From harness-engineering
Implement a feature using Explorer+Workers parallel agents for maximum flow efficiency. Use when implementing a non-trivial feature in a single session, especially when 'use flow-parallel', 'implement in parallel', or 'Explorer+Workers' is mentioned. Do NOT use for simple single-file changes or debugging tasks.
npx claudepluginhub toru-oizumi/claude-harness-engineering --plugin harness-engineeringThis skill uses the workspace's default tool permissions.
Implement features at maximum throughput using the Explorer+Workers pattern: one Explorer gathers shared context, then multiple Workers execute design, test design, and security analysis in parallel.
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.
Implement features at maximum throughput using the Explorer+Workers pattern: one Explorer gathers shared context, then multiple Workers execute design, test design, and security analysis in parallel.
Phase 0: Branch creation
Phase 1a: Explorer (sequential) — gather shared context
Phase 1b: Workers × 3 (parallel) — design, test-design, security
Phase 2: Implementors × 3 (parallel) — code, test-code, security-check
Phase 2.5: Quality gate (build → security-lint → lint → format)
Phase 2.6: Test code generation (sequential)
Phase 3: Three Amigos review + commit
| Signal | Action |
|---|---|
| User says "use flow-parallel" | Run this skill |
| Feature touches 3+ files | Suggest this skill |
| UI + backend changes in same task | Use UI Flow |
| Pure backend/logic change | Use Non-UI Flow |
| Simple single-file fix | Use standard implementation instead |
Before starting:
If scope is too broad: decompose the task first.
| Indicator | Flow Type |
|---|---|
--ui flag | UI Flow |
--no-ui flag | Non-UI Flow |
| Keywords: screen, widget, UI | UI Flow |
| Keywords: domain, usecase, API | Non-UI Flow |
git checkout <base-branch> && git pull origin <base-branch>
git checkout -b feature/<name>
subagent_type: Explore
run_in_background: false
Output: Shared context (file summaries, patterns, ownership assignments) Timeout: Orange signal — narrow feature scope
Agent A (Design): general-purpose, background, max 8 turns
Agent B (Test): general-purpose, background, max 8 turns
Agent D (Security): general-purpose, background, max 8 turns
Pass shared context from Explorer to each Worker.
Agent E (Implementation): general-purpose, background
Agent F (Test Design): general-purpose, background, max 10 turns — creates Fakes + YAML scenarios
Agent G (Security): general-purpose, background, max 8 turns
project build
project lint && project auto-fix
project format
Signal levels: security violation → Red stop | lint > 10 → Orange | lint 1–9 → Yellow note
Agent F' (Test Code): general-purpose, foreground, max 12 turns
Converts Agent F's YAML scenarios to actual test files.
Review from Business / Development / Testing perspectives. Rejection → return to relevant phase.
quality-gate — Pre-PR quality checkspre-task-diagnostic — Readiness check before starting