The What Should I Eat Generator is a free, in-browser decision aid that picks one of 24 original food ideas from a small, transparent editorial pool so you can stop staring at the fridge. It is built around the very specific problem of ordinary meal indecision: you have a kitchen, you have time, you have ingredients — you only need a starting direction. Rather than scrolling a recipe site, opening a delivery app, or scrolling back through your saved posts, you set two or three filters, draw one idea, and treat the result as a conversation starter. The pool is deliberately small so it can stay auditable, the random step uses browser Web Crypto so it stays fair, and no personal data leaves your device.
"What should I eat" is one of the most common food searches on the web, and the reason is not usually a missing recipe. Most home cooks already know how to make a sandwich, a stir-fry, or a bowl of yogurt. The problem is that decision fatigue kicks in before the cooking starts, especially when a household has to agree on one meal or when a single person has been staring at the same pantry for too long. A good answer to this kind of search is narrow in scope: it should produce a direction, not a lecture, and it should respect the fact that the user knows their own allergies, appliances, and budget better than any tool ever could. The What Should I Eat Generator is designed for exactly that low-stakes moment.
What the generator actually does
The generator works from a fixed inventory of 24 original ideas, organized as 4 meal categories crossed with 3 mood categories, with exactly two ideas in every meal-and-mood combination. The four meals are breakfast, lunch, dinner, and snack. The three moods are fresh, cozy, and hearty. The math is 4 × 3 × 2 = 24 unique records, and that is the whole pool — there is no recipe database, no cuisine index, and no nutrition calculator behind the widget.
When you click the generator, the browser first applies your active filters against the full pool of 24, then displays how many ideas remain eligible. From those eligible entries, it draws one at random using an unsigned 32-bit value from crypto.getRandomValues, with rejection sampling so no position can become biased by the modulo shortcut. If you click again while another eligible idea exists, the previous result is removed from the next draw, so the same entry does not appear twice in a row. The eligible count stays visible because the no-repeat rule shrinks only the next draw, not the overall pool.
How to pick a meal idea in three steps
- Choose a meal, food mood, and rough time limit, or leave any filter open for a larger idea pool. Start by picking the meal you are actually facing — breakfast, lunch, dinner, or snack — then choose a mood that fits how you want the food to feel. Fresh leans toward crisp produce and brighter combinations, cozy toward warm and familiar food, and hearty toward more substantial formats. The time limit is optional and rough: it caps the rough preparation estimate the idea carries. Leaving any filter open is fine and increases your eligible pool.
- Pick a food idea and review its title, short direction, time estimate, and eligible-pool count. Once the filters are set, the tool shows how many ideas still match. Drawing returns one title with a short description and a broad preparation-time label such as 10, 20, or 30 minutes. Read the title and the short direction as a starting point, and treat the time label as a planning estimate rather than a promise.
- Verify ingredients, allergies, dietary needs, food safety, and an appropriate recipe before copying or using the idea. The generator is a direction, not a complete ingredient list or cooking method. Before you commit, confirm that the suggestion fits the people eating it, find a recipe you trust, and check every ingredient and step for allergens, dietary needs, and safe handling.
How the filters and result fields work
The three filters are independent and combine by intersection — every active filter must still match for an idea to be eligible. Leaving all three open returns the full pool of 24. The table below summarizes what each filter actually controls, so you can predict the eligible count before you draw.
| Filter | What it does | What happens if you leave it open |
|---|---|---|
| Meal | Limits results to one of breakfast, lunch, dinner, or snack. | All four meal categories remain eligible. |
| Mood | Limits results to fresh, cozy, or hearty editorial categories. | All three moods remain eligible. |
| Time limit | Includes only ideas whose label is at or below the cap (15, 30, or 45 minutes). | The full collection of editorial time labels is eligible. |
The 24-entry pool is built as 4 meals × 3 moods × 2 ideas, so the math is 4 × 3 × 2 = 24 unique records. The time labels themselves are editorial estimates, not cooking times. The contract defines them as positive integers no greater than 60 minutes, used only for filtering, and assumes that common ingredients and basic kitchen equipment are already available. Actual preparation can be shorter or longer, and the label deliberately excludes shopping, thawing, marinating, unfamiliar techniques, cleanup, or interruptions.
From idea to actual meal: what to check next
The result the generator returns is a direction, not a finished plan. Treat the title and short description as a prompt, then do the real food decision yourself. Confirm that the meal and rough effort actually fit the moment — for example, a 30-minute dinner suggestion only helps if you have thirty minutes. Search for a recipe you trust that matches the idea, read the full ingredient list, and check every item against the allergies, intolerances, and dietary needs of everyone who will eat it. Verify safe storage and cooking guidance for any animal products, raw grains, or fermented items the recipe involves, and adjust portions for the number of people at the table. If a suggestion is impractical, draw again or widen one filter — opening up the mood, raising the time cap, or leaving the meal open all grow the eligible pool quickly.
Privacy is part of the verification story. The generator runs entirely in your browser. Filters and the current result live only in the open page, nothing is uploaded or saved by the widget, and clipboard access is requested only when you press the Copy idea button. The tool does not look at your browsing history, location, delivery apps, health information, restaurant availability, pantry contents, or previous meals. If you want a closer look at how that local-only behavior is built, the privacy walkthrough explains the implementation in more detail.
When the generator is the wrong tool
The generator intentionally has no dietary checkbox, and that is by choice rather than an oversight. A title like tomato soup, yogurt bowl, stir-fry, or quesadilla does not establish which ingredients, allergens, preparation surfaces, or substitutions will actually be used, because the same named dish can be suitable or unsuitable depending on the recipe. The tool therefore does not label ideas vegan, vegetarian, gluten-free, halal, kosher, low-sodium, diabetic-friendly, pregnancy-safe, or allergy-safe. For medically prescribed diets, serious allergies, infant feeding, pregnancy concerns, swallowing problems, or other health-related needs, rely on qualified guidance and verified ingredient information rather than a random general prompt. The right use case is ordinary low-stakes indecision: you want a broad idea, understand your own constraints, and will make the real food decision yourself.
The mood labels are practical interface categories, not health claims. Fresh points toward crisp produce and brighter combinations, cozy toward warm familiar food, and hearty toward more substantial formats. None of those words make nutritional, calorie, health, dietary, or emotional-wellness claims, and the meal labels describe when an idea might fit rather than enforce a rule — a breakfast wrap can be dinner, and a snack mix can be part of lunch. The time labels are similarly practical planning estimates used only by the filter. A time label is not a food-safety cooking time and does not tell you how long a particular cut of meat, egg, or grain must be held at temperature.
Common situations where it helps
- You are cooking for one and tired of rotating through the same three lunches.
- A household cannot agree on dinner and you want a neutral starting suggestion.
- You have about twenty minutes and want a direction that respects the cap.
- You want a snack idea that is more specific than "have some crackers."
- You are brainstorming meal prep options and need a short list of moods and formats to react to.
The generator is at its best when you treat it as a small, honest randomizer. It will not rank local restaurants, provide a recipe, order food, calculate nutrition, or compare prices, and it will not pretend that a dish name tells you what is actually in the dish. What it will do is give you one of 24 directions, drawn fairly from the pool that matches your filters, with the immediate repeat avoided so two clicks in a row feel like two different draws. For the everyday question of what should I eat, that is usually enough to break the stall and let the rest of the decision — the recipe, the ingredients, the timing — happen the way it normally does.