# Dried pasta: research notes

Written 2026-10-19 for the category-expertise run. Scope: dried durum wheat pasta plus whole durum, egg, legume and gluten-free grain alternatives as mechanism families. Fresh pasta and Asian noodles are out. Quote labels: **verbatim** (returned exactly by a fetch or extracted from a PDF on disk), **fetched excerpt** (a fetch's paraphrase of the page), **search excerpt** (summary returned by web search, page itself not fetched).

## Seed

Seed run: `seed.py --category dried-pasta` returned the olive-oil harvest-date example and the persona "A nurse, who thinks in who this is actually for and what goes wrong in use."

1. The dried-pasta equivalent of harvest date is the drying profile: temperature and hours. Experts and premium producers rank it alongside the bronze die, it changes nutrition (lysine loss, furosine) and flavour, almost no listing carries a number, and backfilling it means reading producer pages. Kept as the facet `drying_profile`, marked researched, backfill the top of the catalog only.
2. What the nurse notices first: who the box is for and what goes wrong at the stove. Who it is for became the glycemic band (legume pasta for people managing blood sugar, and the fact that overcooking any pasta raises its glycaemic index) and the allergen cross-contact note on legume and gluten-free pasta. What goes wrong in use became the doneness-window trade-off for gluten-free pasta (hard to mush in a minute) and the foaming of legume pasta. Dropped: a "forgiveness window" facet (not on listings, nobody measures it), sodium (dried pasta has none, so it does not split anything), and "who is this for" as a cold-start split (base family already asks it).

## Sources

### Standards body: Consorzio di Tutela della Pasta di Gragnano IGP, "Il prodotto"
URL: https://www.pastadigragnanoigp.it/il-prodotto/
- **verbatim**: "utensili esclusivamente in bronzo, che permettono di conferire alla pasta una superficie rugosa" → IGP pasta must be extruded through bronze dies. Confidence high.
- **verbatim**: "L'essiccazione varia a seconda dei formati e comunque avviene ad una temperatura compresa tra 40° e 85° C per un periodo compreso tra le 4 e le 60 ore" → IGP drying window 40 to 85 °C, 4 to 60 hours. Confidence high.
- **verbatim**: "acqua della falda acquifera di Gragnano, in una percentuale non superiore al 30%" → water share of the dough ≤30%.

### Standards body: Disciplinare.it, "Pasta di Gragnano IGP, modifica del disciplinare 2020"
URL: https://www.disciplinare.it/pasta-di-gragnano-igp-modifica-del-disciplinare-di-produzione-2020.html
- **fetched excerpt (quotes returned)**: "proteine minimo: 13% sulla sostanza secca", "ceneri max.: 0.86% sulla sostanza secca", "non superiore al 12,5% sul prodotto finito" (moisture), "tra 40 e 85°C per un periodo compreso tra le quattro e le sessanta ore" → IGP minimum protein 13% dry basis, moisture ≤12.5%, drying 40–85 °C for 4–60 h. Confidence high.
- Disagreement: a 2013 trade article (foodweb.it, search excerpt) gives 40–80 °C and 6–60 h, which looks like the pre-2020 text. We use the consortium's current wording.

### Regulator: DPR 9 febbraio 2001 n. 187 (Italian decree on flours and pasta), MASAF PDF, extracted with pdftotext
URL: https://www.masaf.gov.it/flex/files/2/f/6/D.34ca305e98ded6c87bfc/DPR_187_2001.pdf
- **verbatim** art. 6 table: "Pasta di semola di grano duro 12,50 [umidità massima] 0,90 [ceneri max] 10,50 [proteine min, azoto x 5,70] 4 [acidità]" → legal floor for durum semolina pasta: moisture ≤12.5%, protein ≥10.5% on dry matter. Confidence high.
- **verbatim** art. 6.4: "e' vietata la fabbricazione di pasta secca preparata con sfarinati di grano tenero" and 6.5: "tollerata la presenza di farine di grano tenero in misura non superiore al 3 per cento" → in Italy dried pasta is durum only, soft wheat tolerated to 3%.
- **verbatim** art. 6.2: "pasta di semola integrale di grano duro" ... table row "12,50 1,40 1,80 11,50 6" → whole durum pasta: protein ≥11.5% dry basis.
- **verbatim** art. 8.1: "almeno quattro uova intere di gallina, prive di guscio, per un peso complessivo non inferiore a duecento grammi di uovo per ogni chilogrammo di semola" and 8.2 "proteine ... non inferiore a 12,50" → egg pasta needs ≥200 g egg per kg semolina and ≥12.5% protein dry basis.
- Note: the law's protein is on dry matter. At 12.5% moisture, 10.5% dry basis ≈ 9.2 g per 100 g as sold. Listings show as-sold values.

