Second Brain Debugging

Overview

This skill adapts systematic debugging for the Second Brain stack:

• Nuxt 4 with Vue 3 Composition API
• @nuxt/content v3 with SQLite and minimark format
• @nuxt/ui v4 components
• D3.js for knowledge graph visualization

Core principle: ALWAYS trace data flow before attempting fixes.

Stack-Specific Gotchas

Before debugging, internalize these common pitfalls:

The Four Phases

Phase 1: Identify the Layer

Data flows through these layers:

Content File (Markdown)
    ↓ Parsed by Nuxt Content
Minimark AST (body.value array)
    ↓ Queried via queryCollection
API Endpoint (server/api/*.ts)
    ↓ Returned as JSON
Vue Composable (useBacklinks, useMentions)
    ↓ Rendered in component
User Interface

First question: Which layer is broken?

1. Content layer - Check frontmatter, markdown syntax
2. AST layer - Check minimark structure: [tag, props, ...children]
3. Query layer - Check .select('body'), path matching
4. API layer - Check response structure, error handling
5. Client layer - Check composable logic, reactivity

Quick diagnostic:

# Check if content is queryable
pnpm nuxi dev
# Visit http://localhost:3000/api/graph
# Empty edges? → Layer 2-3 issue
# Empty nodes? → Layer 1 issue
# Correct data but UI wrong? → Layer 4-5 issue

Phase 2: Trace Data Flow

For content/link issues:

1. Check raw content file:

   • Frontmatter has title and type?
   • Wiki-links use correct format [[slug]]?
   • File in /content/ directory?

2. Check minimark extraction:

   // In server/api/graph.get.ts, temporarily add:
   console.log('Body structure:', JSON.stringify(item.body, null, 2))
   // Verify: { type: 'minimark', value: [[...arrays...]] }
   // NOT: { type: 'minimark', value: [{...objects...}] }
   

3. Check query selection:

   // Must explicitly select body:
   queryCollection(event, 'content')
     .select('path', 'stem', 'title', 'type', 'body') // ← body required!
     .all()
   

4. Check API response:

   curl http://localhost:3000/api/graph | jq '.edges | length'
   curl http://localhost:3000/api/backlinks | jq 'keys | length'
   

For graph visualization issues:

1. Check nodes exist before filtering
2. Check edge source/target types (string vs object after D3)
3. Check filter state in URL query params

Phase 3: Hypothesis Testing

Form ONE hypothesis:

• "The body isn't being selected because..."
• "Links aren't extracted because minimark format is..."
• "The cache is stale because..."

Test minimally:

# Run relevant tests first
pnpm test:unit -- minimark  # For link extraction
pnpm test:run -- graph      # For API issues

Add targeted logging:

// In extractLinksFromBody (server/utils/minimark.ts)
console.log('Body type:', body?.type)
console.log('Body value is array?', Array.isArray(body?.value))
console.log('First node:', JSON.stringify(body?.value?.[0]))

Phase 4: Fix and Verify

1. Create failing test first:

   // tests/unit/utils/minimark.test.ts
   it('extracts links from nested minimark', () => {
     const body = {
       type: 'minimark',
       value: [['a', { href: '/target' }, 'Link text']]
     }
     expect(extractLinksFromBody(body)).toContain('target')
   })
   

2. Implement fix in ONE place

3. Verify with full test suite:

   pnpm test:run
   pnpm lint:fix && pnpm typecheck
   

4. If fix doesn't work after 3 attempts:

   • Question the architecture
   • Is the data model fundamentally sound?
   • Should slug normalization be centralized?

Quick Debugging Commands

# Run all tests
pnpm test:run

# Run specific test file
pnpm test:unit -- minimark
pnpm test:run -- graph.nuxt

# Type check
pnpm typecheck

# Start dev server for manual testing
pnpm dev

# Check API responses
curl -s http://localhost:3000/api/graph | jq '.nodes | length'
curl -s http://localhost:3000/api/graph | jq '.edges | length'
curl -s http://localhost:3000/api/backlinks | jq 'keys'
curl -s "http://localhost:3000/api/mentions?slug=test&title=Test" | jq

Key Files by Subsystem

Content Pipeline

• /content/*.md - Raw content files
• /content.config.ts - Collection schema (title, type required)
• /modules/wikilinks.ts - [[link]] to anchor transformation

Link Extraction

• /server/utils/minimark.ts - Extract links from AST
• Body format: { type: 'minimark', value: [...arrays...] }
• Link format: ['a', { href: '/slug' }, 'text']

API Endpoints

• /server/api/graph.get.ts - Nodes and edges for D3
• /server/api/backlinks.get.ts - Reverse link index
• /server/api/mentions.get.ts - Unlinked mentions search

Client Composables

• /app/composables/useBacklinks.ts - Fetch backlinks
• /app/composables/useMentions.ts - Fetch mentions
• /app/composables/useGraphFilters.ts - Graph filter state

Pages

• /app/pages/[...slug].vue - Content page
• /app/pages/graph.vue - Knowledge graph

Common Fixes

"Graph shows nodes but no edges"

// Fix: Add .select('body') to query
queryCollection(event, 'content')
  .select('path', 'stem', 'title', 'type', 'tags', 'body') // ← ADD body
  .all()

"Links not being extracted"

// Check minimark format - must be ARRAY based
// Wrong: { tag: 'a', props: { href: '/' }, children: [] }
// Right: ['a', { href: '/' }, 'text']

// Fix extractLinksFromMinimark to handle arrays:
function extractLinksFromMinimark(node: unknown): string[] {
  if (!Array.isArray(node)) return []
  const [tag, props, ...children] = node
  // ...
}

"Stale backlinks after content edit"

// Force refresh with key
const { data: backlinks, refresh } = await useAsyncData(
  `backlinks-${Date.now()}`, // or add content hash
  () => $fetch('/api/backlinks')
)

"Content page 404"

// Check slug normalization
// Route uses: /atomic-habits (with slash)
// queryCollection expects path: '/atomic-habits' (with slash)
// Backlinks use slug: 'atomic-habits' (no slash)

// Ensure consistent normalization:
const slug = route.path.startsWith('/') ? route.path : `/${route.path}`

Red Flags - STOP

If you catch yourself:

• Guessing which layer the bug is in
• Adding fixes to multiple files at once
• Skipping the test before fixing
• Not checking minimark format
• Assuming body is included by default

STOP. Return to Phase 1.