@murphai/murph 1.3.0 → 1.3.2

package/node_modules/@murphai/assistant-engine/skills/nutrition-strategy/references/named-diets/dash.md

@@ -0,0 +1,183 @@
+# DASH
+
+Last evidence review: **2026-08-09**
+
+This progressive-disclosure reference supplements the parent `nutrition-strategy` contract. Apply the parent first. This file adds DASH-specific evidence and execution details; it does not own diagnosis, blood-pressure targets, medication changes, or clinician-prescribed sodium, potassium, or fluid limits.
+
+## What DASH is
+
+DASH means Dietary Approaches to Stop Hypertension. It is a flexible, blood-pressure-oriented eating pattern, not a branded food list, cleanse, weight-loss program, or rigid menu.
+
+The common pattern emphasizes:
+
+- vegetables and fruit;
+- whole grains and other minimally processed starches;
+- beans, lentils, nuts, and seeds;
+- fat-free or low-fat dairy in the classic version;
+- fish, poultry, eggs, tofu, or lean unprocessed meat as desired; and
+- unsaturated oils and other satisfying fats in practical amounts.
+
+It generally reduces sodium, sugar-sweetened drinks and frequent sweets, fatty or processed meats, and foods high in saturated fat. These are pattern-level priorities, not purity rules or absolute bans.
+
+Meaningful variants include:
+
+- **Standard-sodium DASH:** official materials commonly use a ceiling near 2,300 mg sodium per day.
+- **Lower-sodium DASH:** commonly uses about 1,500 mg per day and lowered blood pressure further in short controlled feeding research, but is more restrictive and not an automatic target for every user.
+- **Lactose-free or lactose-reduced DASH:** keeps the classic dairy structure with tolerated products.
+- **Dairy-free DASH-style:** replaces dairy with nutrient-aware plant foods. It can preserve the broad pattern, but it is not the exact version tested in the landmark trials.
+- **DASH-like higher-protein or higher-unsaturated-fat variants:** short feeding studies found that replacing some carbohydrate with protein or unsaturated fat could improve selected blood-pressure and lipid outcomes. This does not require a high-protein or low-carbohydrate diet.
+
+Clinician-set sodium, potassium, or fluid instructions override these variants. Do not convert a DASH reference number into a personalized prescription from chat.
+
+## What the evidence supports
+
+- **Blood pressure versus usual or control diets:** DASH has unusually strong named-diet evidence for this specific surrogate outcome. The landmark trials were randomized controlled feeding studies, and a 2020 meta-analysis of 30 randomized trials found average between-group reductions of about 3.2 mm Hg systolic and 2.5 mm Hg diastolic, with moderate certainty. Effects varied across populations and implementations.
+- **DASH plus sodium reduction:** sodium reduction lowered blood pressure independently, and the combination produced the largest reductions in the short DASH-Sodium feeding trial. The standard and lower-sodium variants therefore have a real evidence basis rather than being marketing categories.
+- **Guideline support:** the 2025 United States blood-pressure guideline strongly supports a heart-healthy pattern such as DASH, lower sodium, and higher dietary potassium for adults, while placing medication and target decisions in clinical care.
+- **Plausible mechanisms:** lower sodium and saturated fat, more potassium-, calcium-, magnesium-, and fiber-rich foods, and displacement of highly processed foods are plausible contributors. No single nutrient proves the effect of the whole pattern.
+
+## What the evidence does not establish
+
+- **Not a medication replacement:** a useful dietary pattern does not justify delaying evaluation, stopping medication, changing a dose, or ignoring a clinician's action plan.
+- **No guaranteed individual effect:** average trial results do not predict one person's response. Baseline blood pressure, sodium exposure, medication, kidney function, energy intake, food access, and adherence can all matter.
+- **Limited duration and real-world certainty:** the landmark feeding trials lasted weeks and provided participants' food. Free-living execution over months or years is less controlled and usually more difficult.
+- **Direct health outcomes remain uncertain:** lower rates of cardiovascular events and diabetes have been associated with greater DASH adherence in prospective cohorts, but those findings are observational and vulnerable to confounding. Long-term DASH-specific randomized trials of heart attack, stroke, or mortality are lacking.
+- **No universal superiority over every healthy pattern:** a 2026 systematic review in adults with elevated blood pressure or hypertension found no statistically significant blood-pressure advantage over other dietary interventions in head-to-head randomized comparisons; heterogeneity was substantial and certainty ranged from moderate to very low. DASH is better supported than most fad diets for blood pressure, but that does not make it uniquely best for every person or goal.
+- **Not inherently a weight-loss diet:** weight can change when total intake changes, but DASH's defining evidence does not require intentional weight loss. Route intentional body change to its owner.
+
+## Why a user may choose it
+
+Common reasons include wanting a guideline-supported food pattern for blood-pressure-oriented eating, preferring ordinary grocery foods over specialty products, seeking a less restrictive alternative to elimination diets, or wanting a structure that can work for omnivorous, vegetarian, family, and culturally specific meals. Name recognition, rankings, and testimonials may increase demand, but they do not strengthen the trial evidence.
+
+Clarify whether the real job is a definition, a practical food-system change, blood-pressure interpretation, medication-linked care, or intentional weight change. The DASH label alone does not settle that routing decision.
+
+## Tradeoffs and adequacy checks
+
+### Sodium burden
+
+The lower-sodium version can require substantial label reading, cooking, and restaurant negotiation. Do not default to exact milligram tracking. Start with the one or two foods that contribute the most sodium in the user's actual routine: often breads, sauces and condiments, soups, cured or deli foods, cheese, packaged meals, or restaurant food. Compare like-for-like labels or use a smaller amount before adding a full tracking system.
+
+Do not focus only on the salt shaker. Also do not remove so much flavor or convenience that the plan becomes unsustainable or intake falls. Acids, herbs, spices, aromatics, toasted seeds, and small amounts of strongly flavored condiments can preserve enjoyment.
+
+### Potassium and kidney or medication risk
+
+DASH naturally raises potassium through foods such as beans, lentils, potatoes, leafy greens, tomatoes, squash, fruit, yogurt, and some fish. That is useful for many adults, but more potassium is not universally safer.
+
+Do not encourage a potassium supplement or potassium-chloride salt substitute without a clinician or pharmacist check when the user has kidney disease, impaired potassium handling, a history of high potassium, significant heart disease, or medicines that affect potassium. Relevant medicines can include renin-angiotensin system drugs, mineralocorticoid receptor antagonists, potassium-sparing diuretics, and prescribed potassium; this list is not exhaustive.
+
+### Dairy and non-dairy adaptations
+
+The classic evidence base used fat-free or low-fat dairy. Lactose-free or lactose-reduced milk and yogurt can preserve that structure. Lactose-free dairy is still dairy and is not suitable for a milk allergy.
+
+For a dairy-free version, flexible options include unsweetened fortified soy milk or yogurt, calcium-set tofu, beans, leafy greens, nuts, seeds, and tahini. Check that the replacement pattern supplies meaningful protein, calcium, and vitamin D rather than assuming every plant beverage is nutritionally equivalent. If the pattern also removes all animal foods, use the narrower vegan reference when available and hand supplement or adequacy questions to `micronutrients-supplements`.
+
+### Fiber, energy, and protein
+
+A rapid increase in beans, whole grains, vegetables, fruit, nuts, and seeds can cause bloating, gas, or stool changes. Add one repeatable food or swap at a time over days to weeks, and keep ordinary fluid intake appropriate to thirst and any clinical plan.
