The Kitchen Table

This is the prompt we use in our answer detection example. The agent looks at open questions in your community and decides what to do with each one: mark a reply as the answer, nudge the asker to confirm, or just log it and leave the thread alone.The model only recommends. Your script checks its output and decides whether to act. If you haven’t read it yet, start with Before you automate: what we’ve learned building AI agents for communities. It explains why the agent is built this way.How to use itReplace everything in [brackets] with details from your own community. Keep the output format exactly as it is. Your script depends on it. Test it in Claude first. Paste the prompt, then paste one thread at a time as JSON, and compare the result with what you would have decided. Run it in dry-run mode on 30 to 50 real threads before you let it write anything.The promptYou review open questions in the [COMMUNITY NAME] community and decide whether one of the replies already answers the questio

I had a bad tomato summer three years ago. The market tomatoes were pale and tasteless by mid-August and every salsa I made that season was flat in a way I could not fix with technique. Out of frustration I put six roma tomato plants in containers on a south-facing balcony the following spring. By July I had more tomatoes than I needed and every single one tasted more of tomato than anything I had bought in years. Now: twelve plants, a tomatillo section, two types of chile, and a growing conflict with my balcony space. The salsas I make in season are the reason I keep the garden. The rest of the year I use good canned tomatoes and remember what is possible.

I spent a year systematically testing different dairy products in Mexican cooking contexts: crema, sour cream, crème fraîche, queso fresco, queso Oaxaqueño, Chihuahua cheese, cotija, requesón, and several commercial substitutes. Key findings: Mexican crema is genuinely different from sour cream — higher fat, more pourable, less sour. Not interchangeable for applications where you want a sauce-like consistency. Queso fresco and cotija are different cheeses used for different purposes: fresh vs cured, mild vs salty, crumble texture vs finer crumble. Oaxacan quesillo is the best melting cheese in Mexican cooking and most home cooks outside Mexico do not have access to it. Low-moisture mozzarella is the closest functional substitute for melting.

After cooking indoors exclusively for years, I built an outdoor station this summer and the change to my cooking is significant. Outdoor photo documentation. Setup: a large commercial propane burner on a steel-frame table, a cast iron comal mounted on the second ring, a charcoal chimney for direct flame work, and a stainless prep table alongside. What changed: I can cook at higher heat than any indoor setup because ventilation is unlimited. The comal temperature I can reach outdoors is genuinely different from indoors — hotter and more even. Charring tomatillos, chiles, and corn directly over flame without setting off the smoke alarm. For anyone who has outdoor space: even a single outdoor burner on a table changes what is possible.

I have cilantro aversion (it tastes strongly of soap to me — genetic, nothing to be done). I spent a year systematically testing substitutions in dishes that typically use cilantro and want to share the results. For tacos and burritos: flat-leaf parsley plus extra lime works best. The parsley provides green herbal freshness; the lime provides the bright finish note. Not identical but the closest functional result. For salsa verde: epazote (where available) or Mexican oregano. Both read as 'Mexican herb' without cilantro's specific soapy quality. For guacamole: just skip it entirely. Cilantro-free guacamole with good lime, salt, and fresh onion is excellent on its own terms. For pico de gallo: flat-leaf parsley is the substitute. Add it at the last second — it wilts faster than cilantro.

I see this come up and I want to put down my own position clearly. The burrito in its current most common form — large flour tortilla, rice, beans, protein, multiple toppings — is a border and northern Mexican food that evolved through the 20th century in Sonora and Chihuahua and in the US border states. It is not ancient, it is not pan-Mexican, and it is not identical to what you find in most parts of Mexico. But 'not ancient' and 'not pan-Mexican' do not mean 'not Mexican.' Food evolves. The burrito has Mexican DNA — the tortilla, the beans, the chile-based proteins — and developed in places where Mexican food culture was central. Arguing about authenticity in food is usually less interesting than understanding the history. The burrito's history is specific and interesting. The authenticity debate is not.

Not all cookbooks are technique books. Here are three that changed how I think rather than just what I cook. Diana Kennedy's The Art of Mexican Cooking: the most rigorous documentation of regional Mexican food in English. Not prescriptive — investigative. Reading it changed my understanding of what regional means. Nuevolution by Enrique Olvera: a chef's cookbook that is also a philosophy of Mexican food. The mole pages changed how I approach complexity. Salt, Fat, Acid, Heat (Samin Nosrat): not Mexican but the fat and acid chapters rewired how I think about every dish, including Mexican food. The acid chapter is the most useful 30 pages I have read about cooking. Books that teach you to think are worth more than books that teach you to follow.

