TL;DR Star any moment in Claude Code. AI enriches it. Find it months later. One click back to source.
The Problem
You had a conversation with Claude three weeks ago. It solved your exact problem. The explanation was perfect.
Now you need it again. Where is it?
KNOWLEDGE DECAY KNOWLEDGE DECAY
Day 1 Insight discovered
│
▼
Day 45 "I solved this before..."
│
▼
Day 46 Ask Claude again ──────▶ LOST context
You remember the solution existed. You don’t remember where.
281 sessions. 46 projects. No way to mark what mattered.
The Solution
Star it when you see it.
Claude explains the domain model...
Should Order contain line items or reference them?
Order should own LineItems. They share a lifecycle—
when an order is deleted, its items go too. This is an Aggregate Root pattern.
💡 This clarifies the whole design
★ Click star
Save to Library
Title Order aggregate boundary
Why saving? Key decision on aggregate roots_
★ Save
✨ AI Enriching...
Summary Order as aggregate root owning LineItems
Tags DDD · aggregates · domain-model
Insight Shared lifecycle = same aggregate boundary
✓ Enrichment complete
⏳ 4 months later...
🔍 "aggregate boundary"
★ Order aggregate boundary Jan 31 · DDD · aggregates
Found in 0.02s
Click → Navigate to source
Should Order contain line items or reference them?
Order should own LineItems. They share a lifecycle...
Full session context restored
Design decision preserved ✓
See Star Enrich Find Navigate
That’s the whole workflow:
- See something valuable
- Star it with one click
- Note why it matters (optional but helpful)
- AI enriches automatically
- Find it months later
- Navigate back to full context
What You Can Star
Four levels of granularity:
BOOKMARK TYPES BOOKMARK TYPES
SESSION ────────── Entire session
│ "The whole auth refactor"
│
└─▶ CONVERSATION ── One Q&A exchange
│ "That useEffect explanation"
│
└─▶ ARTIFACT ── File, command, output
│ "The bash one-liner"
│
└─▶ MESSAGE ── Single response
"This specific insight"
Star the whole session. Or just one perfect response. Whatever captures the value.
AI Enrichment
After you star something, Tacit enriches it:
ENRICHMENT ENRICHMENT
Bookmark ────▶ Claude Haiku ────▶ Enriched
│
┌─────────────────┘
│
├── Title
├── Summary
├── Tags
└── Intent/Outcome
You starred it as “Order aggregate question.”
Four months later, you search “aggregate boundary.”
Found.
The Library
All your bookmarks in one place:
LIBRARY VIEW LIBRARY VIEW
📚 Library
──────────────────────────────────
🔍 Search... [All] [Sessions]...
──────────────────────────────────
📌 PINNED
★ Auth System Refactor · 2d ago
──────────────────────────────────
▼ TODAY
★ useEffect cleanup ···· 3h ago
★ CI parallel tests ···· 5h ago
──────────────────────────────────
▶ YESTERDAY
▶ THIS WEEK
▶ EARLIER
Filter by type. Search across everything. Pin what matters most.
Click any bookmark to jump to the source. Full context restored.
Staleness Detection
Content changes. The Library tracks it.
STALENESS STALENESS
Bookmark ────▶ hash: abc123
│
time passes
│
Source ──────▶ hash: xyz789 ──▶ ⚠️ Content changed
If the original session was modified, you’ll see a warning. Your reference might be outdated.
The Workflow
When to Star Star as you work, not after. When Claude explains something perfectly, that’s the moment. The why it matters is fresh. Capture it then.
The habit:
- Something clicks → Star it
- Add a note about why
- Keep working
- Find it when you need it
Every starred insight is future-you saying thanks.
Coming Next
Library is foundational. Building on it:
- Collections — Group related bookmarks
- Export — Share curated knowledge
- Patterns — Find themes across bookmarks
- Suggestions — AI recommending what to save
Library is available in Tacit. Start at tacit.sh