Claude Code — Feature Showcase

Installed Plugins

Superpowers v6.4.1

Brainstorming, written plans, subagent-driven development with built-in review, systematic debugging, red/green TDD, git worktrees and verification before completion.

claude-plugins-official

Compound Engineering v3.30.5

36 skills from brainstorm to ship: planning, prototyping, code review, debugging, browser testing, PR babysitting, product strategy and compounding learnings.

EveryInc

Frontend Design

Distinctive, production-grade interfaces that avoid generic AI aesthetics: typography, color, layout and motion decisions made deliberately.

claude-plugins-official

Playwright

Microsoft's browser automation MCP server: open pages, click, fill forms and take screenshots to check UI work in a real browser.

claude-plugins-official

Context7

Live documentation for any library, framework or SDK, version-specific, so answers do not depend on stale training data.

claude-plugins-official

GitHub

Official GitHub MCP server: issues, pull requests, reviews, checks and code search, authenticated with the gh login.

claude-plugins-official

Code Review

Reviews a diff, branch or PR for correctness bugs at a chosen effort level, from a quick pass to a deep multi-agent cloud review.

claude-plugins-official

Swift LSP v1.0.0

SourceKit-LSP code intelligence, diagnostics and completions for Swift.

claude-plugins-official

Claude HUD v0.8.0

Configurable heads-up display for the session: context, usage, tools and agents at a glance.

jarrodwatts

Development Workflow Pattern

The full skill-driven lifecycle. Each stage is backed by a dedicated skill that enforces engineering rigor.

Step 1

Brainstorm

/brainstorming

→

Step 2

Write Plan

/writing-plans

→

Step 3

Execute

/executing-plans

→

Step 4

TDD

/test-driven-development

→

Step 5

Review

/ce-code-review

→

Step 6

Verify

/verification-before-completion

→

Step 7

Ship

/finishing-a-development-branch

Parallel Execution

Independent tasks are dispatched to parallel sub-agents via /dispatching-parallel-agents or /subagent-driven-development. Each agent works in isolation, then results merge back.

Git Worktrees

Feature work is isolated using /using-git-worktrees. Each worktree gets its own branch and working directory so parallel features never conflict.

Systematic Debugging

Bugs trigger /systematic-debugging before any fix is proposed. Root cause analysis first, then targeted fix, then verification.

Persistent Memory

File-based memory stores user preferences, feedback, project context, and external references across sessions. Indexed via MEMORY.md.

Configuration (settings.json)

settings.json

{
  "enabledPlugins": {
    "superpowers@claude-plugins-official":               true,
    "compound-engineering@compound-engineering-plugin":  true,
    "frontend-design@claude-plugins-official":           true,
    "playwright@claude-plugins-official":                true,
    "context7@claude-plugins-official":                  true,
    "github@claude-plugins-official":                    true,
    "code-review@claude-plugins-official":               true,
    "swift-lsp@claude-plugins-official":                 true,
    "claude-hud@claude-hud":                             true
  },
  "extraKnownMarketplaces": {
    "compound-engineering-plugin": { "source": { "source": "github", "repo": "EveryInc/compound-engineering-plugin" } },
    "claude-hud": { "source": { "source": "github", "repo": "jarrodwatts/claude-hud" } }
  },
  "effortLevel": "high",                              // max reasoning depth
  "statusLine": { "type": "command",
                  "command": "~/projects/claude-code-setup/sync-status.sh" },  // keeps this page current
  "skipDangerousModePermissionPrompt": true             // faster iteration
}

Available Skills (Slash Commands)

Superpowers

/brainstorming /writing-plans /executing-plans /test-driven-development /systematic-debugging /requesting-code-review /receiving-code-review /verification-before-completion /finishing-a-development-branch /dispatching-parallel-agents /subagent-driven-development /using-git-worktrees /writing-skills /diagnosing-superpowers

Compound Engineering — Workflow

/ce-brainstorm /ce-plan /ce-work /ce-code-review /ce-debug /ce-ideate /ce-compound /ce-compound-refresh /ce-commit /ce-commit-push-pr /ce-worktree /ce-resolve-pr-feedback /ce-babysit-pr /ce-simplify-code /ce-sweep /ce-handoff /lfg /wtf

