Case: You need to add a new API-backed feature to an existing service. The skill inspects the repository, confirms the fixed stack, designs a vertical end-to-end slice, compares alternatives, and emits a /plan handoff another session can implement directly. ## Quick Start Ask the agent to use the design skill to turn your feature request into an evidence-backed design and implementation plan before any code is written.

Frequently Asked Questions about design

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design a new feature before writing code?

Invoke the design skill with your requirements and constraints. It inspects the existing system, researches unfamiliar technologies, designs components and boundaries, compares alternatives, and produces a /plan handoff ready for implementation.

How do I choose between deterministic code, direct model calls, and agents?

The skill classifies each AI capability using a decision tree: deterministic code when no model judgment is needed, a direct call when one request suffices, a deterministic workflow when code owns the steps, and an agent only when the model must choose actions or next steps.

When should I use design instead of troubleshoot?

Use design when the causal mechanism is already proven and the remediation changes architecture, interfaces, data ownership, or cross-component workflows. Use troubleshoot when the failure mechanism is unknown or the repair stays inside one existing private boundary.

Can the design skill implement code or create project scaffolding?

No. It never writes production code, tests, migrations, or scaffolding. It ends with a /plan handoff, and may optionally emit a one-way scaffold handoff for an explicit scaffold-project invocation after design approval.

Does the design skill work for greenfield projects with no code?

Yes. After confirming no relevant source exists, it skips codebase discovery and designs from requirements, research-backed technology choices, and the intended deployment context, naming the initial project structure and bootstrap order.