+
+Avoid turning "low fat" into low energy, low satiety, or a diet built from sweetened low-fat products. Preserve enough food, protein, carbohydrate for demand, and satisfying unsaturated fat. Frailty, low appetite, under-fueling, or high training demand outranks tighter sodium or serving targets.
+
+## Lowest-risk practical execution
+
+Use the smallest version that solves the user's problem. A useful starting sequence is:
+
+1. **Add one DASH anchor to a routine meal:** a fruit or vegetable, beans or lentils, a whole grain or potato, or a tolerated dairy or fortified alternative.
+2. **Change one frequent sodium source:** compare two breads or sauces, choose no-salt-added when practical, drain and rinse canned beans, use half the seasoning packet, or put a salty condiment on the side.
+3. **Build one repeatable meal scaffold:** a protein or legume, a produce item, a grain or starchy food, and enough fat or sauce for satisfaction.
+
+Concrete flexible options include:
+
+- oats with fruit, plain yogurt or fortified soy yogurt, and nuts or seeds;
+- beans or lentils with rice or corn tortillas, vegetables, and avocado or an unsaturated oil;
+- fish, tofu, chicken, or eggs with potatoes or a whole grain and vegetables;
+- a bean-based soup or stew with extra vegetables and bread, using lower-sodium stock when feasible; or
+- an emergency meal of drained beans, frozen vegetables, and a microwave grain, followed by fruit or yogurt if wanted.
+
+These are examples, not a menu or compliance test.
+
+### Affordability
+
+Use dried or rinsed canned beans, lentils, oats, rice, potatoes, frozen vegetables, store-brand dairy or fortified alternatives, eggs, peanut butter, and seasonal or canned fruit. "No salt added" is useful when available, but comparing brands or draining and rinsing can be a more realistic fallback. Spend effort on foods eaten often rather than expensive specialty products.
+
+### Cultural and family fit
+
+Keep the user's staple foods and meal identity. Adjust proportions, cooking method, or the dominant sodium source rather than replacing the cuisine. Examples could include dal with rice and vegetables, beans with corn tortillas and cabbage, a tofu-and-vegetable stir-fry with sauce used to taste, or a familiar stew with more legumes and a lower-sodium stock. Serve sauces, cheese, cured meat, or salty garnishes separately when family members need different amounts.
+
+### Restaurants and travel
+
+Restaurant sodium is difficult to estimate, so do not pretend precision. Choose a meal with a vegetable, fruit, bean, fish, poultry, tofu, or whole-grain component; ask for sauce or dressing on the side or less added salt when feasible; and use the next ordinary meal rather than compensation after a salty meal. A convenience meal that preserves adequate intake is better than skipping food to stay "perfect."
+
+## Common failure modes
+
+- treating 1,500 mg as a moral pass/fail rule or a universally safer target;
+- tracking every milligram before identifying the few major sodium sources;
+- focusing on table salt while ignoring packaged foods, sauces, and restaurant meals;
+- replacing dairy with a sweetened, low-protein, unfortified beverage and assuming adequacy is unchanged;
+- increasing fiber abruptly and interpreting predictable discomfort as proof the pattern is unsuitable;
+- cutting sodium or fat so aggressively that meals become unsatisfying or energy and protein intake fall;
+- using potassium supplements or salt substitutes without checking kidney and medication context;
+- changing or skipping blood-pressure medicine because of a few readings;
+- turning DASH into a short cleanse, rigid serving-count challenge, separate family menu, or weight-loss promise; and
+- escalating restriction when blood pressure remains a clinical concern instead of handing off.
+
+## Monitoring, stop conditions, and off-ramps
+
+Use low-burden feedback first:
+
+- Is the meal structure repeatable within the user's budget, culture, schedule, and household?
+- Are hunger, energy, training, and intake still adequate?
+- Are fiber and dairy or non-dairy substitutions tolerated?
+- Did the user change one meaningful sodium source rather than merely add tracking work?
+
+When blood pressure is the actual outcome and the user already has a home-monitoring plan, use a validated upper-arm cuff, the clinician's measurement schedule and action thresholds, and averages or trends rather than one isolated reading. Do not use cuffless wearable estimates for diagnosis or medication decisions.
+
+Pause the new sodium or potassium change and seek prompt clinical guidance for recurrent lightheadedness, fainting, unusual weakness, palpitations, confusion, or vomiting, especially when blood-pressure or potassium-affecting medicine is involved. Follow the parent's urgent-care gate for severe symptoms or acute deterioration.
+
+Useful off-ramps include:
+
+- keep the produce, legume, whole-grain, and meal-structure gains while dropping a burdensome numeric sodium target;
+- use standard-sodium rather than lower-sodium DASH when the stricter version is not safe, adequate, or sustainable;
+- choose another evidence-based heart-healthy pattern when it fits the user's culture, household, access, or preferences better; or
+- stop tightening food rules and hand off when blood pressure remains concerning, symptoms appear, or medication or kidney decisions become central.
+
+Changing course is adaptation, not failure.
+
+## Groups needing extra caution
+
+Do not personalize sodium, potassium, fluid, or blood-pressure treatment from this reference alone for users with:
+
+- kidney disease, impaired potassium handling, prior high potassium, or dialysis;
+- significant heart disease, fluid limits, or clinician-prescribed sodium or potassium limits;
+- blood-pressure, diuretic, potassium, or other medication that can change blood pressure or electrolyte safety;
+- pregnancy or breastfeeding;
+- age under 18;
+- frailty, sarcopenia, low appetite, substantial unintentional weight loss, or probable under-fueling;
+- a history or signs of an eating disorder or harmful restriction;
+- post-bariatric anatomy, a severe food allergy, or another therapeutic diet; or
+- high-volume heat training, unusually high sweat losses, or an individualized electrolyte plan.
+
+General meal ideas may still help, but the relevant clinician or registered dietitian should own condition-specific targets and medication-linked changes.
+
+## Murph handoffs
+
+- Use `cardiometabolic-health` when blood-pressure readings, diagnosis, risk, targets, or marker interpretation are the main job.
+- Use `chronic-illness-support`, care navigation, and the user's clinician or pharmacist when kidney disease, heart disease, medication, prescribed sodium or potassium limits, or therapeutic blood-pressure management is central.
+- Use `food-journal` for low-pressure observation of actual meals, restaurant context, or the few recurring sodium sources; do not create a DASH adherence score.
+- Use `gut-digestion` for persistent bloating, constipation, diarrhea, or a symptom-driven fiber strategy.
+- Use `body-composition` when intentional weight or waist change is the real goal, then return here for meal execution.
+- Use `micronutrients-supplements` for potassium supplements, salt-substitute labels, or calcium, vitamin D, vitamin B12, or other adequacy questions that cannot be answered from food structure alone.
+- Use the parent `nutrition-strategy` hydration and performance lane when training, heat, sweat, or fueling demand makes a generic lower-sodium target unsafe or impractical.
+- Use `experiment-onboarding` only after the user chooses one specific, safe change with a bounded window and stop conditions.
+
+## Source register
+
+Reviewed **2026-08-09**. Stable DOI, PubMed, and official-source links are provided without relying on popularity rankings or testimonial evidence.
