Use when adding new error messages to React, or seeing "unknown error code" warnings.
npx skills add 30eggis/claude-frontend-skills --skill "spec-it"
Install specific skill from multi-skill repository
# Description
Frontend specification generator. Transforms PRD/requirements into comprehensive frontend specs with wireframes, component specs, and test plans.
# SKILL.md
name: spec-it
description: "Frontend specification generator. Transforms PRD/requirements into comprehensive frontend specs with wireframes, component specs, and test plans."
allowed-tools: AskUserQuestion, Skill
argument-hint: "[mode] [--resume
permissionMode: bypassPermissions
spec-it: Mode Router
Intelligent router that guides users to the appropriate spec-it mode based on their needs.
Trigger Keywords
This skill activates on semantic matches for:
- "frontend spec", "UI spec", "design spec"
- "PRD to spec", "requirements to design"
- "wireframe", "component spec", "test spec"
- "spec-it" (explicit)
Workflow
Step 1: Check for Direct Mode Specification
IF args contains "stepbystep" OR "step-by-step" OR "manual":
Skill(spec-it-stepbystep, args)
STOP
IF args contains "complex" OR "hybrid" OR "milestone":
Skill(spec-it-complex, args)
STOP
IF args contains "auto" OR "automation" OR "full-auto":
Skill(spec-it-automation, args)
STOP
IF args contains "fast" OR "quick" OR "rapid":
Skill(spec-it-fast-launch, args)
STOP
IF args contains "--resume":
# Resume needs to know which mode - check session _meta.json
# New structure: .spec-it/{sessionId}/(plan|execute)/_meta.json
Read: .spec-it/{sessionId}/plan/_meta.json OR .spec-it/{sessionId}/execute/_meta.json
route to appropriate skill based on mode stored in meta
STOP
Step 2: Ask User for Mode Selection
AskUserQuestion: "Which spec-it mode would you like to use?"
Options:
1. "Step-by-Step (Recommended for learning)"
- Chapter-by-chapter approval
- Maximum control
- Best for: Small projects, first-time users
2. "Complex/Hybrid"
- 4 milestone approvals
- Auto-validation between milestones
- Best for: Medium projects
3. "Full Automation β Execute"
- Minimal approval (only final review)
- Auto-proceeds to implementation (spec-it-execute)
- Best for: Large projects, experienced users
4. "Fast β Execute (Recommended for prototyping)"
- Skip brainstorming, components, tests
- Quick wireframes with design trends
- Auto-proceeds to implementation
- Best for: Rapid prototyping, design validation
Step 3: Route to Selected Mode
CASE selection:
"Step-by-Step" β Skill(spec-it-stepbystep)
"Complex/Hybrid" β Skill(spec-it-complex)
"Full Automation β Execute" β Skill(spec-it-automation) # Auto-executes after spec
"Fast β Execute" β Skill(spec-it-fast-launch) # Auto-executes after spec
Mode Comparison
| Feature | Step-by-Step | Complex | Automation | Fast |
|---|---|---|---|---|
| Approvals | Every chapter | 4 milestones | Final only | Final only |
| Control | Maximum | Balanced | Minimal | Minimal |
| Speed | Slowest | Medium | Fast | Fastest |
| Learning | Best | Good | Limited | None |
| Project Size | Small | Medium | Large | Prototype |
| Auto-Execute | No | No | Yes | Yes |
| Components | Full | Full | Full | Skipped |
| Tests | Full | Full | Full | Skipped |
Direct Invocation
For experienced users who know which mode they want:
# Step-by-Step mode
/spec-it-stepbystep
# Complex/Hybrid mode
/spec-it-complex
# Full Automation mode (auto-executes)
/spec-it-automation
# Fast mode (auto-executes)
/spec-it-fast-launch
# Resume any mode
/spec-it-stepbystep --resume 20260130-123456
/spec-it-complex --resume 20260130-123456
/spec-it-automation --resume 20260130-123456
/spec-it-fast-launch --resume 20260130-123456
Output
All modes produce the same output structure:
.spec-it/{sessionId}/plan/
βββ _meta.json, _status.json
tmp/
βββ 00-requirements/
βββ 01-chapters/
βββ 02-screens/
β βββ wireframes/
β βββ [html/, assets/] # If Stitch mode
βββ 03-components/
βββ 04-review/
βββ 05-tests/
βββ 06-final/
βββ final-spec.md
βββ dev-tasks.md
βββ SPEC-SUMMARY.md
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.