I have been pairing mezcal with home-cooked Mexican food for a year and want to share what I have found works. Carnitas: a smoky mezcal is the obvious choice but I prefer a joven mezcal with light smoke — the fat in the carnitas provides enough depth that the mezcal does not need to add more. A heavily smoked mezcal competes. Chile verde: anything citrus-forward. The green chile and tomatillo base is already complex and acidic; a bright mezcal echoes rather than fights it. Mole negro: the most complex food in this category and probably the one place I reach for an añejo. The aged spirit can stand up to the depth of a proper mole negro without being lost in it.

I was making enchiladas and at the filling stage decided the filling was too good to be contained — pulled pork in a red guajillo sauce with a small amount of refried black beans and queso fresco. I almost rolled it into a burrito instead. I made the enchiladas anyway. Photo: the baking dish, 8 enchiladas in a line, dark red sauce over the top, cotija over that. But I also made one burrito with the same filling. Side-by-side comparison was interesting. The enchilada won — the sauce pooled around the rolled tortilla and saturated the outside as it baked. The burrito was correct and clean. The enchilada was more luxurious. Different dishes for different occasions.

Documentation of a first attempt at fresh corn tortillas, with the honest mistakes included. Attempt 1: too dry — cracked on the press, cracked on the comal. Looked like a failed attempt at pottery. Attempt 2: water added, still slightly too dry. Some cracking at the edges. Attempt 3: got the texture to soft clay. No cracking on the press. Placed on a too-cool comal — steamed rather than toasted. Attempt 4: hot comal. First tortilla that looked like a tortilla. The photo of attempt 4 is not impressive by any objective measure. But the smell of fresh masa on a hot comal is the reason people do this. Once you smell it, you understand what masa harina is trying to replicate.

For anyone who has been cooking from canned beans and wants to understand what they are missing, here is the full process with photos. Stage 1 (raw dried pinto beans): small, firm, fully dry. Colour: light tan. Stage 2 (soaked overnight): swollen to double size, skin slightly looser. Stage 3 (after one hour cooking): beginning to soften, broth still pale. Stage 4 (after two hours): the broth has darkened significantly. The first beans are beginning to split. Stage 5 (done): full softness, broth is dark and deeply flavoured. The smell at this stage is the best smell in my kitchen. The broth in stage 5 is the ingredient you are throwing away when you use canned beans.

I am a convert and I want to document the conversion. Forty years of cooking on non-stick pans. Good ones, replaced when they scratched, always adequate. Then I inherited a cast iron skillet from a relative and started using it out of curiosity. Three months later: the cast iron is the only pan I use for anything that involves direct browning. The carnitas crisp is better. The tortilla char is better. The potato crust is better. The egg situation is complicated (I am still learning the seasoning) but improving. Photos: carnitas in cast iron at the crisping stage (you can see the direct Maillard crust), vs my memory of the non-stick equivalent which had more of a sauté colour than a proper crust.

The tomato plants this year were more productive than I expected. Eleven kilos of ripe roma tomatoes, all ready within the same two-week window. Photo: the full harvest on the kitchen counter. It is a problem I am very happy to have. Plan so far: 3 kilos into fire-roasted red salsa for freezing. 2 kilos into a slow-cooked tomato passata. 2 kilos into a chile colorado braise base. 2 kilos into slow-roasted whole tomatoes (stored in olive oil). 2 kilos still unallocated. Suggestions welcome for the last 2 kilos. Bonus points if it involves dried chiles.

The tomatillos are producing well this year and I set up a backyard station for making salsa verde in volume. Photo documentation. Setup: outdoor gas burner, large comal, blender, sterilised jars. The workflow: tomatillos from the vine, straight to the comal for charring, into the blender with toasted serrano and charred onion, adjusted with salt and lime, into hot sterilised jars. I made 11 jars this weekend. They will last until spring. The photo of the row of jars with the dark green-grey colour of properly charred tomatillo salsa is the most satisfying thing I have made this year.

Side-by-side photo because I think the visual argument is more convincing than anything I can write. Left: supermarket flour tortilla, the most popular brand locally. Right: fresh flour tortilla made with good lard. The differences visible in the photo: the lard tortilla has slight layering at the edge, a visible variation in surface texture from the fat distribution, and a warmer colour from the fat content. The shop one is uniformly pale and smooth. In the fold test (which I also photographed): the lard tortilla folded without cracking; the shop one developed a small crack at the fold point. Bake a lard tortilla. Once.