+
+1. **2025 guideline for prevention, detection, evaluation, and management of high blood pressure in adults.** Guideline and official summary; supports DASH as one heart-healthy pattern while keeping targets and medication care clinical. [DOI](https://doi.org/10.1161/CIR.0000000000001356) | [Official summary](https://professional.heart.org/en/science-news/2025-high-blood-pressure-guideline/top-things-to-know)
+2. **NHLBI DASH Eating Plan, Following DASH, and Health Benefits of DASH.** Official pattern definition, 2,300 mg and 1,500 mg sodium variants, lactose-free options, and summaries of the controlled feeding trials; pages updated 2026-02-25. [Eating plan](https://www.nhlbi.nih.gov/health/dash-eating-plan) | [Following DASH](https://www.nhlbi.nih.gov/health/dash/following-dash) | [Health benefits](https://www.nhlbi.nih.gov/health/dash/health-benefits)
+3. **A clinical trial of the effects of dietary patterns on blood pressure.** Randomized controlled feeding trial; 459 adults, eight-week intervention, short-term blood-pressure surrogate. [DOI](https://doi.org/10.1056/NEJM199704173361601) | [PMID 9099655](https://pubmed.ncbi.nlm.nih.gov/9099655/)
+4. **Effects on blood pressure of reduced dietary sodium and the DASH diet.** Randomized controlled feeding trial; 412 adults, three sodium levels, strongest short-term effect from lower sodium plus DASH. [DOI](https://doi.org/10.1056/NEJM200101043440101) | [PMID 11136953](https://pubmed.ncbi.nlm.nih.gov/11136953/)
+5. **Dietary Approaches to Stop Hypertension diet and blood pressure reduction in adults with and without hypertension: a systematic review and meta-analysis of randomized controlled trials.** Thirty randomized trials, 5,545 participants; moderate-certainty evidence for blood-pressure reduction versus control. [DOI](https://doi.org/10.1093/advances/nmaa041) | [PMID 32330233](https://pubmed.ncbi.nlm.nih.gov/32330233/)
+6. **Effectiveness of the DASH diet versus alternative dietary interventions for hypertension management: a systematic review and meta-analysis.** Published online 2026-07-23; no statistically significant blood-pressure superiority over comparator diets, substantial heterogeneity, and moderate-to-very-low certainty. [DOI](https://doi.org/10.1016/j.cpcardiol.2026.103397) | [PMID 42492868](https://pubmed.ncbi.nlm.nih.gov/42492868/)
+7. **DASH dietary pattern and cardiometabolic outcomes: an umbrella review of systematic reviews and meta-analyses.** Separates controlled-trial biomarker evidence from prospective-cohort event associations; direct outcome certainty was lower. [DOI](https://doi.org/10.3390/nu11020338) | [PMID 30764511](https://pubmed.ncbi.nlm.nih.gov/30764511/)
+8. **How potassium can help prevent or treat high blood pressure.** Official safety source for kidney disease, medication, supplements, and potassium-based salt substitutes; reviewed 2025-08-14. [Official page](https://www.heart.org/en/health-topics/high-blood-pressure/changes-you-can-make-to-manage-high-blood-pressure/how-potassium-can-help-control-high-blood-pressure)

package/node_modules/@murphai/assistant-engine/skills/nutrition-strategy/references/named-diets/intermittent-fasting.md

@@ -0,0 +1,153 @@
+# Intermittent fasting
+
+Use this progressive-disclosure reference only after the parent `nutrition-strategy` contract selects the named-diet lane. The parent remains authoritative for routing, safety, tracking burden, clinical boundaries, and answer shape. This file distinguishes intermittent-fasting variants and adds diet-specific execution guidance; it is not a registered skill or a therapeutic fasting protocol.
+
+## Decision summary
+
+- Intermittent fasting is a family of timing and intermittent-energy-restriction patterns, not one diet. Time-restricted eating, 16:8, 14:10, 5:2, alternate-day fasting, and one meal a day are not interchangeable.
+- In adults studied for overweight, obesity, or cardiometabolic risk, fasting patterns can produce modest short-term changes in weight and surrogate markers, but they are generally not superior to a sustainable non-fasting energy-restriction pattern. Evidence for adverse effects, long-term durability, and direct health outcomes is limited.
+- Earlier time-restricted eating has a modest average advantage over late time-restricted eating for some surrogate outcomes, but eating-window duration has not shown a consistent dose-response. A shorter window is not automatically better.
+- For an adult without caution flags who wants a simple timing boundary, prefer the least restrictive viable version: first remove unwanted late eating or use a roughly 12-hour overnight gap; if a named fasting pattern is still desired, start with 14:10. Do not jump to 5:2, alternate-day fasting, or OMAD.
+- Adequate food, medication safety, eating-disorder recovery, sleep, social life, and training fuel outrank keeping a fasting streak.
+
+## What the label can mean
+
+| Pattern | Common meaning | Material distinction |
+|---|---|---|
+| **Time-restricted eating (TRE)** | Food and energy-containing drinks are kept within a recurring daily window; total energy is not necessarily prescribed. | The window may be early, middle, late, or self-selected. TRE does not tell Murph whether the person eats for 6, 8, 10, or 12 hours, or whether intake falls. |
+| **14:10** | A 14-hour fast and 10-hour eating window each day. | Usually leaves room for three eating occasions and is the least restrictive common named version. It is not a biological threshold. |
+| **16:8** | A 16-hour fast and 8-hour eating window each day. | Often compresses intake to two or three occasions. A noon-to-8 p.m. version differs from an 8 a.m.-to-4 p.m. version in sleep, training, family, and circadian fit. |
+| **5:2** | Five usual-intake days and two days each week with little or no energy; research often uses modified, not zero-energy, fast days. | It is a weekly intermittent-energy-restriction pattern, not daily TRE. Consecutive and nonconsecutive fast days create different practical burdens. |
+| **Alternate-day fasting (ADF)** | Alternating usual-intake days with zero- or very-low-energy days. | Restriction occurs about every other day, so fueling, social, and rebound-eating pressure is usually greater than with 5:2 or moderate TRE. |
+| **One meal a day (OMAD)** | One meal in a short daily window, often implying about 22-23 hours without food. | This is a highly compressed daily pattern with direct evidence much thinner than for broader TRE. It makes adequate total intake and distributed protein or training fuel harder. |
+
+A label also may not specify water, energy-free drinks, milk, supplements, or medication timing. Do not turn "clean fasting" rules into a purity test. Prescribed medication, hydration needs, and safe fueling take precedence over preserving a technical fast.
+
+## What the evidence supports
+
+### Comparative efficacy
+
+- The 2026 Cochrane review found that intermittent fasting made little to no difference in weight loss or quality of life compared with regular dietary advice in adults with overweight or obesity. Included trials lasted no longer than 12 months, and adverse-event evidence was uncertain.
+- A 2025 network meta-analysis of 99 randomized trials found broadly similar weight and cardiometabolic effects for intermittent fasting and continuous energy restriction. ADF had the only average advantage over continuous restriction for body weight, about 1.3 kg, which was below that review's prespecified threshold for a clinically important difference. Most trials were short, and most participants had an existing health condition.
+- A 10-hour TRE trial published in 2026 found short-term reductions in weight and fat in selected adults with obesity. That result shows that a feasible TRE protocol can work for some people; it does not establish superiority, long-term durability, or direct disease prevention.
+- A one-year 5:2 trial found similar modest weight outcomes to standard brief dietary advice. Participants initially liked the simple rule, but reported adherence fell substantially over time. Preference can make a pattern useful without making it more efficacious.
+
+### Meal timing versus energy intake
+
+- TRE often reduces eating opportunities and therefore energy intake. Trials that prescribe similar energy intake in both groups generally find little or no added weight-loss advantage from the time window itself.
