Water Tracker

Hydration tracking and advice from conversational mentions. All persistent data lives in ~/Clawic/data/water/ — log.md (daily entries), memory.md (learned containers, baseline, patterns), config.yaml (declared preferences); this skill reads and writes only that folder. Health context stays local: nothing is uploaded, shared, or sent anywhere, and no credentials or identifiers are stored. If you have data at an old location (~/water/ or ~/clawic/water/), move it to ~/Clawic/data/water/. First use: read setup.md; file formats in memory-template.md.

When To Use

• User mentions drinking anything ("had water with lunch", "finished my bottle") and wants intake tracked
• User asks how much water they should drink, or whether they drink enough
• User reports thirst, dark urine, headache, cramps, or fatigue and hydration is a plausible factor
• User plans or logs exercise, sauna, a flight, altitude, or hot weather — or is sick (fever, vomiting, diarrhea, hangover) — and asks what or how much to drink
• Mode: act-as tracker (log without commentary) plus advise (targets and adjustments) when asked
• Not for meal or calorie logging (route to calories) and not for diagnosing illness (see Red Flags)

Quick Reference

Depth on demand: logging.md mention→ml parsing, calibration, day boundaries · exercise.md sweat rate, endurance, hyponatremia · environment.md heat, cold, altitude, flights · illness.md fever, GI losses, hangover, kids and elderly · electrolytes.md sodium, ORS, sports drinks · conditions.md stones, pregnancy, fluid restriction, medications · habits.md pattern detection, summaries, habit design · quality.md tap, bottled, filters, travel safety · setup.md first run · memory-template.md file formats.

Core Rules

1. Daily fluid target = 30-35 ml x body weight in kg. Worked example: 70 kg → 2100-2450 ml. Weight unknown → default 1600 ml (women) / 2000 ml (men), the ~80% drinking-fluid share of EFSA adequate total-water intake (2.0 L women / 2.5 L men, food water included). Never quote 3.7 L or 2.7 L as a drinking target: those NAM figures are total water including the 20-30% that comes from food.
2. Count all non-alcoholic drinks at 100%: coffee, tea, milk, soup, sparkling water. Up to 400 mg caffeine per day produces no net fluid loss in habitual drinkers (Armstrong). Alcohol counts as 0 ml.
3. Estimate first, ask later. Use the Logging Defaults table, flag the entry as estimated, and only ask when the ambiguity changes the day total by more than 20% ("bottle" could be 500 or 1000 ml → ask; "a glass" → just log 250).
4. Judge status by urine, not by totals hit. Pale straw = on target regardless of ml logged; dark (5 or above on the 8-level Armstrong chart) at midday = deficit, drink 500 ml now; completely colorless at every void = overshooting, ease off. First morning urine is always dark and proves nothing.
5. Thirst is a lagging signal: it appears at roughly 1-2% body mass already lost, and measurable performance drop starts near 2% (ACSM). For planned long efforts, schedule intake instead of waiting for thirst. Adults over 65 have blunted thirst: schedule for them by default.
6. Cap intake rate at about 1 L per hour sustained; the gut absorbs little more, and plain water beyond sweat losses dilutes sodium. Body weight GAIN during exercise = overdrinking, stop fluids (exercise hyponatremia hit 13% of Boston Marathon finishers in the Almond study).
7. History of kidney stones changes the target: drink enough to produce at least 2.5 L urine per day (AUA), which means roughly 3 L intake. This overrides Rule 1 upward.

Logging Defaults

Default sizes when the user does not specify (store personal overrides in memory.md after one calibration; parsing rules in logging.md):

Reporting default: silent logging, summary only when asked or per the reporting variable. Never send missed-glass reminders; nagging is the main reason users abandon hydration tracking (habits.md for what works instead).

Target Adjustments

Stack on top of the Rule 1 baseline; conditions.md restrictions override everything.

Red Flags

Anything in this table suspends the protocols above: route to a clinician.

Output Gates

Before replying to any hydration-related message, verify:

• Drink mention → entry appended to log.md before anything else; reply is a one-line acknowledgment at most, no unsolicited advice?
• Any target quoted → computed from stored weight or the sex defaults (Rule 1), never a flat 2 L or "8 glasses"?
• About to suggest drinking more → today's log checked AND no Red Flags row matches?
• About to ask a question → answer not already in memory.md, and it is either the one-time container calibration or a decision the defaults cannot resolve?
• Any number given → traceable to a rule or table in this skill, not improvised?

Configuration

User-dependent variables. Defaults apply until the user states a preference; store them in ~/Clawic/data/water/config.yaml.

Preference areas — customizable dimensions; a stated preference gets recorded in config.yaml and applied:

• Beverages — drinks the user wants excluded or counted differently ("don't count soda") — modifies Rule 2 at log time
• Containers — personal container sizes; learned once via calibration, stored in memory.md, never re-asked
• Schedule — fasting windows, pre-bed cutoff, front-loading — shapes when suggestions land (habits.md)
• Activity — training days and sport — pre-applies the exercise adjustment on those days (exercise.md)
• Medical context — declared conditions or clinician-set numbers — override Rule 1 per conditions.md; never inferred from symptoms
• Tone — summary verbosity, encouragement vs bare numbers — affects habits.md summaries only

Traps

Where Experts Disagree

• Drink to thirst vs programmed drinking. The exercise-hyponatremia consensus (Hew-Butler) pushes drink-to-thirst for endurance events because overdrinking kills and underdrinking rarely does; ACSM programs intake for hard efforts in heat where thirst lags real losses. Boundary: fast, hot, or over-4-h efforts and adults over 65 → schedule it; recreational sessions → thirst plus the sweat-rate test (exercise.md).
• Water as a weight-loss or skin tool. Pre-meal water (about 500 ml) showed a modest weight-loss effect in one dieting RCT (Dennis); skin-appearance claims lack evidence. Offer the pre-meal trick when asked; never promise skin or metabolism effects.
• Is plain water the best rehydrator? For everyday life, yes. After heavy one-shot losses, drinks with sodium and nutrients are retained measurably longer (Maughan's beverage hydration index — details in electrolytes.md); the boundary is losses, not preference.

Related Skills

More Clawic skills, get them at https://clawic.com/skills/water (install if the user confirms):

• fasting — eating-window rules and electrolyte protocol for fasts over 24 h
• calories — meal and drink logging when the user tracks food, not just fluids
• running — race-day fueling and pacing where fluid plans meet training plans
• gym — workout logging that feeds the exercise adjustment automatically

Feedback

• If useful, star it: https://clawic.com/skills/water
• Latest version: https://clawic.com/skills/water

Part of Clawic, the verified skill library. Get this skill: https://clawic.com/skills/water.