### Regulator: FoodTimes, "Origin of wheat in pasta, Italian decree notified to Brussels"
URL: https://www.foodtimes.eu/consumers-and-health/origin-of-wheat-in-pasta-italian-decree-notified-to-brussels/
- **fetched excerpt (quotes returned)**: labels must carry "'Wheat growing country'" and "'Country of milling'"; when several countries are involved "'EU', 'non-EU,' 'EU and non-EU'"; if ≥50% from one country, that country "followed by the words 'and other countries …'". Confidence high.

### Regulator: Food Compliance International, "Italy introduces mandatory origin labelling for rice and durum wheat pasta"
URL: https://foodcomplianceinternational.com/industry-insight/news/1002-italy-introduces-mandatory-origin-labelling-for-rice-and-durum-wheat-pasta
- **fetched excerpt**: labels must include "both the country where the wheat was cultivated and the country of milling"; compliance by February 2018; "introduced as an experiment and will temporarily apply until end of 2020". Confidence high.

### Regulator: Laboratorio Chimico Camera di Commercio Torino, "Decreti origine nazionali: proroga al 31 dicembre 2024"
URL: https://www.lab-to.camcom.it/news/2024/1/2/decreti-origine-nazionali-proroga-al-31-dicembre-2/
- **verbatim**: "Prorogato fino al 31 dicembre 2024" for "l'origine del grano per la pasta di semola di grano duro". Confidence high.
- **search excerpt** (portale-etichettatura.lab-to.camcom.it, 2025-02-14): DM 23 December 2024, published in Gazzetta Ufficiale 12 February 2025, extends the same obligation to 31 December 2025. Confidence medium (not fetched).

### Academic: Lucisano et al. 2008, "Influence of die material on pasta characteristics", Food Research International 41(6) 646–652
URL: https://www.sciencedirect.com/science/article/abs/pii/S0963996908000690
- Page returned 403. **search excerpt** of the abstract: "Die inserts lined with Teflon give a smooth and bright-yellow appearance to the product, whereas bronze inserts give a rough surface"; bronze "higher porosity, which caused a 20–30% reduction in the spaghetti breaking strength"; rough surface "made the deposition of eggs easier" for insects. Confidence medium on its own, high once the 2022 review below repeats the same numbers.

### Academic: Bresciani, Pagani, Marti 2022, "Pasta-Making Process: A Narrative Review", Foods 11(3):256
URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC8834582/
- **verbatim**: "Teflon gives the product a smooth and bright yellow appearance, while bronze inserts produce a rough surface."
- **verbatim**: "Bronze-extruded spaghetti is more porous and therefore more fragile (breaking strength decreases by 20–30%) than Teflon-extruded products." → second source for the 20–30% figure. Confidence high.
- **verbatim**: "High temperature drying cycles (>65 °C) are effective in improving the sensory characteristics of pasta, especially in the case of pasta made with semolina low in protein." → high-temperature drying compensates for weak semolina. Confidence high.
- **verbatim**: "Pasta dried at a low temperature had low amounts of furosine, which is the most widely used marker for assessing the extent of the Maillard reaction."
- **verbatim**: "Durum wheat semolina characterized by high protein content and strong gluten—able to withstand the physical stress occurring during extrusion, drying, and cooking—is the ideal raw material for high-quality pasta."
- **verbatim**: "Dry pasta is also characterized by long shelf life, up to three years, thanks to its low humidity (generally lower than 12.5%)."
- Mentions "cooking loss values lower than 3 g/100 g pasta" without the context we needed; not used as a threshold.