+- Timing may still have independent circadian effects. A 2026 network meta-analysis found early TRE modestly better than late TRE for body weight and fasting insulin, while rankings by eating-window length were inconsistent. These are surrogate outcomes, not proof that an early window prevents disease or extends life.
+- A short, intended-isocaloric crossover trial found that early and late 8-hour windows shifted circadian timing but did not produce clinically meaningful cardiometabolic differences. This supports a plausible clock effect while limiting claims about its health importance.
+- Do not promise that 16 hours, 20 hours, or any other duration activates a clinically validated "autophagy stage." Cellular mechanisms and ketone production are plausible research pathways, not a consumer clock for guaranteed outcomes.
+
+### What is not established
+
+- No fasting variant has high-certainty randomized evidence for lower all-cause mortality, cardiovascular events, cancer, dementia, or longer human lifespan.
+- Evidence is dominated by short- to medium-term trials in adults with overweight, obesity, or metabolic risk. It transfers poorly to pregnancy, breastfeeding, adolescence, frail older adults, people with low intake, eating-disorder recovery, lean athletes, or high-volume training.
+- Short-term weight or biomarker changes do not establish long-term adherence, diet quality, psychological safety, or net benefit for a specific person.
+
+## Why users may choose it
+
+Common reasons include fewer food decisions, a clear boundary around late-night eating, dislike of calorie tracking, a schedule that naturally favors later or earlier meals, appetite preference, weight or biomarker goals, religious or cultural familiarity with fasting, and claims about metabolism, longevity, or "discipline." Clarify the actual job before recommending a protocol.
+
+Treat a preference for simple rules as useful fit information. Do not treat hunger tolerance, longer fasts, or a tighter window as virtue. Religious fasting may have distinct food, fluid, timing, and community rules; do not silently replace it with a research TRE protocol.
+
+## Lowest-risk practical execution
+
+1. **Confirm the goal and owner.** A definition or ordinary meal-timing question can stay here. Intentional fat loss, muscle gain, calorie targets, or a plateau belongs with `body-composition`; marker, condition, or medication goals belong with clinical owners.
+2. **Start wider than the user's maximum tolerance.** First try a consistent overnight gap near 12 hours or a 14:10 pattern. Tighten only when the person is eating enough, feels well, and can name a benefit that the tighter window adds.
+3. **Fit the window around real life.** Protect sleep opportunity, work, family meals, culture, and the most demanding training. If early and late windows fit equally well, an earlier finish is a reasonable preference; do not force an early window that causes missed recovery food, very early waking, or social isolation.
+4. **Protect meal quality and distribution.** Keep roughly three meaningful protein-containing eating occasions when feasible. Build meals from flexible options: eggs, yogurt, milk, tofu, beans, fish, poultry, meat, or fortified alternatives; oats, bread, rice, potatoes, pasta, tortillas, fruit, or other demand-matched carbohydrate; vegetables or another fiber-rich food; and nuts, seeds, avocado, olive oil, cheese, sauce, or another satisfying fat or flavor.
+5. **Keep an easy fallback.** Examples include yogurt with fruit and cereal; a bean-and-rice bowl; eggs or tofu with toast and fruit; a sandwich with milk or a fortified alternative; or a frozen meal plus produce. Convenience food is compatible with a useful eating window.
+6. **Return to the next normal meal after a deviation.** Do not compensate for a late meal, social event, or large meal by extending the next fast.
+
+### Variant-specific guidance
+
+- **14:10:** Preferred first named version. It often preserves breakfast, lunch, and dinner or three shifted eating occasions without requiring a separate diet.
+- **16:8:** Use only when an 8-hour window still permits adequate meals and training fuel. If intake becomes rushed, the last meal becomes very large, or protein is compressed into one or two weak occasions, widen to 14:10.
+- **5:2:** Not a default first step. Do not invent a fast-day calorie target or rigid menu in this reference. If an appropriate user has already chosen it, favor nonconsecutive lower-demand days, ordinary balanced eating on the other days, and no "earning" or repayment cycle.
+- **ADF:** Do not present its small average short-term weight advantage as a reason to accept greater burden. Frequent low-intake days commonly collide with training, social meals, and adequate intake; 14:10 or a regular daily pattern is the preferred off-ramp.
+- **OMAD:** Do not recommend as a routine starting pattern. One meal makes total energy, protein distribution, carbohydrate around exercise, fiber, micronutrients, and medication-with-food needs harder to meet and can promote an overly large meal or rebound eating. Offer two or three eating occasions within a wider window instead.
+
+## Exercise, protein, and under-fueling
+
+- Fasted exercise is optional, not a marker of fitness and not clearly superior for performance or fat loss. A short easy session may be fine when the person feels normal; this does not justify applying the same approach to long, hard, high-volume, or repeated sessions.
+- Place the eating window so a demanding session can receive familiar carbohydrate beforehand and carbohydrate plus protein afterward. A banana with yogurt, toast with eggs, cereal with milk, a sandwich, or rice with tofu or poultry are options. Breaking the fast is the correct repair when fuel is needed.
+- Daily adequacy and distributed protein matter more than defending a narrow window. If a strength, muscle, or recovery goal cannot fit the parent's protein distribution guidance, widen the window rather than cramming everything into one meal.
+- Watch for persistent fatigue, falling performance, slow recovery, recurrent illness or injury, dizziness, feeling unusually cold, irritability, sleep disruption, menstrual change, reduced libido, food preoccupation, or unexpected weight loss. These are reasons to stop tightening the window and assess under-fueling, not proof of low motivation.
+
+## Common failure modes and repairs
+
+| Failure mode | Lowest-burden repair |
+|---|---|
+| Skipping food all day, then eating rapidly or beyond comfort at night | Move food earlier, widen to 14:10 or regular meals, and add a planned protein-plus-carbohydrate meal or snack before extreme hunger. |
+| Recurrent rebound or loss-of-control eating after fasting | Stop the fasting protocol and restore regular nourishment; route eating-disorder risk rather than adding more rules. |
+| A late window worsens sleep, reflux, or family fit | End earlier if feasible, widen the window, or use regular meals. Route persistent digestive symptoms to `gut-digestion`. |
+| A strict early window removes post-training food | Move or widen the window, or eat the recovery meal and end the fast. Training recovery outranks timing theory. |
+| Protein, produce, or total food is crowded out | Add another eating occasion and use easy foods; do not solve low total intake with protein alone. |
+| Excess caffeine is used to suppress hunger | Reduce the fasting demand and eat. Do not use stimulants as a substitute for adequate food or sleep. |
+| Weekday rigidity produces weekend compensation | Use a flexible 12-hour or 14:10 baseline instead of all-or-none rules. |
+| A missed window is treated as failure | Resume at the next ordinary meal. No punishment, extra fasting, or compensatory exercise. |
+| The person tracks every minute without a decision benefit | Drop timer precision. A repeatable meal boundary and a few feedback signals are enough. |
+
+## Monitoring, stop conditions, and off-ramps
+
+Use the lowest useful monitoring burden. A short fit check can ask whether the pattern preserves stable daytime energy, ordinary hunger rather than extremes, sleep, training quality and recovery, social flexibility, and enough food. Do not require calorie logging or daily weighing. Glucose monitoring belongs to an existing clinician-directed plan, not this reference.
+
+**End the fast and follow the person's existing medical plan** for shakiness, sweating, confusion, faintness, or other suspected low-glucose symptoms. Seek urgent help for confusion, fainting, chest pain, severe weakness, inability to keep fluids down, severe dehydration, or symptoms that do not improve.
