Manage Concepts Database

MANDATORY for ALL concept operations: add, update, verify, merge duplicates, sync.

Core Rules

1. ALWAYS verify before adding - run verification BEFORE creating JSON
2. ALWAYS fix-concepts after changes - run bun run fix-concepts AFTER adding/updating concepts
3. ALWAYS sync after changes - run sync AFTER modifying concepts
4. ≥90% confidence = STOP - review suggested duplicates
5. 70-89% confidence = REVIEW - manually compare matches
6. <70% confidence = PROCEED with caution

Scripts Reference

Quick Workflow: Add Single Concept

Step 1: Quick Check (5s)

grep -ri "CONCEPT_NAME" src/data/concepts/ 2>/dev/null | head -5

If matches → stop, verify manually.

Step 2: Database Verify (10s)

bun scripts/verify-concept.ts --name "Concept Name" --summary "Brief summary"

Step 3: Find Related Notes

find $OBSIDIAN_VAULT_LOCATION/30\ Areas -type f -name "*KEYWORD*.md" 2>/dev/null | grep -v ".smart-env" | head -5

Convert path to URL: spaces→+, remove .md, prefix https://notes.dsebastien.net/

Step 4: Find Related Articles

for f in $(grep -ril "KEYWORD" "$OBSIDIAN_VAULT_LOCATION/30 Areas/33 Permanent notes/33.04 Creations/Articles/"*.md 2>/dev/null | grep -v "(Draft)"); do
  grep -m1 "^title:" "$f" | sed 's/title: //'
  grep -m1 "^url:" "$f" | sed 's/url: //'
done

Step 5: Create Concept JSON

{
    "id": "concept-id",
    "name": "Concept Name",
    "summary": "One-sentence summary",
    "explanation": "Detailed explanation.",
    "tags": ["tag1", "tag2"],
    "category": "Category",
    "icon": "FaLightbulb",
    "featured": false,
    "aliases": [],
    "relatedConcepts": [],
    "relatedNotes": [],
    "references": [{"title": "Name - Wikipedia", "url": "https://en.wikipedia.org/wiki/...", "type": "website"}],
    "articles": [],
    "books": [],
    "tutorials": [],
    "datePublished": "YYYY-MM-DD",
    "dateModified": "YYYY-MM-DD"
}

Step 6: Fix & Sync

bun run fix-concepts
bun scripts/sync-concepts-db.ts

Bulk Concept Addition (Parallel)

For 10+ concepts, use sub-agents:

Task tool (subagent_type="general-purpose"):
"Create concept JSON for [NAME]:
1. Read source note at [PATH] if available
2. Generate explanation if source thin
3. Search related notes in $OBSIDIAN_VAULT_LOCATION/30 Areas/ (32.02 or 33.02 Content)
4. Search related articles in .../33.04 Creations/Articles/ (skip Draft)
5. Create JSON at src/data/concepts/[id].json
6. Add Wikipedia reference if available"

After all complete:

bun run fix-concepts
bun scripts/sync-concepts-db.ts
bun run build 2>&1 | tail -5

Merge Duplicates

# Review both
cat src/data/concepts/{source-id}.json
cat src/data/concepts/{target-id}.json

# Merge (combines tags, aliases, references; deletes source)
bun scripts/merge-duplicates.ts --source {source-id} --target {target-id} --strategy merge-fields
bun run fix-concepts
bun scripts/sync-concepts-db.ts

URL Construction: Related Notes

File: $OBSIDIAN_VAULT_LOCATION/30 Areas/33 Permanent notes/33.02 Content/Note Name.md
URL:  https://notes.dsebastien.net/30+Areas/33+Permanent+notes/33.02+Content/Note+Name

Rules: spaces→+, remove .md, path starts 30+Areas/...

Category Assignment

Tag Rules

• Plural: strategies, businesses, decisions
• Exceptions (singular): gerunds (brainstorming), uncountable (knowledge), fields (psychology)
• Hyphenated: well-being, systems-thinking

Check existing:

grep -h '"tags"' src/data/concepts/*.json | tr ',' '\n' | tr -d '[]"' | sed 's/^[[:space:]]*//' | sort -u

Cross-References

# Find by tag
grep -l '"TAGNAME"' src/data/concepts/*.json | xargs -n1 basename | sed 's/.json$//'

# Find by category
grep -l '"category": "CATEGORY"' src/data/concepts/*.json | xargs -n1 basename | sed 's/.json$//'

# Find by keyword
grep -l 'KEYWORD' src/data/concepts/*.json | xargs -n1 basename | sed 's/.json$//'

Add cross-references in both directions via relatedConcepts array.

List Existing Concepts

# Count
ls src/data/concepts/*.json | wc -l

# IDs
ls src/data/concepts/*.json | xargs -n1 basename | sed 's/.json$//' | sort

# Names
grep -h '"name"' src/data/concepts/*.json | sed 's/.*"name": "//;s/".*//' | sort

# Tags
grep -h '"tags"' src/data/concepts/*.json | tr ',' '\n' | tr -d '[]"' | sed 's/^[[:space:]]*//' | sort -u

# Categories
grep -h '"category"' src/data/concepts/*.json | sed 's/.*"category": "//;s/".*//' | sort -u

Generate Concept ID

echo "Parkinson's Law" | tr '[:upper:]' '[:lower:]' | sed "s/'//g" | sed 's/ /-/g' | sed 's/[^a-z0-9-]//g'
# → parkinsons-law

Database Maintenance

# Stats
sqlite3 concepts.db "SELECT (SELECT COUNT(*) FROM concepts) as concepts, (SELECT COUNT(*) FROM concept_aliases) as aliases, (SELECT COUNT(*) FROM concept_tags) as tags;"

# Rebuild
rm concepts.db && bun scripts/init-concepts-db.ts

Handling Light Source Content

When source note is thin (1-2 sentences):

• Generate comprehensive explanation using your knowledge
• Cover: what, how, why, when to apply
• Add relevant references (Wikipedia, books, papers)
• Still link source note in relatedNotes

Skip concept only if:

• Just a quote (no actionable concept)
• Confirmed duplicate
• Person/book name without conceptual content
• Article reference, not a concept