### Academic: Giannetti et al. 2013, "Furosine as a pasta quality marker", Journal of Food Science
URL: https://ift.onlinelibrary.wiley.com/doi/10.1111/1750-3841.12163
- Wiley and Semantic Scholar both refused the abstract. **search excerpt**: furosine "ranging from 107 to 506 mg/100 g of protein"; very-high-temperature drying "345 to 506", low temperature "107 to 186". Confidence medium. Used only to size the furosine gap; direction is confirmed by the review above and by the De Cecco paper.

### Academic: PMC9689111, "Influence of Some Spaghetti Processing Variables on Technological Attributes and the In Vitro Digestion of Starch" (2022)
URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC9689111/
- **verbatim**: "reducing cooking from fully cooked to al dente and using pasta of very high protein content (17%), reduced starch digestion extent." and "reducing the cooking time to more like traditional Italian al dente texture rather than fully cooked or sometimes overcooked pasta could be a strategy to lower pasta GI". Confidence high.

### Academic: PMC7911001, "The Glycaemic and Insulinaemic Response of Pasta in Chinese and Indians Compared to Asian Carbohydrate Staples" (2021)
URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC7911001/
- **verbatim**: "semolina spaghetti and wholegrain spaghetti can be classified as low GI products, with a GI of 53 and 54, respectively" and "The compact protein structure entraps the starch granules leading to reduced or delayed access for digestive enzymes". Confidence high for the mechanism; the absolute GI disagrees with the primer below.

### Academic: PMC6722931, "Evaluation of Cooking Quality ... of Pasta Added with Oat Bran and Apple Flour" (2019)
URL: https://pmc.ncbi.nlm.nih.gov/articles/PMC6722931/
- **verbatim**: "The cooking loss (CL), expressed as grams of matter loss/100 g of raw pasta, was evaluated by determining the amount of solids lost into the cooking water" and "A cooking loss of 12% or less was considered acceptable and indicative of good quality pasta". Confidence high for the definition, medium for the threshold (one paper citing another). **search excerpt** from a second paper (Cooking Properties and Heat Damage of Dried Pasta, ResearchGate): typical values "6.40-6.50 g/100 g of raw spaghetti".

### Specialist review: America's Test Kitchen, "Why some pasta absorbs more or less sauce"
URL: https://www.americastestkitchen.com/how_tos/10415-why-some-pasta-absorbs-more-or-less-sauce
- **verbatim**: "Pasta made the traditional way is extruded through bronze plates, or dies, that give it a more textured surface. The cheaper, faster, modern way to extrude pasta is through slick Teflon dies that create a sleek surface."
- **verbatim**: "noodles extruded through bronze dies ... were far better at gripping sauce than the smooth noodles pushed through Teflon plates, which allowed more sauce to drip off."
- **verbatim**: "Pasta extruded through bronze dies will look coarser on the surface, while noodles extruded through Teflon dies look perfectly smooth."
- **fetched excerpt**: traditionally made pasta needs about 30 percent more cooking water for the same sauce consistency. Confidence high.

### Specialist review: America's Test Kitchen, "An ode to Italian pasta"
URL: https://www.americastestkitchen.com/articles/3349-an-ode-to-italian-pasta
- **verbatim**: "dried at a relatively low temperature (about 100 degrees) for as long as three days, compared with the much faster method of drying pasta at 180 degrees for 7 to 10 hours." (Fahrenheit: about 38 °C vs 82 °C.)
- **verbatim**: "research has shown that there is not much textural difference between the two when cooked." → the counter-view on slow drying. Confidence high that ATK says this.
- **verbatim**: "Most manufacturers use dies coated with nonstick materials such as Teflon" and "While a few larger companies such as De Cecco use bronze dies, most bronze die–cut pasta is made by small companies."

### Specialist review: America's Test Kitchen, "What is bronze-cut pasta"
URL: https://www.americastestkitchen.com/articles/6898-what-is-bronze-cut-pasta
- **verbatim**: "When dough is pressed through the bare metal, it winds up with a coarser exterior and better grip."; "bronze-cut brands will often say that or 'bronze-die' on the label."; under magnification "the bronze-cut De Cecco had a distinctly scaly texture".
- **fetched excerpt**: named bronze-die brands De Cecco, Rustichella d'Abruzzo; Teflon-die brands Ronzoni, Barilla, Divella. Confidence high.