+
+**Stop the protocol and return to regular nourishment** for recurrent dizziness, headaches or nausea; sleep worsening; persistent fatigue; training decline; menstrual change; food preoccupation; recurrent rebound or loss-of-control eating; substantial unintentional weight loss; or any sign that the window is displacing needed food or medication.
+
+Use the least restrictive off-ramp that resolves the problem:
+
+- 16:8 to 14:10, then to a simple overnight gap or regular meals;
+- 5:2 or ADF to a consistent daily pattern;
+- OMAD to two meals plus a snack, then to three ordinary eating occasions when useful.
+
+If a tighter pattern offers no clear valued benefit, keeping it is unnecessary. Do not wait for harm or treat a wider window as failure.
+
+## Groups needing extra caution
+
+Do not provide a self-directed fasting protocol when any of these are present:
+
+- pregnancy or breastfeeding;
+- adolescence or anyone under 18;
+- frailty, sarcopenia, low appetite, low intake, substantial unintentional weight loss, or recovery from prolonged very low intake;
+- a known or suspected eating disorder, severe restriction, recurrent loss-of-control eating, purging, compulsive exercise, intense food or weight fear, or number sensitivity;
+- current under-fueling or a training load that already exceeds available recovery;
+- post-bariatric surgery or another clinician-managed nutrition plan.
+
+Require clinician or registered-dietitian involvement before fasting when diabetes, recurrent hypoglycemia, kidney disease, advanced liver disease, significant heart disease, a relevant endocrine disorder, or a fluid restriction is material. The same applies to insulin or another medicine that can cause hypoglycemia, any medicine that must be taken with food, or a medicine whose safety changes with dehydration or major meal-timing changes. Do not recommend medication adjustments.
+
+Acute illness, surgery or injury recovery, heat exposure, endurance events, and other periods with increased food or fluid needs are reasons to suspend the fasting goal or obtain individualized guidance.
+
+## Murph handoffs
+
+- `body-composition`: intentional weight, fat, muscle, recomposition, calorie, or plateau strategy; return here for meal execution only.
+- `cardiometabolic-health`, `chronic-illness-support`, and care navigation: glucose, lipids, blood pressure, disease-specific goals, hypoglycemia, medication timing, or therapeutic fasting.
+- `food-journal`: low-pressure observation of timing, hunger, rebound eating, sleep, symptoms, or training response when retrospective data would change a decision.
+- `gut-digestion`: persistent reflux, constipation, nausea, diarrhea, bloating, or symptom-driven elimination work.
+- `micronutrients-supplements`: supplement evidence, label review, dosing, or a specific adequacy concern that food structure cannot answer.
+- `experiment-onboarding`: only after the user chooses one bounded, safe version with feedback and stop conditions.
+- `behavior-followthrough`: reminders or repeated implementation repair after the plan is accepted.
+
+## Source register
+
+Reviewed **2026-08-09**. Prefer the synthesis-level sources for overall claims; use individual trials to understand a specific protocol, not to override the full evidence base.
+
+| Source | Evidence type and population | Decision use and limits |
+|---|---|---|
+| [Intermittent fasting for adults with overweight or obesity (Cochrane, 2026)](https://doi.org/10.1002/14651858.CD015610.pub2) | Systematic review of 22 randomized trials, 1,995 adults; interventions up to 12 months. | Best current anchor for little-to-no comparative difference in weight or quality of life and for uncertain adverse-event evidence. Limited to adults with overweight or obesity and short follow-up. |
+| [Intermittent fasting strategies and cardiometabolic risk factors (BMJ, 2025)](https://doi.org/10.1136/bmj-2024-082007) | Network meta-analysis of 99 randomized trials, 6,582 adults; TRE, ADF, whole-day fasting, continuous restriction, and usual intake. | Distinguishes variants. Supports broadly similar effects to continuous restriction and only a small ADF weight advantage that did not reach the review's clinical-importance threshold. Outcomes were mainly weight and risk markers. |
+| [Timing and eating duration of TRE (BMJ Medicine, 2026)](https://doi.org/10.1136/bmjmed-2024-001071) | Systematic review and network meta-analysis of 41 randomized trials, 2,287 participants. | Supports a modest early-versus-late TRE advantage for some surrogate outcomes and inconsistent rankings by window length. No direct disease-event evidence. |
+| [Calorie restriction with or without TRE (2022)](https://doi.org/10.1056/NEJMoa2114833) | Twelve-month randomized trial in 139 adults with obesity; both groups received calorie restriction. | Supports no clear added weight or metabolic benefit from an 8-hour early window when energy restriction was shared. Population and prescribed intervention limit generalization. |
+| [Intended-isocaloric early versus late TRE (2025)](https://doi.org/10.1126/scitranslmed.adv6787) | Two-period crossover trial in 31 women with overweight or obesity; two weeks per 8-hour window. | Shows meal timing can shift circadian phase without clinically meaningful short-term cardiometabolic differences. Small, short, and single-population. |
+| [TREAD 10-hour TRE trial (2026)](https://doi.org/10.1002/oby.70228) | Twelve-week randomized trial in adults with obesity. | Recent evidence that a moderate 10-hour window can reduce weight and fat in a selected group. Does not establish superiority, durability, or direct health outcomes. |
+| [TREAT 16:8 trial (2020)](https://doi.org/10.1001/jamainternmed.2020.4153) | Twelve-week randomized trial of noon-to-8 p.m. TRE in 116 adults with overweight or obesity. | Shows that a common late 16:8 schedule is not reliably superior to structured meals for weight or cardiometabolic markers. |
+| [Randomized trial of the 5:2 diet (2021)](https://doi.org/10.1371/journal.pone.0258853) | One-year pragmatic trial in 300 adults with obesity. | Supports similar modest outcomes to standard advice, higher initial preference, and substantial adherence decline. Does not justify treating 5:2 as superior. |
+| [TRE and sleep systematic review (2024)](https://doi.org/10.3389/fnut.2024.1419811) | Systematic review of randomized trials lasting at least eight weeks. | Short- to medium-term TRE did not usually worsen sleep, but findings were limited and population-dependent. Personal sleep fit still needs monitoring. |
+| [Standards of Care in Diabetes - 2026: nutrition and fasting](https://doi.org/10.2337/dc26-S005) and [hypoglycemia](https://doi.org/10.2337/dc26-S006) | Current major clinical guideline. | Supports medication review, glucose-safety planning, and hypoglycemia precautions rather than self-directed fasting in medication-linked diabetes. Not a general endorsement of fasting. |
+| [Consensus statement on Relative Energy Deficiency in Sport (2023)](https://doi.org/10.1136/bjsports-2023-106994) | Multidisciplinary sport-health consensus. | Supports prioritizing adequate energy availability and taking performance, recovery, reproductive, bone, and psychological warning signs seriously. It does not diagnose under-fueling from chat data. |
+| [Practical guideline on nutrition and hydration in geriatrics (2022)](https://doi.org/10.1016/j.clnu.2022.01.024) | Major clinical nutrition guideline for older adults. | Supports protecting energy, protein, hydration, muscle, and function in frailty or malnutrition risk rather than adding restriction. |
+| [Pediatric guidance on preventing obesity and eating disorders (2016)](https://doi.org/10.1542/peds.2016-1649) | Clinical report focused on adolescents. | Supports avoiding dieting and meal skipping as self-directed weight-control tools in youth and favoring sustainable health behaviors. |
+| [Eating-disorders recognition and treatment guideline NG69](https://www.nice.org.uk/guidance/ng69) | Current official guideline, updated with surveillance. | Supports prompt recognition and specialist care when restriction, loss-of-control eating, purging, compulsive exercise, or medical instability is present. It is not evidence that fasting causes every eating disorder. |
+| [Official research summary on intermittent fasting and aging](https://www.nia.nih.gov/news/research-intermittent-fasting-shows-health-benefits) | Public research summary from a national aging institute. | Separates plausible mechanisms and animal findings from the lack of long-term human lifespan evidence; supports rejecting transformation and longevity promises. |

package/node_modules/@murphai/assistant-engine/skills/nutrition-strategy/references/named-diets/ketogenic.md

@@ -0,0 +1,149 @@
+# Ketogenic diet
+
+Apply the parent `nutrition-strategy` contract first. This progressive-disclosure reference adds ketogenic-diet distinctions, evidence, and failure handling. It does not register a skill, replace the parent's routing or safety rules, authorize medication changes, or turn clinician-managed ketogenic diet therapy into a wellness plan.
