lifetract

Life + Traction — A CLI that lets AI agents query your health and time-tracking data.

Go · modernc.org/sqlite · Single static binary · JSON output

AI Agent Skill: pi-skills/lifetract

What This Does

Unifies Samsung Health CSV exports (8 types) and aTimeLogger SQLite (18 categories) into a single SQLite database (lifetract.db). AI agents query health and time data via JSON — sleep, heart rate, steps, stress, exercise, weight, HRV, and how you spend your hours.

Every record carries a Denote ID (YYYYMMDDTHHMMSS), enabling cross-referencing with denotecli on the same time axis.

Samsung Health and aTimeLogger are widely-used apps with standard export formats. lifetract is a general-purpose tool — anyone using these apps can use it. Data itself is private; public dataset is planned separately.

./run.sh update                     # Place new data + rebuild DB (one step)
lifetract today                     # Today's unified summary
lifetract read 2025-10-04           # Full day: health + time tracking
lifetract timeline --days 30        # 30-day cross-sectional view
lifetract time --days 30            # Time category analysis

Install

git clone https://github.com/junghan0611/lifetract.git
cd lifetract
./run.sh build    # → ~/.local/bin/lifetract

Requires Go 1.21+. Nix users: nix build (CGO_ENABLED=0 static binary).

Updating Data

Export Samsung Health + aTimeLogger backup from your phone into a date folder, then one command:

# Put exports in self-tracking-data/YYYYMMDD/, then:
./run.sh update

Automatically: moves Samsung Health folder → replaces aTimeLogger DB → rebuilds lifetract.db.

Architecture

./run.sh update (or lifetract import --exec)
  Samsung Health CSVs ────────────┐
  aTimeLogger SQLite ─────────────┼→ lifetract.db → JSON API
                                  │
lifetract <command>               │
  DB exists? → SQLite query (~90ms)
  No DB?     → CSV fallback (~300ms)

DB Tables

| Table | Source | Rows | |-------|--------|------| | sleep | Samsung Health | 4,489 | | sleep_stage | Samsung Health | 78,591 | | heart_rate | Samsung Health | 62,036 | | steps_daily | Samsung Health | 9,692 | | stress | Samsung Health | 25,768 | | exercise | Samsung Health | 2,195 | | weight | Samsung Health | 283 | | hrv | Samsung Health | 1,058 | | atl_category | aTimeLogger | 18 | | atl_interval | aTimeLogger | 13,918 |

Total: 198,030 rows, 36MB (as of 2026-03-10)

Commands

| Command | Description | |---------|-------------| | status | Data source availability and stats | | import [--exec] | Build DB (dry-run / execute) | | today | Today's unified health + time summary | | read <id> | Query by Denote ID (day or event level) | | timeline [--days N] | Date-indexed cross-sectional view | | sleep [--days N] [--summary] | Sleep session analysis | | steps [--days N] | Daily step counts | | heart [--days N] | Heart rate trends | | stress [--days N] | Stress levels | | exercise [--days N] | Exercise sessions | | time [--days N] [--category X] | Time category analysis (aTimeLogger) | | export | Public-safe export plan |

Flags

| Flag | Default | Description | |------|---------|-------------| | --days N | 7 | Query period | | --data-dir DIR | ~/repos/gh/self-tracking-data | Data root | | --shealth-dir DIR | Auto-detect latest | Samsung Health directory | | --summary | false | Summary mode | | --category CAT | all | Time category filter | | --exec | false | Execute mode (for import) |

Denote ID System

Day level:   20251004T000000  ← same key as denotecli journals
Event level: 20251004T233000  ← individual sleep/exercise events

Cross-reference:

lifetract read 2025-10-04       # Body state + time usage that day
denotecli search "20251004"     # Notes and journal entries that day

Data Sources

| Source | Format | Period | Location | |--------|--------|--------|----------| | Samsung Health | CSV export | 2017-03 – present | self-tracking-data/samsunghealth_*/ | | aTimeLogger | SQLite DB | 2021-10 – present | self-tracking-data/atimelogger/database.db3 |

aTimeLogger Categories (Indistractable Framework)

| Class | Categories | |-------|------------| | Traction | Deep work, Reading, Practice, Exercise, Walking, Self-talk | | Maintenance | Sleep, Nap, Meals, Prep, Chores, Commute, Shopping | | Distraction | Distraction, YouTube, Short break, Leisure | | Family | Family |

Project Structure

lifetract/
├── run.sh              # Build, test, update
├── flake.nix           # Nix packaging (CGO_ENABLED=0)
├── docs/plan.md        # Roadmap
└── lifetract/          # Go source
    ├── main.go         # CLI routing
    ├── config.go       # Configuration + auto-detect
    ├── helpers.go      # Shared utils (Denote ID, time parsers)
    ├── csv.go          # Samsung Health CSV parser + record types
    ├── db.go           # SQLite schema + connection
    ├── db_query.go     # DB query functions
    ├── import.go       # Import manifest
    ├── import_exec.go  # CSV→DB conversion
    ├── export.go       # Public-safe export + category policy
    ├── query.go        # Query commands (DB↔CSV routing)
    ├── timeline.go     # Timeline builder
    ├── read.go         # Denote ID lookup
    └── *_test.go       # 47 tests, 68% coverage

Related Projects

| Project | Role | |---------|------| | denotecli | Qualitative data (3,000+ notes/journals) — same Denote ID axis | | self-tracking-data | Raw data repository |

Data Update Log

| Date | Total Rows | DB Size | Samsung Health | aTimeLogger | Notes | |------|------------|---------|----------------|-------------|-------| | 2025-10-06 | 183,635 | 33MB | ~2025-10 (77 CSVs) | ~2025-10 (13,102 intervals) | Initial build | | 2026-03-10 | 198,030 | 36MB | ~2026-03 (78 CSVs) | ~2026-03 (13,918 intervals) | +14,395 rows, 5 months added |

Author: @junghanacs · Apache 2.0

Lifetract

Install / Use

README