### Specialist review: Cook's Illustrated spaghetti taste test
URL: https://www.americastestkitchen.com/taste_tests/553-spaghetti
- The page now redirects to a recipe. **search excerpt** of the test text: De Cecco Spaghetti No. 12 won, "the lab confirmed these strands as the strongest of all the samples, with the lowest percentage of 'cook loss'". Confidence medium (not read on the page).

### Spec sheet: De Cecco, "Slow drying for high-quality pasta" (2019 PDF, extracted with pdftotext)
URL: https://www.dececco.com/wp-content/uploads/2019/07/LENTA-ESSICCAZIONE_EN_2019.pdf
- **verbatim**: "between 1980 and 1990, pasta could be dried in just a few hours by using temperatures of 80-90°C."
- **verbatim**: "assorted pasta products exhibit heat damage (identified with the analysis of furosine ...) of higher than 300mg per 100g of protein." and "wholegrain pasta samples analysed ... values of furosine greater than 500-600mg per 100g of protein".
- **verbatim**: "De Cecco's durum wheat semolina pasta has average levels of furosine between 50 and 200 mg per 100g of protein".
- **verbatim**: "its texture and firmness in cooking depend heavily on the grain used, which must be high quality" (on low-temperature drying).
- **verbatim**: "using the fresh semolina coarse-grain within hours of grinding. The dough is made with lots of cold water and long kneading times, followed by bronze die-drawing and slow drying at low temperatures."
- Producer document; confidence medium for its own numbers, high for the direction, which the academic review confirms.

### Spec sheet: De Cecco Spaghetti n° 12 product page (dececco.com it_it)
URL: https://www.dececco.com/it_it/product/spaghetti-n-12/
- Page returned 403. **search excerpt**: cooking time 12 minutes, al dente 10; "trafilata al bronzo"; slow drying "up to 18 hours"; 14% protein. Confidence medium.

### Spec sheet: Pasta Garofalo N° 9 Spaghetti (UK product page)
URL: https://www.pasta-garofalo.com/uk/product/n-09-spaghetti/
- **fetched excerpt with quotes**: ingredients "Durum wheat semolina."; per 100 g energy 351 kcal, protein 14 g, fibre 3.0 g, salt <0.01 g; cooking time 10 minutes; "Bronze drawn", "Rough to the touch thanks to the processing with bronze drawing"; Pasta di Gragnano PGI; "Country of cultivation: EU and non-EU countries", "Country of milling: Italy"; "May contain soy and mustard". Confidence high. Shows a premium producer publishes protein per 100 g, die, PGI and wheat origin on the page.

### Spec sheet: Casa Rustichella, Spaghetti Lunghi 500 g
URL: https://casarustichella.com/en/prodotto/spaghetti-lunghi-500g/
- **verbatim**: ingredients "Durum wheat semolina, water."; protein "13,5 g" per 100 g; "bronze drawn and dried at low temperature" (no hours on the page). Confidence high.

### Primer: DeLaurenti blog, "Rustichella d'Abruzzo: the Epicurean's Pasta"
URL: https://blog.delaurenti.com/2022/01/07/rustichella-dabruzzo-the-epicureans-pasta/
- **verbatim**: "Drying the pasta at low temperatures, a process that takes 40–60 hours, preserves the flavor of the wheat and retains its nutritional value." Confidence medium (retailer blog).

### Primer: Made in Tuscany, "Martelli traditional Italian pasta since 1926"
URL: https://www.madeintuscany.it/site/dt_portfolio/martelli-traditional-italian-pasta/?lang=en
- **verbatim**: "dried at a 'low traditional temperature' (33-36°C) for about 50 hours (depending on the weather)"; "slowly kneading the best durum-wheats with cold water. The bronze-drawing gives the pasta a rough texture." Confidence medium.

### Spec sheet (producer blog): Agrimò, "Furosine and pasta"
URL: https://www.agrimo.it/en/blogs/notizie/furosina-agrimo-pasta
- **verbatim**: "Pasta dried too quickly, above 70–90°C, loses up to 30% of its lysine."; industrial drying "80–110°C"; Agrimò dries "up to 46 hours below 40°C (104°F)". Producer source, confidence medium; lysine direction matches the academic review.