+
+## Choose the lane before answering
+
+- **Ordinary weight-loss or wellness use:** use this reference for definition, fit, comparison, low-risk execution, and troubleshooting in a medically uncomplicated adult. If intentional weight or waist change is the main job, `body-composition` owns the strategy and this reference can support food execution.
+- **Marker-, disease-, or medication-led use:** diabetes, glucose-lowering or blood-pressure medication, a large lipid response, kidney or liver disease, pancreatitis, or another clinical constraint changes the risk. Hand off to `cardiometabolic-health`, `chronic-illness-support`, care navigation, and the user's clinician rather than prescribing the diet or medication adjustments.
+- **Therapeutic ketogenic diet therapy:** classic ketogenic, medium-chain-triglyceride, modified Atkins, and low-glycemic-index therapies used for drug-resistant epilepsy or selected rare metabolic conditions are clinician-managed protocols. Their ratios, ketone targets, laboratory monitoring, supplements, and contraindication screening are not ordinary keto advice.
+
+## Definition and meaningful variants
+
+A ketogenic diet restricts carbohydrate enough to produce sustained **nutritional ketosis**, usually while providing adequate protein and deriving most remaining energy from fat. Research definitions vary; common ordinary versions use less than about 50 g carbohydrate per day, often 20–50 g, but the intake that produces ketosis differs by person and context. “Low carbohydrate” does not necessarily mean ketogenic, and a label does not reveal food quality, energy intake, or whether ketosis is present.
+
+**Nutritional ketosis is not diabetic ketoacidosis (DKA).** Nutritional ketosis is a regulated fuel state without metabolic acidosis. DKA is a life-threatening state of substantial ketone accumulation plus acidosis, usually in diabetes; with an SGLT2 inhibitor, glucose can be only modestly elevated. A home glucose or ketone result must not be used to dismiss concerning symptoms.
+
+Variants that materially change the answer:
+
+- An **unsaturated-fat-forward, minimally processed version** can emphasize fish, olive oil, avocado, nuts, seeds, tofu or tempeh, eggs, other suitable protein foods, nonstarchy vegetables, and modest low-carbohydrate fruit.
+- A **saturated-fat- or processed-meat-heavy version** may meet the carbohydrate rule while worsening diet quality and, for some people, LDL cholesterol or apolipoprotein B (apoB).
+- **Targeted or cyclical keto** adds carbohydrate around training or on selected days. Comparative evidence is sparse; use it as an individual fueling adaptation, not a proven superior variant.
+- **Vegetarian keto** is possible but more planning-intensive because protein, fiber, iron, calcium, iodine, and food variety can become harder to cover.
+- **Carnivore-style eating is a separate, more restrictive pattern.** Do not treat its risks or evidence as interchangeable with a produce-containing ketogenic diet.
+
+## What the evidence supports—and does not
+
+Most ordinary keto trials last weeks to months, relatively few extend beyond one to two years, and most measure weight or laboratory surrogates rather than cardiovascular events, mortality, disability, or quality of life. Implementations and comparison diets vary, so a result for “low carbohydrate” is not automatically a result for sustained nutritional ketosis.
+
+- **Weight and appetite:** keto can reduce appetite for some people and can produce short-term weight loss. Early scale loss includes glycogen-associated water and is not all body fat. In a large Cochrane review, low-carbohydrate diets produced little or no clinically important weight-loss advantage over balanced-carbohydrate weight-loss diets through one to two years. Adherence, energy intake, protein, food quality, and support appear more decisive than ketosis itself.
+- **Glycemia in type 2 diabetes:** low- and very-low-carbohydrate patterns can improve A1C and reduce medication needs in some people over roughly the first six months. Comparative advantages usually diminish with longer follow-up. This is a medication-linked clinical lane because insulin, sulfonylureas, and other therapies may need prompt clinician adjustment; very-low-carbohydrate eating should be avoided with SGLT2 inhibitors because of ketoacidosis risk.
+- **Lipids and cardiovascular risk:** triglycerides often fall and HDL cholesterol may rise, but LDL cholesterol and apoB responses are heterogeneous and can rise substantially. An improved triglyceride-to-HDL pattern does not cancel an apoB increase. Trials have not established that ordinary keto prevents cardiovascular events or extends life.
+- **Training:** increased fat oxidation is a mechanism, not proof of better performance. Compared with higher-carbohydrate diets, evidence in healthy exercising adults is mostly neutral or unfavorable for endurance and high-intensity performance; maximal strength is often similar. Most performance studies are short, and elite-athlete studies have been especially limited.
+- **Therapeutic use:** the best-established direct clinical outcome is seizure reduction in some people with drug-resistant epilepsy under specialist care, with stronger randomized evidence in children than adults. Tolerability and dropout remain material.
+- **Other disease claims:** evidence for cancer, dementia, psychiatric conditions, migraine, polycystic ovary syndrome, and broad “metabolic healing” claims remains condition-specific, preliminary, inconsistent, or surrogate-based. Do not present keto as a stand-alone treatment or advise replacing established care.
+
+Across umbrella reviews, many reported keto associations are supported by low or very-low certainty evidence. Popularity, testimonials, ketone production, lower insulin exposure, or increased fat oxidation can explain interest or a plausible mechanism; none proves comparative long-term benefit.
+
+## Why a user may choose it
+
+Likely reasons include a preference for clear food rules, less hunger on a higher-fat pattern, avoiding refined carbohydrate foods, rapid early scale feedback, a short-term glucose goal, fit with familiar foods, or community influence. A therapeutic indication is a different reason and belongs to a clinical team. Identify the actual goal before optimizing ketosis; the useful part may be fewer refined foods, more protein, or simpler meal structure rather than the ketotic state.
+
+## Diet-specific tradeoffs and adequacy concerns
+
+- **Early fluid and electrolyte shifts:** carbohydrate restriction reduces glycogen-associated water and can increase sodium and water loss early. Headache, thirst, fatigue, lightheadedness, cramps, or a rapid scale drop may follow. These symptoms are not proof that the diet is “working,” and universal salt, potassium, fluid, or electrolyte targets are unsafe. Use ordinary fluids and meals; medication, kidney, heart, blood-pressure, and endocrine contexts need individualized guidance.
