TagTime CLI (tt)

External memory for AI Agents. Save decisions, search context, build knowledge — from terminal.

Current version: 0.1.5 · GitHub · 中文

Why tt?

AI is stateless. Your projects aren't.

Every time you start a new AI conversation, context is lost:

• "We decided to use Zod for validation because..." — forgotten
• "The API error format should be..." — explain again
• "Don't use default exports in this project" — AI doesn't know

tt gives AI persistent memory.

# Save a project decision
tt save "Use Zod for validation - better TS inference than Yup" --title "Validation Choice"

# Later, AI (or you) can retrieve it
tt search "validation"
tt get @1

Works with local folders, Obsidian/Logseq vaults, or TagTime cloud. Local-first, works offline.

---

Quick Start

# 1. Add your notes folder (any folder works)
tt vault add notes ~/Documents/notes
tt mode local

# 2. Save knowledge
tt save "Always use pnpm in this project, not npm" --title "Package Manager"

# 3. Retrieve context
tt search "package manager"
tt get @1

# Cloud mode (sync across devices)
tt login
tt save "Your note"

How AI Agents Use tt

# AI saves a decision during coding
tt save "Chose Prisma over Drizzle - team familiar with Prisma" --tags "decisions"

# AI retrieves context before starting a task
tt search "database orm"

# AI captures learnings
git diff | tt save --title "Auth refactoring - extracted useToken hook"

Install skill for your AI assistant:

tt skill install              # Claude Code (default)
tt skill install -t cursor    # Cursor
tt skill install -t copilot   # GitHub Copilot
tt skill install -t all       # All platforms

---

Core Commands

Save — Capture knowledge

tt save "content"                      # Quick save
tt save "content" --title "Title"      # With title
tt save "content" --tags "tag1,tag2"   # With tags
tt save ./file.md                      # Save file
echo "content" | tt save               # Pipe input

# Capture command output
git log --oneline -10 | tt save --title "Recent commits"
git diff | tt save --title "Today's changes"

Search — Find context

tt search "keyword"                    # Full-text search (fuzzy)
tt search "validation library"         # Semantic-ish search
tt search "auth" --folder "decisions"  # Filter by folder
tt search "api" --tag "conventions"    # Filter by tag
tt search "keyword" --json             # JSON output (for AI)

Get — Retrieve content

tt get @1                   # Get first result from last search
tt get @2                   # Get second result
tt get "path/to/note.md"    # Get by path
tt get @1 --raw             # Content only, no formatting

---

Local Mode

Work with any local folder — plain markdown, Obsidian, or Logseq.

Vault Management

# Add vault
tt vault add <name> <path>
tt vault add notes ~/Documents/notes          # Plain markdown
tt vault add obsidian ~/Obsidian/Vault        # Auto-detects Obsidian
tt vault add logseq ~/Logseq/Graph            # Auto-detects Logseq

# Manage
tt vault list
tt vault use <name>
tt vault info
tt vault remove <name>

# Switch modes
tt mode              # Show current
tt mode local        # Local mode
tt mode cloud        # Cloud mode

---

Cloud Mode

Sync across devices with TagTime cloud.

# Auth
tt login                  # Interactive login
tt login --token KEY      # API key login
tt whoami                 # Check status

# Save/Search/Get — same as local mode
tt save "content" --title "Title"
tt search "keyword"
tt get @1

# Recent
tt recent                 # List recent
tt recent -n 5

# Batch operations
tt import ./notes/*.md              # Import files
tt import ./notes/ -r               # Recursive
tt export -o ./backup/              # Export all
tt export -o ./backup/ --tag "work" # Export filtered

---

Pipe Input

Perfect for capturing command output:

git log --oneline -10 | tt save --title "Recent commits"
git diff | tt save --title "Changes"
docker logs app | tt save --tags "debug"
curl https://api.example.com | tt save -t "API Response"
pbpaste | tt save --title "From clipboard"  # macOS

---

AI Agent Integration

Let AI assistants use tt as their external memory:

tt skill install              # Claude Code (default)
tt skill install -t cursor    # Cursor
tt skill install -t codex     # OpenAI Codex CLI / OpenCode
tt skill install -t copilot   # GitHub Copilot
tt skill install -t all       # All platforms

tt skill status               # Check installation
tt skill uninstall -t all     # Remove

Platform	File Location
Claude Code	~/.claude/skills/tagtime-cli/SKILL.md
Codex / OpenCode	./AGENTS.md
Cursor	./.cursor/rules/tagtime.mdc
GitHub Copilot	./.github/copilot-instructions.md

---

Config

Config: ~/.config/tagtime/config.json

---

Development

pnpm install        # Install
pnpm dev            # Dev mode
pnpm build          # Build
pnpm test           # Test

---

Philosophy

Traditional note tools store information for humans to read later.

tt stores context for AI to use now.

In the age of vibe coding, your knowledge base isn't just for you — it's working memory for AI agents that help you build software. Every decision saved is context that doesn't need to be re-explained.

Human saves decision → tt stores it → AI retrieves it → Better code