help

<tool_restrictions>

MANDATORY Tool Restrictions

BANNED TOOLS — calling these is a skill violation:

• EnterPlanMode — BANNED. This skill outputs information directly.
• ExitPlanMode — BANNED. You are never in plan mode.
• AskUserQuestion — BANNED. This is a read-only info dump, not interactive. </tool_restrictions>

/arc:help

Show every Arc command with context-aware relevance annotations.

---

Step 1: Gather Context (quick, parallel)

Run these checks in parallel. Keep it fast — no deep exploration.

# What framework/stack?
ls package.json next.config.* vite.config.* nuxt.config.* 2>/dev/null | head -5

# Arc artifacts?
ls docs/vision.md docs/arc/specs/*.md docs/arc/plans/*.md 2>/dev/null | head -10

# Design docs?
ls docs/design-context.md docs/arc/specs/design-*.md 2>/dev/null | head -5

# Test setup?
ls vitest.config.* jest.config.* playwright.config.* cypress.config.* 2>/dev/null | head -5

# Has UI? (React/Vue/Svelte components)
ls src/app/**/*.tsx app/**/*.tsx src/components/**/*.tsx components/**/*.tsx 2>/dev/null | head -3

# Has AI features?
grep -rl "from ['\"]ai['\"]" src/ app/ lib/ 2>/dev/null | head -3

# Git state
git log -1 --format=%ci 2>/dev/null
git diff --name-only HEAD~5 2>/dev/null | head -20

# CLAUDE.md or rules?
ls CLAUDE.md .claude/rules/**/*.md rules/**/*.md 2>/dev/null | head -5

# Progress journal?
head -20 docs/arc/progress.md 2>/dev/null

From these checks, build a mental model of what's present and what's missing.

---

Step 2: Output the Command Guide

Present all commands in a single output. Use the context to annotate relevance.

Format

For each command group, output:

## [Group Name]

| Command | What it does | Relevance |
|---------|-------------|-----------|
| `/arc:command` | One-line description | **Relevant** — [why] |
| `/arc:command` | One-line description | *Low relevance* — [why] |

Relevance rules:

• Relevant — the project has the prerequisites AND there's a reason to use it now
• Available — the project has the prerequisites but no urgent reason
• Low relevance — project is missing prerequisites (e.g., no UI = no /arc:design)
• Not applicable — fundamentally doesn't apply (e.g., no codebase at all)

The Catalog

Output ALL of these in order:

---

ENTRY POINTS

Command	What it does	When to use
/arc:go	Understands your codebase, asks what you want to do, routes to the right workflow	Starting a session, unsure where to begin
/arc:help	This command — shows all commands with context	When you want to see what's available

FOUNDATION

Command	What it does	When to use
/arc:vision	Define project goals, purpose, and success criteria	New projects, or when goals are unclear

DESIGN

Command	What it does	When to use
/arc:brand	Create a visual identity system — palette, typography, tone, and generated assets	New projects, rebranding, before design work
/arc:ideate	Turn an idea into a validated design through collaborative dialogue	New features that need thinking before building
/arc:design	Create distinctive, non-generic UI with aesthetic direction and wireframes	When building UI that should be memorable
/arc:naming	Generate and validate project names with domain/GitHub checks	Naming a new project or product

EXECUTE

Command	What it does	When to use
/arc:implement	Scope-aware planning and execution with TDD	Small fixes through substantial features
/arc:ai	AI SDK guidance — correct patterns, deprecated API warnings	Before implementing any AI feature

REVIEW

Command	What it does	When to use
/arc:review	Expert review with parallel specialized agents (security, design, performance, etc.)	Before merging, after implementation
/arc:audit	Mechanical verification plus comprehensive audit	Quick checks, focused audits, before shipping

TEST

Command	What it does	When to use
/arc:testing	Test strategy and execution — unit, integration, E2E with specialist agents	Creating test plans, running suites, fixing failures
SHIP

Command	What it does	When to use
/arc:letsgo	Production readiness checklist	Final gate before production deployment

CROSS-CUTTING

Command	What it does	When to use
/arc:responsive	Audit and fix mobile responsiveness with visual verification	After building desktop-first UI
/arc:seo	Deep SEO audit — meta tags, structured data, crawlability	Before launching public-facing pages
/arc:commit	Smart commit with auto-splitting across domains	When ready to commit changes
/arc:suggest	What to work on next — analyzes codebase, issues, and debt	Starting a session, unsure what to tackle
/arc:document	Capture solved problems as searchable documentation	After solving a non-obvious problem
/arc:tidy	Clean up completed plans in docs/arc/plans/	When plan files accumulate
/arc:rules	Apply Arc's coding rules to the project	Setting up a new project with Arc conventions
/arc:deps	Dependency audit — outdated packages, CVEs, batch upgrades	Periodic maintenance, before shipping
/arc:hooks	Install Claude Code hooks for auto-formatting, linting, context	Setting up development environment

UTILITY

Command	What it does	When to use
/arc:prune-agents	Kill orphaned Claude subagent processes	When agents didn't exit cleanly

---

Step 3: Contextual Recommendations

After the catalog, add a short section:

## Recommended Right Now

Based on what I found in your project:

1. **[Command]** — [specific reason based on context]
2. **[Command]** — [specific reason based on context]
3. **[Command]** — [specific reason based on context]

Pick 2-4 commands that make the most sense given:

• What's missing (no vision doc → suggest /arc:vision)
• What's stale (old plans → suggest /arc:tidy)
• What just changed (recent UI edits → suggest /arc:design or /arc:audit --harden)
• What could be improved (no tests → suggest /arc:testing)

---

Rules

• Don't invoke any other skills. This is information only.
• Don't ask questions. Output the catalog and recommendations, then stop.
• Keep context gathering under 5 seconds. Quick checks only, no deep exploration.
• Show ALL commands. Don't hide irrelevant ones — annotate them so users learn they exist.
• Be specific about relevance. "Low relevance — no UI components found" not just "Low relevance."