Claude Settings

Run a technical UI quality audit across accessibility, performance, responsive behavior, theming, and design anti-patterns.

Perform a final UI quality pass for spacing, alignment, states, motion, copy, and visual consistency before shipping.

Run a guided browser-based QA flow against a local or staging URL using the persistent browser runtime when available.

Read-only UI quality analysis across accessibility, performance, responsive behavior, theming, and design anti-patterns.

Realign UI work to the repository's design system, tokens, spacing, and established component patterns.

Split an execution plan into independently executable vertical-slice task documents.

Internal Moonshot Architecture stage for writing C4 model artifacts and ADRs from accepted architecture decisions.

Internal Moonshot Architecture stage for extracting architecturally significant requirements and quality attribute scenarios from normalized requirements.

Detects code changes, auto-updates related docs (PROJECT.md, README, CHANGELOG, generated docs), and bootstraps project documentation structure.

Safely commit explicit Git changes and optionally close out Kernel project knowledge.

Use for advisory retrospective collection, daily pattern analysis, and local improvement proposal or issue draft generation.

Collect, review, and explicitly import project knowledge from Codex/Claude sessions and the current codebase.

Record decisions, issues, and handoff notes during or after work.

Simplifies and refines recently modified code for clarity, consistency, and maintainability while preserving exact functionality.

Update project memory and commit when the user explicitly wants both.

Create distinctive, production-grade frontend interfaces with high design quality. Use for React/web UI implementation and redesign work where visual direction matters.

Review planning artifacts for product value, timing, and scope control before execution.

Review planning artifacts for architecture integrity, dependency shape, and execution readiness.

Runs runtime/browser verification for web projects using URL health checks and optional E2E commands.

Analyzes agent failures and suggests system improvements (context/skills/agents/rules).

Checks essential information and project status before starting work and emits readiness signals for orchestration.

Gather persistent design context for a project and save it to .impeccable.md so future UI work has real product context.

Keep product-definition work moving by recording ambiguity as assumptions or blockers instead of stopping the workflow.

Review non-trivial implementation changes for quality and regression risk before completion or merge.

Use when the user asks for a rich explanation of a code change, diff, branch, or PR. Produces HTML output.

Default Codex command-skill entrypoint for Moon Relay Kernel task routing and adaptive workflow execution. Selecting this skill activates Kernel workflow for that task; it does not force unselected ordinary Codex tasks into Kernel.

Reviews code for security vulnerabilities. Use when security concerns are detected or before deploying critical changes.

Produce product-definition artifacts and a provenance-bound Task Contract Seed before Kernel execution.

Analyze the current repository and create or refresh the project bootstrap docs (`.claude/PROJECT.md`, `workflow/README.md`, `docs/design/README.md`, `docs/glossary/README.md`, `docs/daily/README.md`, `TEST_GUIDE.md`, `docs/analysis/README.md`) to match the project.

Runtime-adaptive plan validation skill (Plan Reviewer rubric). Use after writing context.md for complex feature/refactor work.

Run required checks and decide whether implementation has enough evidence to be treated as complete.

Parses design assets (PDF, CSS) to generate a draft design spec.

Resolves build and compilation errors. Use when build fails or type errors occur.

Create, refresh, and organize project-scoped account-root master and phase plans for phase-based work.

Complete Moonshot Relay account-root installation after installing the repository with Agent Skills CLI.

Use when work is still at the idea-to-plan stage and needs bounded product-definition artifacts before implementation.

코드베이스 전수 분석 및 인터랙티브 개발자 가이드(슬라이드북) 제작 스킬. 주니어 개발자 눈높이의 "Why(기술적 이유)" 중심 해설, C4 인프라 토폴로지(5대 망 분리), E2E 트랜잭션 바운더리, 100% 전수 소스코드 리뷰(라인별 사수 1:1 코칭 노트 매핑), 하단 빈 공간(Dead Space) 제로 2-Tier 레이아웃, 3대 엣지 시나리오, 16대 실무 함정 매트릭스, 자가진단 퀴즈 스테이션을 포함한 단일 Standalone HTML을 자동 생성합니다.

Performs implementation in the chain and records completion state and changed files in analysisContext. Use during implementation.

Convert a PRD or existing codebase objective into an evidence-grounded architecture design package before implementation planning.

Use for bounded implementation work that already has enough context and does not need the phase harness.

Use for large, phase-based, or long-running implementation work that should run from a prepared plan package.

Runs parallel Agent Teams for review, research, verification, and implementation workflows.

Review a product-definition artifact only for next-stage readiness, not stylistic quality.

Checks whether a downstream project has a usable PROJECT.md contract and routes to project-md-refresh when it does not.

Review UI code for Web Interface Guidelines compliance. Use when asked to "review my UI", "check accessibility", "audit design", "review UX", or "check my site against best practices".

Produce architecture artifacts and a provenance-bound Task Contract Seed before Kernel execution.

Build and query the regenerable account-root Code Index for the current Kernel project.

Checks whether a downstream task has the minimum context contract and routes to context-builder when it does not.

Synthesize implementation-ready UI design specs and planning/policy specs from live Figma files, Figma URLs, node IDs, screenshots, or large Figma planning boards. Use when a user asks to organize Figma design specs, planning specs, product policy specs, interaction states, screen behavior, or update an existing spec from added Figma cases.

Classifies a user request into task types (feature, modification, bugfix, refactor) and extracts intent keywords. Use at the start of PM analysis.

Explicitly build and refresh the current project's local MemoryGraph knowledge graph.

Promote explicitly approved reusable project or harness knowledge after review, replay, rollback, and release evidence gates.

Pre-implementation execution gate based on four principles (think before coding, simplicity first, surgical changes, goal-driven execution).

Skill-level phase execution adapter that routes prepared phase work to in-session coordinator execution, with delegated-terminal kept as legacy compatibility only.

Enforce red-green-refactor evidence before behavior-changing implementation work.

Determines phase and execution chain based on analysisContext (task type, complexity, signals). Use after uncertainty detection.

Use before implementation in strict runs to confirm isolated workspace setup and baseline evidence.

Maintain the Moonshot harness and downstream .claude installs. Use when adopting external skill or harness patterns, improving moonshot-orchestrator or moonshot-phase-runner policy, fixing runtime parity or completion-gate fixtures, updating commit-memory defaults, or syncing shared .claude assets into target projects while preserving project-local PROJECT.md, memory, settings, logs, and task docs.

Internal Moonshot Architecture stage for reviewing whether an architecture package is ready for implementation handoff.

Checks whether a downstream project has enough verification contract information and blocks strict completion when it does not.