name: brainstorm description: Scan codebase, propose improvements AND features autonomously aliases: ["what next", "whatnext", "what-next"] allowed-tools: Bash, Read, Grep, Glob, Task, TaskCreate, TaskUpdate, TaskList, Write, Edit model: opus user-invocable: true

Brainstorm

Philosophy: User doesn't know what to focus on. YOU scan, analyze, propose, and create stories - without asking.

Usage

Phase 1: Quality Scan (Parallel)

Launch 4 scans simultaneously using Task tool with run_in_background: true:

Task({ subagent_type: "Explore", model: "haiku", run_in_background: true,
  prompt: "Find TODOs/FIXMEs in [PROJECT_PATH]. Report: count, file:line, content." })

Task({ subagent_type: "Explore", model: "haiku", run_in_background: true,
  prompt: "Find console.log statements in [PROJECT_PATH] (skip test files). Report: count, files." })

Task({ subagent_type: "Explore", model: "haiku", run_in_background: true,
  prompt: "Find hardcoded colors (text-white, bg-black, #hex, rgb) in [PROJECT_PATH]. Report: count, files." })

Task({ subagent_type: "Explore", model: "haiku", run_in_background: true,
  prompt: "Find large files (>300 lines) and 'any' type usage in [PROJECT_PATH]. Report: file, lines, issues." })

Phase 2: Feature Ideation (Autonomous)

After scans complete, read project context:

• CLAUDE.md - goals, roadmap, known issues
• README.md - what the app does
• package.json - name, description, dependencies

Then analyze and propose 3-8 features:

• Missing features - what similar apps have that this doesn't
• UX improvements - based on component structure found
• Integration opportunities - based on installed packages
• Performance wins - based on patterns observed

Be specific: "Add Cmd+K search modal" not "Improve UX"

Phase 3: Present Everything

Brainstorm Complete
═══════════════════
Scanned 247 files in 45 seconds.

Quality Issues
┌──────────────────┬───────┬──────────────────┐
│ Category         │ Count │ Status           │
├──────────────────┼───────┼──────────────────┤
│ TODOs/FIXMEs     │ 0     │ ✅ Clean         │
│ console.log      │ 12    │ ⚠️ In 4 files    │
│ Hardcoded colors │ 6     │ ⚠️ In shadcn/ui  │
│ Large files      │ 3     │ ⚠️ >500 lines    │
└──────────────────┴───────┴──────────────────┘

Feature Ideas
┌───┬─────────────────────────────────┬────────┐
│ # │ Idea                            │ Effort │
├───┼─────────────────────────────────┼────────┤
│ 1 │ Add keyboard shortcuts (Cmd+K) │ Medium │
│ 2 │ Offline mode (PWA ready)        │ High   │
│ 3 │ Export to PDF                   │ Low    │
└───┴─────────────────────────────────┴────────┘

Create stories?
- "quality" → cleanup tasks only
- "features" → feature tasks only
- "all" → everything

Targeted Mode

When user says brainstorm X:

• Skip quality scan entirely
• Read files related to X topic
• Propose 3-5 specific ideas for X
• Immediately create stories

Rules

• Never ask "what do you want?" - analyze and propose
• Don't over-generate - 3-8 feature ideas max
• Be specific - concrete features, not vague improvements
• Note effort - Low/Medium/High for each
• Skip shadcn/ui colors - note them but don't prioritize (library defaults)
• Auto-create for top recommendation - then offer more

Token Cost

• 4 parallel Haiku scans: ~20K tokens
• Context reads: ~5K tokens
• Time: 30-60 seconds
• Much cheaper than reading entire codebase