### Enthusiast community: Noodledex, "De Cecco vs Barilla: the bronze die difference"
URL: https://www.noodledex.com/guides/de-cecco-vs-barilla
- **verbatim**: "Barilla extrudes most SKUs through Teflon and dries fast at high temperatures."; "De Cecco dries at low temperature and lists a certified drying time for each shape (orecchiette No. 91 packs say 9 hours)."; "De Cecco usually runs 1-2 minutes longer than Barilla for the same shape"; "The starch on the surface is what makes the sauce cling — rinsing washes it away." Confidence medium.

### Enthusiast / primer: Vincenzo's Plate, "Pasta brands ranked"
URL: https://www.vincenzosplate.com/ranking-italian-pasta-brands/
- **verbatim**: "If it says 11% or 12%, the pasta will cook, but it won't have much resilience." and recommends "13% or even 14% protein"; bronze pasta looks "slightly pale, almost dusty and a little rough"; artisan pasta "dried slowly at lower temperatures, sometimes for 20, 30, even 40 hours"; industrial "dried very quickly at high temperatures, sometimes even in just a few hours". Confidence medium.
- **search excerpt** (same page, numbers not confirmed on fetch): Rummo "14.5%", Garofalo "14%", De Cecco "around 14%", Barilla "around 12.8%". Barilla's own EU label below says 13 g.

### Primer: ChefShop, "How to choose better dried pasta"
URL: https://chefshop.com/blogs/field-notes/how-to-choose-better-dried-pasta
- **verbatim**: "Look for a rough, matte, slightly powdery-looking surface rather than a shiny yellow one."; "Bronze creates friction as the dough passes through, giving the pasta a rougher, more textured surface."; "Look for pasta made with durum wheat semolina." Confidence medium.

### Primer (Italian): Cibo360, "Scegliere la pasta di qualità: trafila e metodo di essiccazione"
URL: https://www.cibo360.it/qualita/confronti/scegliere_pasta_qualita.htm
- **fetched excerpt with quotes**: bronze pasta "si presenta opaca" and "molto ruvida"; Teflon pasta "è lucida, di un giallo più chiaro e brillante"; on nutrition between the two: "No, non cambia nulla: i valori nutrizionali della pasta rimangono sempre quelli". Confidence medium. Supports: the die changes surface, not the nutrition panel.

### Other: Modern Retail, "How Barilla is trying to go premium with its new pasta line"
URL: https://www.modernretail.co/marketing/how-barilla-is-trying-to-go-premium-with-its-new-pasta-line/
- **verbatim** (re-fetched after the blind check): "Al Bronzo SKUs are sold at the suggested retail price of $2.99 per box"; "Barilla's blue boxes that sell for around $1.99"; "Al Bronzo shapes are created using Barilla's new micro-engraved bronze dyes". The page does not mention protein or cooking time. Confidence high for price and die.
- Dropped: the ">14% protein" and "two more minutes" claims came from other search results (Daily Meal, Eat This); Eat This, fetched, gives price and the die claim only, and barilla.com refused the fetch. Not in the document.

### Other: WonderHowTo, "Best grocery store pasta: a chef-approved buying guide" (cites an Agrofood.it review)
URL: https://food-hacks.wonderhowto.com/news/best-grocery-store-pasta-a-chef-approved-buying-guide/
- **search excerpt**: "a research review from Agrofood.it found no peer-reviewed study that directly measured whether that roughness makes sauce stick better"; "the sauce-clinging payoff is chef experience, not a controlled kitchen test." Confidence medium. Kept as the dissenting side on sauce cling.

### Other: CBC, "Italy's farmers 'demonizing' Canadian wheat" and The Western Producer, "Count on Canadian durum"
URLs: https://www.cbc.ca/news/world/italy-canada-durum-wheat-dispute-1.4191615 ; https://www.producer.com/farm-family/count-on-canadian-durum-to-produce-delicious-pasta/
- **search excerpt**: Canadian Western Amber Durum "recognized for its high protein content, gluten strength"; Italian durum "with its lower protein content, can't be made into pasta without the addition of higher-protein wheat from elsewhere"; Italian objections centre on pre-harvest glyphosate desiccation. Confidence medium; the two sides are national lobbies, so both quotes are partisan.