+- **Constipation and fiber:** removing grains, legumes, and much fruit can sharply reduce fiber and stool volume. Constipation is among the most commonly reported adverse effects. Protect fiber with nonstarchy vegetables, avocado, berries if they fit, nuts, seeds, chia or ground flax, and adequate fluid; increase fiber gradually. Persistent or painful symptoms belong to `gut-digestion` and clinical care when red flags are present.
+- **Food quality and micronutrients:** a fat-first implementation can displace protein, produce, calcium-rich foods, potassium-rich foods, magnesium, folate, thiamin, and food variety. Supplements do not automatically repair a poor pattern and belong to `micronutrients-supplements` when evidence, labels, dose, or safety is the question.
+- **Training and recovery:** reduced carbohydrate availability can impair repeated sprints, high-volume lifting, team sport, and hard endurance work, particularly during adaptation. Falling output, poor recovery, dizziness, sleep disruption, recurrent injury, menstrual changes, or escalating food preoccupation are reasons to restore fuel or leave the diet—not challenges to out-discipline.
+- **LDL-C and apoB:** the average response hides wide variation. Diets rich in butter, coconut oil, fatty processed meats, and cheese may be especially problematic for some people. A substantial rise deserves reassessment even when weight, glucose, triglycerides, or HDL improve.
+- **Practical burden:** restaurants, travel, family meals, cultural staples, cost, and limited convenience choices can make strict ketosis brittle. A less restrictive lower-carbohydrate or Mediterranean-style pattern may preserve the useful core with less burden.
+
+## Extra caution: do not run an ordinary self-directed trial
+
+Use a clinician-led handoff rather than ordinary keto execution when any of these apply:
+
+- type 1 diabetes, current SGLT2-inhibitor use, or a history of DKA;
+- insulin, sulfonylurea, or another glucose-lowering medicine; clinically important blood-pressure treatment can also require closer observation as intake and fluid balance change;
+- pregnancy or breastfeeding, age under 18, frailty, low intake, substantial unintentional weight loss, or suspected under-fueling;
+- current or prior eating disorder, binge–restrict cycling, purging, compulsive exercise, intense fear around food or weight, or marked number sensitivity;
+- kidney disease, significant liver disease, a history of pancreatitis, severe hypertriglyceridemia, or a known or suspected inherited disorder of fatty-acid transport, oxidation, ketone use, or related metabolism;
+- a proposed therapeutic use for epilepsy, cancer, neurologic or psychiatric disease, or another diagnosed condition.
+
+Known disorders that prevent safe use of fat as a major fuel can make ketogenic therapy dangerous and cannot be screened out in chat. Do not advise starting keto to “test” whether one is present. Never tell a user to stop, reduce, or time medication differently to make keto possible.
+
+## Lowest-risk ordinary execution
+
+For a medically uncomplicated adult who still chooses an ordinary trial:
+
+1. **Define one job and a bounded review point.** Examples are steadier appetite, simpler weekday meals, or tolerability during a specific training block. Ketosis itself is not the outcome, and routine ketone testing is unnecessary for ordinary wellness use.
+2. **Keep a protein anchor at regular meals.** Flexible options include fish, eggs, poultry, tofu, tempeh, plain Greek yogurt or cottage cheese if suitable, and unprocessed meat in amounts that fit the person's preferences.
+3. **Build in produce and fiber rather than adding fat alone.** Options include leafy greens, broccoli or cauliflower, zucchini, peppers, mushrooms, green beans, avocado, a small serving of berries, nuts, seeds, chia, and ground flax.
+4. **Favor unsaturated fats.** Olive oil, avocado, nuts, seeds, and fish are stronger defaults than building the diet around butter, coconut oil, processed meats, cheese, cream, or specialty “fat” snacks.
+5. **Keep one realistic fallback.** Examples: eggs or tofu with frozen vegetables; canned fish with a bagged salad and olive oil; plain Greek yogurt with nuts and berries if it fits; or a prepared protein with microwave vegetables. The fallback should prevent missed meals, not enforce perfection.
+6. **Use normal hydration, not forced electrolyte loading.** Drink to ordinary thirst and conditions, eat regular meals, and do not add a fixed sodium or potassium regimen. Persistent lightheadedness, palpitations, weakness, or inability to hydrate is a stop signal.
+7. **Do not stack restriction.** Avoid combining keto by default with fasting, a large calorie deficit, dehydration practices, or abrupt increases in training. Do not compensate after a higher-carbohydrate meal; return to the next ordinary meal.
+
+Exact carbohydrate counting is optional, not automatic. When precision is genuinely needed, use verified labels and the parent exact-food workflow; do not invent “net carbohydrate” values or turn ketone readings into an adherence score.
+
+## Common failure modes
+
+- Chasing the highest ketone number, buying exogenous ketones, or treating ketosis as the health outcome.
+- Calling early water loss fat loss or using rapid scale change as proof of superior efficacy.
+- Letting added fat displace protein, vegetables, fiber, or total dietary variety.
+- Replacing carbohydrate mainly with butter, coconut oil, processed meat, cheese, and packaged keto products.
+- Ignoring constipation, declining training, poor sleep, low energy, menstrual changes, or food preoccupation as an unavoidable “adaptation.”
+- Starting without a diabetes-medication plan, self-adjusting medication, or using a normal-ish glucose reading to dismiss possible DKA.
+- Treating a higher-carbohydrate meal as failure, then fasting, overexercising, or restricting harder.
+- Continuing despite a large LDL-C or apoB rise because triglycerides, HDL, weight, or glucose moved favorably.
+
+## Monitoring, stop conditions, and off-ramps
+
+### Monitor with the lowest useful burden
+
+During the first days and weeks, watch energy, thirst, dizziness, nausea, bowel function, sleep, mood, food preoccupation, training output, recovery, and whether regular meals remain possible. At the chosen review point, ask whether the actual goal improved enough to justify restriction, cost, social burden, and side effects.
+
+For longer use or when cardiovascular risk is relevant, a clinician can decide whether baseline and follow-up LDL-C, non-HDL cholesterol, apoB, triglycerides, glucose measures, blood pressure, kidney or liver markers, and medication review are appropriate. Do not promise that weight loss protects against an adverse apoB response. Therapeutic ketogenic diets require their own specialist monitoring plan.
+
+### Stop and reassess
+
+Leave the self-directed trial and seek appropriate care for:
+
+- repeated hypoglycemia or concerning glucose changes, medication-related symptoms, or inability to follow an existing diabetes sick-day plan;
+- persistent dizziness, fainting, palpitations, severe weakness, vomiting, inability to keep fluids down, or signs of significant dehydration;
+- persistent constipation, marked abdominal pain or distension, vomiting, or blood in stool;
+- sustained decline in training or daily function, recurrent injury, menstrual change, unexpected weight loss, or other under-fueling clues;
+- escalating restriction, bingeing, compensatory behavior, anxiety, or obsessive ketone, carbohydrate, food, or weight checking;
+- a substantial LDL-C, non-HDL cholesterol, or apoB increase, especially if it persists after replacing saturated fats with unsaturated fats.
+
+**Urgent DKA lane:** nausea or vomiting, abdominal pain, rapid or deep breathing, confusion, unusual drowsiness, severe weakness, or acute deterioration in a person with diabetes—especially with SGLT2-inhibitor use, illness, fasting, or reduced intake—needs urgent medical assessment. DKA can occur without extreme hyperglycemia. Do not use chat, a home glucose value, or a low urine-ketone result to rule it out.
+
+### Practical off-ramps
+
+There is no wellness requirement to remain in ketosis, and no moral or metabolic failure in leaving it.
