From klair-legacy
Guides users through creating comprehensive technical specifications using spec-driven development methodology. Use when users need to document features, requirements, or system designs before implementation. Helps structure specifications with proper sections, requirements, and technical details.
npx claudepluginhub ai-builder-team/ai-builder-plugin-marketplace --plugin klair-legacyThis skill uses the workspace's default tool permissions.
This skill guides you through creating comprehensive technical specifications following spec-driven development best practices. You'll receive a complete, well-structured spec document that aligns stakeholders before implementation begins. The process is highly collaborative—the agent will discuss each section with you to understand requirements, get your approval, then write it to the spec fil...
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
This skill guides you through creating comprehensive technical specifications following spec-driven development best practices. You'll receive a complete, well-structured spec document that aligns stakeholders before implementation begins. The process is highly collaborative—the agent will discuss each section with you to understand requirements, get your approval, then write it to the spec file before moving to the next section.
Use this skill when users need to create a technical specification, write a feature spec, document requirements, draft a design doc, or formalize a project proposal. Specifications created by this skill are stored within feature folders at features/<domain>/<feature-id>/specs/.
This skill generates comprehensive technical specifications with the following characteristics:
--- separators between major sectionschecklist.md filefeatures/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.mdIMPORTANT: Do NOT read other existing specs in the repository. There are varying standards and approaches being tested. ONLY follow the template in template.txt for structure and guidelines.
IMPORTANT: Features must exist before specs can be created. Use the feature-workflow skill to create features first.
Locate Existing Feature
features/index.md to get list of domainsls features/<domain-folder>/FEATURE.md file: Check if features/<domain>/<feature-id>/FEATURE.md existsDetermine Spec Number and Name
features/<domain>/<feature-id>/specs/^\d{2}- (e.g., 01-, 02-)highest + 1, zero-padded to 2 digits (e.g., 01, 02, ... 99)01phase-1-0-foundation, accordion-refactor, api-integration<number>-<spec-name> (e.g., 04-accordion-refactor)features/<domain>/<feature-id>/specs/<number>-<spec-name>/ doesn't existConfirm Spec Location
features/<domain>/<feature-id>/specs/<number>-<spec-name>/. Proceed?"Create Directory Structure
mkdir -p features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp
Initialize spec.md with metadata
gh api user --jq '.login' to get the current user's GitHub ID for the Owner fieldfeatures/<domain>/<feature-id>/specs/<number>-<spec-name>/spec.md with title and metadata:# [Spec Name from user]
**Status:** Draft
**Created:** [Today's date YYYY-MM-DD]
**Last Updated:** [Today's date YYYY-MM-DD]
**Owner:** [GitHub ID from gh api user command]
---
Check if research has already been provided by the user.
Option A: Research Already Provided
If the user has already provided feature research (e.g., in a file, in their message, or from a previous exploration session):
features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md for referenceOption B: No Research Provided - Conduct Research Directly
If no research has been provided, conduct the research yourself using Glob, Grep, and Read tools:
Find relevant files and patterns:
Identify similar implementations:
Map dependencies and integration points:
Document findings in features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md:
Based on the research, identify any HIGH-LEVEL clarifications needed:
DO NOT ask about:
These details should be deferred to the implementation phase.
Create all section drafts at once using parallel Task calls.
Launch 4 parallel general-purpose agents (one per section) using a SINGLE message with 4 Task tool calls:
Agent 1: Overview Draft
- subagent_type: "general-purpose"
- model: "haiku"
- description: "Draft Overview section"
- prompt: "Read features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md. Based on the research and our discussion about [feature], draft the Overview section following the template in .claude/skills/creating-specs/template.txt. Write your draft to features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/overview.md. Be concise and cover what is being built and how it works. Do NOT include a paragraph explaining why the implementation is needed or its benefits. Return 'DONE' when complete."
Agent 2: Out Of Scope Draft
- subagent_type: "general-purpose"
- model: "haiku"
- description: "Draft Out Of Scope section"
- prompt: "Read features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md. Based on the research and our discussion about [feature], draft the Out Of Scope section following the template in .claude/skills/creating-specs/template.txt. Write your draft to features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/out-of-scope.md. Include numbered items with rationale for each exclusion. Be concise - cover only what's needed. Return 'DONE' when complete."
Agent 3: Functional Requirements Draft
- subagent_type: "general-purpose"
- model: "haiku"
- description: "Draft Functional Requirements section"
- prompt: "Read features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md. Based on the research and our discussion about [feature], draft the Functional Requirements section following the template in .claude/skills/creating-specs/template.txt. Write your draft to features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/requirements.md. Use FR1/FR2/FR3 format. Each FR MUST include a Success Criteria. Be concise - cover only what's needed. Return 'DONE' when complete."
Agent 4: Technical Design Draft
- subagent_type: "general-purpose"
- model: "haiku"
- description: "Draft Technical Design section"
- prompt: "Read features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec-research.md. Based on the research and our discussion about [feature], draft the Technical Design section following the template in .claude/skills/creating-specs/template.txt. Write your draft to features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/technical-design.md. Include subsections: Files to Reference, File(s) to Modify, Data Flow. Return 'DONE' when complete."
After all 4 agents complete:
CRITICAL: Review sections in this EXACT order. Do NOT skip ahead.
Section Order:
For EACH section (in order):
Present Draft: Read tmp/<section>.md and present to user
Iterate Until Approved:
tmp/<section>.md as requestedWrite to Main Spec:
features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec.md---) after the section (except for last section)Check for Understanding Changes (Self-Reflection):
State Tracking:
Enforcement Example:
[Currently reviewing Overview]
User: "Let's skip to Technical Design"
Assistant: "Let's finish reviewing the Overview section first before moving to Technical Design. Are you ready to approve the Overview, or would you like to make changes?"
After all sections are approved and written to spec.md:
Read the final features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec.md
Generate comprehensive checklist.md based on:
Format checklist as:
# Implementation Checklist
## Data Layer
- [ ] [Task from requirements or technical design]
## Business Logic
- [ ] [Task from requirements or technical design]
## UI/UX
- [ ] [Task from requirements or technical design]
## Code Quality
- [ ] Run type checker and fix all errors
- [ ] Follow existing code patterns
Write to features/<domain>/<feature-id>/specs/<number>-<spec-name>/checklist.md
Ask about tmp/ folder:
tmp/ folder with section drafts?"features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/ as-is for referencefeatures/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/ entirelyExecute user's choice:
rm -rf features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmpSummary:
features/<domain>/<feature-id>/specs/<number>-<spec-name>/spec.mdfeatures/<domain>/<feature-id>/specs/<number>-<spec-name>/checklist.mdAfter spec and checklist are complete, automatically update the feature's Changelog.
Read FEATURE.md
features/<domain>/<feature-id>/FEATURE.mdPrepare Changelog Entry
| {YYYY-MM-DD} | [<number>-<spec-name>](specs/<number>-<spec-name>/spec.md) | {Brief description} |Update FEATURE.md
Verify Update
Example Changelog Entry:
| 2025-12-15 | [04-phase-1-0-foundation](specs/04-phase-1-0-foundation/spec.md) | Implement foundation for table refactoring |
See template.txt for the complete spec template with guidelines and examples for each section. The template includes 5 core sections:
features/<domain>/<feature-id>/specs/<number>-<spec-name>/tmp/spec-research.md, not in tmp/---) to separate major sections