### Primer: GlycemicSnap, "Pasta and blood sugar"
URL: https://glycemicsnap.com/blog/pasta-and-blood-sugar
- **verbatim**: "Standard durum wheat spaghetti, cooked al dente, has a GI of approximately 42-48."; "Overcooked, mushy pasta can jump to a GI of 55-65"; "Whole wheat pasta (GI ~42) is slightly lower than white pasta (GI ~45-48)"; legume pasta "GI ~25-35". Cites a 2018 BMJ Open meta-analysis and a BBC experiment, not a database. Confidence low-medium; absolute numbers differ from PMC7911001 (53/54) but both say low-to-medium and both say overcooking raises it.

### Spec sheet: Banza, Classic Chickpea Spaghetti
URL: https://www.eatbanza.com/products/spaghetti
- **verbatim**: ingredients "Chickpeas, Pea Starch, Tapioca, Xanthan Gum"; claims "50% More protein", "3X Fiber", "25% Fewer Net Carbs" versus average pasta; cooks "about 8-10 minutes". Nutrition panel is an image. Confidence high for what the page states.

### Other: MyFoodDiary / SnapCalorie entries for Banza spaghetti (search excerpt)
URL: https://www.myfooddiary.com/foods/7241784/banza-spaghetti-chickpea-pasta
- **search excerpt**: per 100 g protein 19.3 g, fiber 8.8 g, 333 kcal. Confidence medium. Cross-check: Gluten Safety ranking (fetched) says "A single serving can contain 14 grams of protein and 8 grams of fiber" for Banza (per 2 oz).

### Other: Barilla Protein+ spaghetti (recipal / H-E-B / Stop & Shop, search excerpt)
URL: https://www.recipal.com/ingredients/525045-nutrition-facts-calories-protein-carbs-fat-barilla-protein-plus-spaghetti
- **search excerpt**: per 56 g serving 10 g protein, 5 g fiber, 190 kcal; ingredients semolina, durum flour, lentil flour, pea protein, chickpea flour, barley, spelt. So a wheat-plus-legume blend at about 18 g protein per 100 g. Confidence medium.

### Retail listing: Open Food Facts, Barilla Spaghetti n.5 (EAN 8076800195057)
URL: https://world.openfoodfacts.org/product/8076800195057/spaghetti-n-5-barilla
- **fetched excerpt**: ingredients durum wheat semolina pasta; per 100 g protein 13 g, fibre 3 g; origin of ingredients "Italy"; cooking time 10 min. Confidence high for what the pack states. Note this is the Italian-market pack; Barilla's wheat origin varies by market.

### Retail listing: Open Food Facts, De Cecco whole wheat spaghetti (US pack, UPC 0024094730124)
URL: https://world.openfoodfacts.org/product/0024094730124/whole-wheat-spaghetti-f-lli-de-cecco-di-filippo
- **fetched excerpt**: ingredients "Whole semolina from (wheat), niacin, ferrous lactate, thiamine, mononitrate, riboflavin, folic acid"; serving 56 g; per 100 g protein 14.29 g, fiber 8.9 g; origin of ingredients "not indicated"; no die or drying text. Confidence high.

### Retail listing: Whole Foods Market, De Cecco 100% whole wheat spaghetti
URL: https://www.wholefoodsmarket.com/grocery/product/de-cecco-100-whole-wheat-spaghetti-no-12-b083ljr5nc
- **fetched excerpt**: present: title, ingredients, serving 56 g, protein 8 g, fiber 4 g, "Vegetarian" tag. Absent: cooking time, country of origin, any bronze-die or slow-drying text. Confidence high.

### Retail listing: Instacart (Berkeley Bowl), Rustichella d'Abruzzo spaghetti 17.6 oz
URL: https://www.instacart.com/products/198205-rustichella-d-abruzzo-spaghetti-17-6-oz
- **fetched excerpt**: present: protein 13 g, fiber 4 g, sodium 10 mg, calories 343 (per 100 g, basis not labelled), size, price. Absent: ingredients, cooking time, description, origin, process text. Confidence high.