Compound Engineering — Product, Design & Docs

/ce-prototype /ce-polish /ce-test-browser /ce-test-xcode /ce-dogfood /ce-doc-review /ce-proof /ce-explain /ce-strategy /ce-product-pulse /ce-pov /ce-noslop /ce-optimize /ce-bakeoff /ce-retune /ce-riffrec-feedback-analysis /ce-promote /ce-setup

Design & Documents

/frontend-design /artifact-design /dataviz /web-artifacts-builder /theme-factory /brand-guidelines /pptx /docx /xlsx /pdf /deep-research /doc-coauthoring

Framework & Utilities

/claude-api /code-review /security-review /simplify /run /init /loop /schedule /update-config /fewer-permission-prompts /keybindings-help /plugin-authoring /workflow-authoring /mcp-builder /skill-creator

Specialized Sub-Agents

Agents run in the background with isolated context, in parallel when the work is independent. Review personas now run inside the review skills (/code-review, /ce-code-review) instead of as standalone agents.

Built-in

Explore

Read-only fan-out search across many files; returns the conclusion, not the file dumps.

Plan

Software architect for implementation strategies, critical files and trade-offs.

general-purpose

Multi-step research and execution with the full tool set.

claude-code-guide

Answers questions about Claude Code, the Agent SDK and the Claude API.

fork

A copy of the current session with its full context, for a side task that should not block the main thread.

statusline-setup

Configures the status line.

Orchestration

Workflow

Deterministic multi-agent scripts: pipelines and parallel fan-out with verification, run in the background.

Agent teams

Named agents that message each other with SendMessage and are listed with ListAgents.

Worktree isolation

An agent can work in its own git worktree so parallel features never conflict.

Remote agents

Agents can run in a cloud environment and on schedules (routines).

Built-in Tools

File Operations

ReadWriteEditGlobGrepNotebookEdit

Execution

BashLSP

Web & Search

WebFetchWebSearchToolSearch

Task Management

TaskCreateTaskUpdateTaskListTaskGetTaskStopTaskOutput

Planning & Orchestration

AgentSkillEnterPlanModeExitPlanModeEnterWorktreeExitWorktree

Scheduling & Remote

CronCreateCronDeleteCronListRemoteTriggerScheduleWakeupMonitorAskUserQuestionWorkflowSendMessageArtifactPushNotification

API Integrations (MCP Servers)

Refero

Real-world UI and UX references from shipped products, used as design inspiration for landing pages and app screens.

Context7

Version-specific documentation for any library, framework or SDK.

GitHub

Repositories, issues, pull requests, reviews and checks through the official GitHub MCP server.

Playwright

Drives a real browser: navigate, click, fill and screenshot to verify UI work.

Gmail

Read, search, draft and organize email.

Google Calendar

Events, availability and meeting scheduling.

Google Drive

Search, read and create Drive files, Docs, Sheets and Slides.

Notion

Pages, databases and comments in Notion workspaces.

Claude Docs

Living documents created and edited together with Claude.

Pre-Approved Permissions

Commands auto-allowed without a confirmation prompt, configured in ~/.claude/settings.json.

python / python3 pip / pip3 install uv npm install / run / build / check npx bun mkdir / chmod tree / ls / find curl / ping timeout kill / pkill git status / diff / log / add / commit gh pdflatex / pandoc huggingface-cli WebSearch WebFetch

Key Design Decisions

High Effort, Always

effortLevel is set to "high" globally. Every task gets maximum reasoning depth — no shortcuts, no shallow passes.

Skill-First Workflow

Before any action, relevant skills are checked and invoked. Skills enforce discipline: brainstorm before build, verify before ship.

Dangerous Mode Enabled

Permission prompts are skipped for faster iteration. Pre-approved commands cover the common toolchain (Python, Node, git, gh). A local safety hook still scores risky shell commands outside bypass mode.

Multi-Marketplace Plugins

Draws from 3 plugin sources: Anthropic Official, EveryInc (Compound Engineering) and jarrodwatts (Claude HUD). Community + first-party.

Agent-Native Architecture

Background sub-agents, forks, agent teams and scripted workflows. Independent work runs in parallel, each agent with its own context.

Cross-Session Memory

Persistent file-based memory indexed by MEMORY.md. Stores user context, feedback, project state, and external references across conversations.

Personal Usage Dashboard

Lifetime Claude Code activity for swchoi1994. Snapshot as of 2026-04-17.

Less More

You’ve used ~289× more tokens than Animal Farm.