Week eleven of my Sunday batch cooking. The system has stabilised and I wanted to document what it looks like now versus what it looked like in week one. Week one: 3 hours of chaotic cooking, two blowouts in the assembly, rice overcooked, forgot to keep the bean broth. Week eleven: 2 hours, everything parallel, zero blowouts, rice correct every time, bean broth reserved and refrigerated in a labelled jar. Photos: the current setup (labelled containers, resting jig for the finished burritos, the broth jar in the fridge). The difference between week one and week eleven is not skill — it is system. The same technique repeated weekly becomes automatic.

Full disclosure: the habanero hot sauce in this burrito was made with a 3:1 habanero to tomato ratio, which was ambitious. Photo: the cross-section. The sauce is visibly contained in a stripe — I used the wet ingredient discipline from this community's assembly threads and it worked. The burrito held. The seal was fine. The eating experience was extremely intense but controlled. The habanero sauce at that ratio has a fruitiness behind the heat that is genuinely beautiful if you can get past the heat level. Would I make it at that ratio again? Probably not for a weeknight. Yes for a deliberate occasion.

I documented the full carnitas process from start to finish with photos at the two-hour, four-hour, six-hour, and eight-hour marks. This was a 2.5kg shoulder cooked in lard at a low oven temperature. Two hours: the meat is beginning to colour but still intact. The fat is clarified. Four hours: the shoulder has slumped and the visible connective tissue is starting to melt. Six hours: you can see the muscle beginning to separate. The fat is darker. Eight hours: the meat pulls apart with a finger. The fat has reduced to about 60% of the original volume. Final crisp in a cast iron took 6 minutes. The result was the best carnitas I have made — yielding, slightly gelatinous at the edge pieces, consistent throughout.

I have been wanting to write this post for a while. Here is a detailed breakdown of my best burrito roll so far, with photos at each stage. Fill weight: 248g. Diameter at widest: 7.2cm. Length: 21cm. Seal method: three-point with a 45-second skillet hold. Fill distribution: rice as base layer across the full width, carnitas centred in a column, beans flanking the protein, avocado as the top layer placed last. Salsa in a controlled stripe across the centre of the protein. Cross-section after cutting: symmetric, clean edges, no leaking, avocado visible as a distinct layer at the top. This is what the numbers look like when everything is right. I have the photo if anyone wants the reference.

Documenting the difference. Before: burritos assembled by feel, no sense of why some held and others did not, fill weight guessed, sealing method informal, tortilla temperature ignored. After eight months: weigh everything, three-point seal, resting step, skillet temperature checked, bean broth kept and used. Blowout rate dropped from roughly 1 in 3 to almost zero. Photos: a burrito from my first month (you can see the structural issues), and one from last week (clean seal, correct geometry, cross-section shows even distribution). This community is a genuine resource. I would not have made the connection between fill weight and blowout probability on my own.

Six weeks of working on this. Six weeks of burritos that looked right until they hit the skillet and unrolled. I have been posting questions here and trying every suggestion. Today: three-point seal, skillet at proper temperature (finally invested in a cast iron thermometer), seam-side down for 45 seconds before moving. The seal held. Photo: cross section after cutting, absolutely clean. No leaking. Proper circular geometry. Thank you to RollAndrei and hot_skillet_only specifically — the skillet temperature explanation was the missing piece.

Batch-cooked Sunday. Four fillings in parallel: carnitas (the main event), black beans with epazote, Mexican red rice, and charred corn. Twelve burritos assembled, six for the week and six for the freezer. The carnitas took four hours but were mostly unattended. The beans were two hours, also unattended. The rice was 35 minutes. The corn took 10 minutes. Photos at each stage: the carnitas resting in the fat before crisping, the beans after two hours (the broth was dark and thick), the assembled burritos lined up before wrapping. This is the most satisfying Sunday cooking I do. The week is easier from Monday.

We have a lot of new members this month and I want to give everyone a space to say hello. Three questions: 1. Where are you cooking from? 2. What are you most trying to improve right now? 3. What has been your best cooking discovery in the last year? I will start: I am based in Austin. Trying to improve my masa hydration consistency. Best discovery: that resting a burrito for five minutes before cutting it dramatically improves the cross-section.

I lurked here for six months before I made my first post. I wanted to say something about that experience because I think it is more common than the member count suggests. Reading the threads on technique without posting felt valuable in a way I did not expect — absorbing how people thought about cooking, what questions they asked, what kinds of answers were considered useful. By the time I posted for the first time I had a framework that would have taken years to build from cookbooks alone. If you are reading this and have not posted yet: you are welcome here. The tone is good.