Toggl Track (REJECTED)

REJECTED — manual tracking violates auto-only rule, kept for audit

Toggl Track (REJECTED)

Why rejected (2026-06-20): Manual time tracking violates the family-wide auto-only-tracking rule locked the same day this service was originally adopted. Wakatime auto-tracks coding time; non-coding time is intentionally not tracked at all rather than manually tracked. Manual timers decay — humans forget to start, fudge stops, skip days — and the resulting data is dishonest, which makes it worse than no data. File kept for audit trail per the family pattern of “flip status, never delete”.

Role (was)

The family’s manual time tracker for non-coding work — meetings, calls, planning, design reviews, deep-work blocks, learning sessions, admin. Adopted 2026-06-20 (Batch 19) as the manual half of a Toggl + Wakatime split; walked back the same day later in 2026-06-20 when the auto-only-tracking rule was locked. See decisions/architecture/time-tracking-wakatime-only for the walk-back narrative.

What replaces it

Nothing. Non-coding time is not a system metric in the family. The auto-only-tracking rule explicitly distinguishes metrics (must auto) from content (manual writing is fine — that’s what the journal is). Time spent on meetings / planning / calls is not a metric the family captures.

Wakatime remains the sole time-tracking pick — it captures coding time automatically via IDE plugin heartbeat, no manual start / stop required.

Free tier (still valid, kept for record)

• Up to 5 users (would have used 1)
• Unlimited time entries
• Unlimited projects + clients + tags
• Basic reports + CSV export
• REST API access (api.track.toggl.com/api/v9)
• Cross-platform clients + Pomodoro timer + idle detection

Card / subscription required?

NO. Free tier sign-up was email or Google sign-in. No payment method requested. Free tier is permanent, not a trial. (Card-on-file posture was fine — the rejection is on the manual-tracking dimension, not the billing dimension.)

Why this was originally picked (then walked back)

When Toggl was adopted earlier in 2026-06-20, the reasoning was:

• Manual timers capture intent + context (what the user meant to spend time on); auto-tracking captures actuals.
• Free unlimited entries with API access is rare among manual trackers.
• 5-user free tier with no card.

The walk-back logic: manual capture decays. The honest read of “what was I doing during this 2-hour gap in Wakatime data?” is “untracked”, not “I should manually fill it in later.” Untracked time is a feature, not a bug — it tells the user they were away from the keyboard, which is information.

What this means concretely going forward

• No Toggl Track account is created. If one was created during the brief Batch 19 adoption window earlier today, it stays unused and will be deleted at next account-cleanup pass.
• No TOGGL_API_TOKEN in Doppler or any runtime secret store.
• No lifestream ingest of Toggl data — the future lifestream pulls from auto sources only (Wakatime API, GitHub commits, npm publishes, CI events, CF Web Analytics — see auto-tracking-everywhere).
• No manual-tracking surface anywhere in the family (no Clockify, no Harvest, no spreadsheet, no app).

Cross-refs

• Auto-only-tracking rule (2026-06-20) — the rule that rejects this service
• Auto-tracking-everywhere decision — the family-wide principle
• Time-tracking — Wakatime only (decision, walk-back) — the rename of the original Batch 19 decision after the walk-back
• Wakatime — sole time-tracking pick
• Productivity services index
• Lifestream JSONL canonical
• No card-on-file rule — separate concern; Toggl was fine on this axis

Edit on GitHub · Back to index