+
+- Move to a less restrictive lower-carbohydrate or Mediterranean-style pattern while keeping the protein, minimally processed foods, and meal defaults that helped.
+- Add flexible carbohydrate foods according to preference and tolerance: beans or lentils, fruit, oats or another whole grain, potatoes or squash, milk or yogurt, or culturally familiar staples. A gradual return can be easier on appetite and bowel comfort, but a special “reverse diet” is not required.
+- For hard or repeated training, restore carbohydrate before, during, or after the relevant sessions rather than protecting ketosis at the expense of performance and recovery.
+- For an adverse lipid response, reduce butter, coconut oil, fatty processed meats, cream, and excess cheese; favor olive oil, nuts, seeds, avocado, fish, and soluble-fiber foods. A large or persistent apoB/LDL-C rise still needs clinical review and may be a reason to stop keto entirely.
+- Anyone using glucose-lowering medication should coordinate carbohydrate changes with the prescribing team rather than tapering the diet or medication alone.
+
+## Murph handoffs
+
+These are internal ownership cues; do not expose route labels in a user-facing answer.
+
+- `body-composition`: intentional fat loss, muscle gain, recomposition, weight, waist, plateau, or calorie strategy.
+- `cardiometabolic-health`, `chronic-illness-support`, care navigation, and the user's clinician: diabetes, medications, blood pressure, LDL-C or apoB response, kidney or liver disease, pancreatitis, severe hypertriglyceridemia, or any therapeutic indication.
+- `gut-digestion`: persistent constipation, diarrhea, reflux, abdominal symptoms, or a fiber-change strategy.
+- `food-journal`: a brief, low-pressure record to learn what was eaten and how energy, bowel function, appetite, or training responded.
+- `micronutrients-supplements`: electrolyte powders, ketone products, MCT products, fiber supplements, multivitamins, labels, doses, and interactions.
+- `experiment-onboarding`: only after the user chooses one safe, bounded change with an outcome and stop conditions.
+- `behavior-followthrough`: recurring support after the plan is judged safe and worthwhile.
+- Eating-disorder-informed clinical and dietetic care: any material restriction, compensation, binge–restrict, or relapse risk.
+
+## Source register
+
+Evidence checked **2026-08-09**. Links are stable DOI records; source summaries state what each source can and cannot establish.
+
+- **American Diabetes Association Professional Practice Committee for Diabetes (2026), “Facilitating Positive Health Behaviors and Well-being to Improve Health Outcomes.”** [DOI 10.2337/dc26-S005](https://doi.org/10.2337/dc26-S005). Current diabetes guidance on short-term low-carbohydrate effects, medication supervision, SGLT2-inhibitor ketoacidosis risk, and groups for whom very-low-carbohydrate plans are not recommended.
+- **Multi-society consensus report (2024), “Hyperglycemic Crises in Adults With Diabetes.”** [DOI 10.2337/dci24-0032](https://doi.org/10.2337/dci24-0032). Defines DKA by ketonemia plus metabolic acidosis and covers euglycemic presentations; it does not evaluate wellness keto efficacy.
+- **Cochrane systematic review (2022), “Low-carbohydrate versus balanced-carbohydrate diets for reducing weight and cardiovascular risk.”** [DOI 10.1002/14651858.CD013334.pub2](https://doi.org/10.1002/14651858.CD013334.pub2). Large randomized evidence synthesis finding little or no clinically important comparative weight or risk-factor advantage through one to two years; low-carbohydrate definitions and interventions were heterogeneous.
+- **Umbrella review of randomized-trial meta-analyses (2023), “Effects of ketogenic diet on health outcomes.”** [DOI 10.1186/s12916-023-02874-y](https://doi.org/10.1186/s12916-023-02874-y). Calibrates certainty across conditions, identifies an LDL-C signal, and emphasizes the lack of long-term direct-outcome trials.
+- **Meta-analysis of randomized clinical trials (2024), “Impact of the ketogenic diet as a dietary approach on cardiovascular disease risk factors.”** [DOI 10.1016/j.ajcnut.2024.04.021](https://doi.org/10.1016/j.ajcnut.2024.04.021). Across short heterogeneous trials, keto lowered several risk factors but increased mean total and LDL cholesterol; it provides surrogate, not cardiovascular-event, evidence.
+- **Randomized trial in healthy adults (2024), “Ketogenic diet but not free-sugar restriction alters glucose tolerance, lipid metabolism, peripheral tissue phenotype, and gut microbiome.”** [DOI 10.1016/j.xcrm.2024.101667](https://doi.org/10.1016/j.xcrm.2024.101667). Shows that fasting glucose, glucose tolerance, apoB, and microbiome signals can move in different directions and change over 4–12 weeks; the small, short trial cannot establish long-term outcomes.
+- **National Lipid Association scientific statement (2019), “Review of current evidence and clinical recommendations on the effects of low-carbohydrate and very-low-carbohydrate (including ketogenic) diets.”** [DOI 10.1016/j.jacl.2019.08.003](https://doi.org/10.1016/j.jacl.2019.08.003). Cardiometabolic tradeoffs, variable LDL response, food-quality priorities, contraindication screening, and monitoring; predates newer trials but remains a major clinical consensus source.
+- **American Heart Association scientific statement (2023), “Popular Dietary Patterns: Alignment With American Heart Association 2021 Dietary Guidance.”** [DOI 10.1161/CIR.0000000000001146](https://doi.org/10.1161/CIR.0000000000001146). Evaluates diet-pattern quality and sustainability rather than proving clinical events; very-low-carbohydrate patterns align poorly when they restrict protective plant foods or favor saturated fat.
+- **Randomized crossover trial (2022), “Effect of a ketogenic diet versus Mediterranean diet on glycated hemoglobin in individuals with prediabetes and type 2 diabetes mellitus.”** [DOI 10.1093/ajcn/nqac154](https://doi.org/10.1093/ajcn/nqac154). Twelve-week-per-phase comparison finding no significant A1C advantage for keto, with higher LDL-C and lower intake of several nutrients; single-site and short-duration limits apply.
+- **International Society of Sports Nutrition position stand (2024), “Ketogenic diets.”** [DOI 10.1080/15502783.2024.2368167](https://doi.org/10.1080/15502783.2024.2368167). Reviews healthy exercising adults; performance is largely neutral or unfavorable versus higher-carbohydrate diets, maximal strength is often similar, and most trials are short.
+- **International adult ketogenic-therapy recommendations (2021), “International Recommendations for the Management of Adults Treated With Ketogenic Diet Therapies.”** [DOI 10.1212/CPJ.0000000000001007](https://doi.org/10.1212/CPJ.0000000000001007). Specialist consensus for therapeutic variants, contraindications, baseline assessment, adverse-effect management, and monitoring; not a wellness-diet protocol.
+- **Cochrane systematic review (2020), “Ketogenic diets for drug-resistant epilepsy.”** [DOI 10.1002/14651858.CD001903.pub5](https://doi.org/10.1002/14651858.CD001903.pub5). Supports seizure reduction in selected patients while showing limited adult evidence and meaningful tolerability/dropout concerns.
+- **Systematic adverse-event analysis (2026), “Adverse events and tolerability of ketogenic diets.”** [DOI 10.1186/s40795-026-01277-5](https://doi.org/10.1186/s40795-026-01277-5). Finds gastrointestinal events, especially constipation, commonly reported across heterogeneous ketogenic interventions; pooled frequencies should not be treated as a precise risk estimate for one ordinary adult implementation.