### Retail listing: You Dream Italy, De Cecco Spaghetti n°12 500 g
URL: https://www.youdreamitaly.com/de-cecco-spaghetti-n12-500-g-pasta-di-semola-di-grano-duro-a-lenta-essiccazione
- **fetched excerpt with quotes**: title "Pasta di Semola di Grano Duro a Lenta Essiccazione"; ingredients "Semola di grano duro, acqua"; "Tempo di cottura: 12 minuti (10 minuti al dente)"; "trafilatura ruvida al bronzo"; "lenta essiccazione" with no hours; origin only "Paese d'origine Italia", no wheat-origin line. Confidence high.

### Spec sheet: Barilla Foodservice full-line poster (PDF, extracted with pdftotext)
URL: https://cafemeetingplace.com/pdf/barilla-full-line-poster.pdf
- **verbatim** headings and cuts: "Over 35 different pasta cuts"; rows group "ANGEL HAIR/ CAPELLI D'ANGELO", "SPAGHETTI", "LINGUINE", "FETTUCCINE" (long) separately from "ELBOWS", "DITALINI", "GEMELLI", "ROTINI", "FARFALLE", "PENNE RIGATE", "RIGATONI" (short), "ORZO", "PASTINA", "MINI FARFALLE" (soup/mini) and "WAVY LASAGNE", "OVEN READY LASAGNE", "JUMBO SHELLS", "MANICOTTI" (oven). Confidence high that producers draw the category as a grid of shape illustrations grouped by cut.

### Primer: Barilla, "Pasta & pasta sauce pairing guide" and "Soup pasta shapes"
URL: https://www.barilla.com/en-us/help-with/pasta-kitchen-tips/pasta-sauce-pairing-guide
- Page returned 403. **search excerpt**: Barilla groups shapes as "Long Shapes, Short Shapes, Mini and Soup Shapes, and Lasagne and Oven Cuts"; long shapes "for smooth, silky sauces"; twists and spirals "create pockets that hold onto smaller ingredients". Confidence medium.

### Specialist review: Gluten Safety, "Gluten-free pasta ranked"
URL: https://glutensafe.io/gluten-free-pasta-alternatives-ranked/
- **verbatim**: brown rice pasta "Undercooked, it is gritty and starchy. Overcooked by even a minute, it becomes mushy and falls apart."; corn pasta "The texture is smoother than rice pasta and holds up reasonably well in cooking." Confidence medium.

### Primer: Feed Me Phoebe, "7 common mistakes people make when cooking gluten-free pasta"
URL: https://feedmephoebe.com/how-to-cook-gluten-free-pasta/
- **verbatim**: "Gluten-free pasta tends to be less forgiving than regular pasta and can go from hard in the center to mushy in a matter of minutes."; "Gluten-free pasta tends to be starchier than regular pasta, which sometimes creates a lot of foam on the top of the pot." Confidence medium. Second source for the narrow doneness window.

### Primer: The IBS Dietitian, "Low FODMAP pasta guide"
URL: https://theibsdietitian.com/blog/low-fodmap-pasta
- **verbatim**: "Chickpea pasta (100g or 1 cup cooked)" as a low-FODMAP serve; wheat pasta "up to 2/3 cup cooked weight". Lentil pasta serve not given. **search excerpt** (wholeisticliving.com): chickpea pasta low-FODMAP at 1 cup cooked; red lentil pasta "up to 1/2 cup". Confidence low; both are secondhand from the Monash app.

