detailed-design

Generates per-screen detailed design documents with mermaid sequence diagrams from project specs.

Updated Aug 1, 2026
One-click install
npx skills add https://github.com/Samind-Samind/projexa --skill detailed-design-samind-samind
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: detailed-design
Source: https://github.com/Samind-Samind/projexa/tree/main/.claude/skills/detailed-design
Command: npx skills add https://github.com/Samind-Samind/projexa --skill detailed-design-samind-samind

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Writing consistent, traceable detailed design documents for every screen or feature of a system is slow and error-prone, especially when each document must include sequence diagrams, error flows, and traceability back to requirements. This Skill guides an interactive workflow that synthesizes existing specs, prototypes, tech-stack decisions, and acceptance criteria into per-screen conceptual design documents. ## Core Features & Use Cases - Per-screen Detailed Design Documents: Creates or updates one Markdown file per screen/feature under docs/02-design/02-technical/detailed-design/, each containing at least one mermaid sequenceDiagram. - AI Component Rule Enforcement: For screens involving AI components (TOR Parser, Design Analyzer, Test Generator, Document Writer), it enforces rules such as JSON-only AI output, source references, and a mandatory human review gate before any data is persisted. - Interactive Scope and Detail-Level Selection: Uses structured questions with at least three options (with pros and cons) to confirm scope, update mode (create, patch, or rewrite), and diagram detail level before writing any file. - Use Case: A system analyst asks to produce detailed designs for the MVP screens; the Skill reads the screen specs, backlog, system design document, and prototypes, confirms the plan, then delegates file writing to a dedicated sub agent. ## Quick Start Ask the assistant to create detailed design documents with sequence diagrams for the MVP screens based on the existing specs and prototypes.

Frequently Asked Questions about detailed-design

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

FAQPage Schema
How do I generate detailed design documents with sequence diagrams?▼

Specify the screens or features by SCR code, name, or module, and the Skill reads the matching specs, backlog, and system design document. It confirms scope and diagram detail level with you, then produces one Markdown file per screen with at least one mermaid sequenceDiagram.

What inputs are required before creating a detailed design document?▼

Each screen in scope needs a spec file under docs/01-requirements/01-spec/ plus the backlog. The Skill also uses the system design document, tech-stack.md if present, prototypes, and acceptance criteria when available.

How are AI components handled in sequence diagrams?▼

Screens involving AI components must show AI returning JSON only, include source references or confidence notes, and contain a human review step before any data is saved. Diagrams may never show AI writing directly to the data layer or generating document files itself.

Can the Skill update an existing detailed design file instead of rewriting it?▼

Yes. When a target file already exists, the Skill asks whether to patch only changed sections, rewrite the whole file, or skip that screen. It never deletes or moves existing content without explicit user confirmation.

Why does the Skill delegate file writing to a sub agent?▼

The main loop handles all design decisions and user interaction, while the detailed-design-writer sub agent only writes the finalized content to files. This separation prevents the sub agent from making unauthorized design decisions.