## Numbers and where they rest

| Number | Sources | Status |
|---|---|---|
| Moisture ≤12.5% | DPR 187 art. 6; IGP disciplinare; Bresciani 2022 | two or more |
| Legal protein floor 10.5% dry basis | DPR 187 art. 6 (regulator text) | one, but it is the law itself |
| IGP protein ≥13% dry basis (≈11.4 g/100 g as sold at 12.5% moisture) | disciplinare.it 2020 | one source for the number; IGP drying range confirmed by consortium. All label comparisons in the document are as sold. |
| IGP drying 40–85 °C, 4–60 h | consortium page; disciplinare.it | two (older 40–80/6–60 noted) |
| Bronze: 20–30% lower breaking strength, higher porosity | Lucisano 2008 (search excerpt); Bresciani 2022 (verbatim) | two |
| High-temperature drying >65 °C improves firmness, esp. weak semolina | Bresciani 2022 | one academic, consistent with ATK "not much textural difference" and De Cecco's "depends heavily on the grain" |
| Fast drying 80–90 °C in roughly 4–10 h (ATK 180 °F for 7–10 h; IGP minimum 4 h); slow below ~60 °C over 9–60 h (De Cecco 9–18 h by shape; Martelli ~50 h at 33–36 °C; Rustichella 40–60 h; IGP max 60 h) | De Cecco PDF; ATK ode; Noodledex; Martelli; Rustichella; Agrimò; IGP | several |
| Furosine: De Cecco's own data (market >300; its semolina pasta 50–200; its wholegrain 200–250; market wholegrain 500–600); Giannetti 2013 survey LT 107–186 vs VHT 345–506 (search excerpt) | De Cecco PDF; Giannetti 2013; Bresciani (direction) | attributed to De Cecco in the document; survey numbers labelled as search excerpt |
| Lysine loss up to 30% | Agrimò (producer) | one; kept as judgement low |
| Premium protein 13–14 g/100 g; mass-market 12.5–13 | Garofalo 14; Rustichella 13.5; Barilla n.5 13; Vincenzo's Plate rule; Barilla Al Bronzo >14; De Cecco 14 (search) | several |
| Legume pasta 18–25 g protein, 8–14 g fibre per 100 g | Banza via MyFoodDiary (19.3/8.8 per 100 g); Gluten Safety (14 g/8 g per 2 oz, an older label); Barilla Protein+ (10 g/5 g per 56 g) | two or more; label versions differ, so the document gives a range |
| Whole durum fibre 7–12 g/100 g vs 3 g white | Open Food Facts De Cecco WW 8.9; Whole Foods 4 g/56 g (7.1); Barilla whole grain 7 g/56 g (12.5, search); Garofalo white 3.0; Barilla white 3 | several |
| GI durum spaghetti al dente 42–54 (GlycemicSnap 42–48; PMC 53/54); overcooked 55–65 (GlycemicSnap only); whole wheat 40–45 (GlycemicSnap table) | GlycemicSnap; PMC7911001; PMC9689111 | sources disagree on the absolute; no delta is computed in the document |
| Legume pasta GI 25–35 | GlycemicSnap only | one; judgement low |
| Cooking loss ≤12% acceptable (PMC6722931); <3 g/100 g cited as good (Bresciani 2022); ~6–6.5 g typical (ResearchGate search excerpt, not fetched, not cited) | PMC6722931; Bresciani 2022 | thresholds disagree; recorded as a judgement, the 6–6.5 figure dropped from the document |
| Origin label: growing country + milling country; EU/non-EU wording; extended to end 2025 | FoodTimes; Food Compliance; lab-to 2024; portale 2025 (search) | two or more |

## Against the existing add-on schema

The add-on `formulations/food-beverages/add-ons/pasta.jsonc` is empty. The standards it composes from already carry `ingredients`, `high_signal_markdown`, `sizing.net_quantity`, `sizing.serving_size`, `servings_per_container`, `nutrition` (with `basis`) and `allergens`. Proposals only; nothing edited.

| Facet in this document | In add-on today | Where it belongs under AGENTS.md | Note |
|---|---|---|---|
| base_family | no (ingredients list implies it) | vertical anchor, as an enum on the leaf | Derivable from `ingredients[0]` plus title words; worth an explicit enum because it is the first cold-start question and a hard filter for celiac and high-protein buyers. |
| shape_class | no | hard filter on the leaf | Comes from the title, not the photo, so it passes schema rule 1. Keep it coarse (long strand, long ribbon, short tube, short shape, soup, sheet). Exact geometry and ridges stay with the image pipeline. |
| die_surface | no | quality currency (enum bronze / smooth / unspecified) | Stated on the pack when bronze; never stated when Teflon. We would happily augment it from producer pages, which is the currency test. |
| protein_per_100g | covered by `nutrition.protein_g` + `basis` + `serving_size` | derived, not a new field | The ranker needs it normalised to 100 g; propose the composer or a post-step compute it rather than adding a field. |
| fiber_per_100g | covered by `nutrition.dietary_fiber_g` | derived | Same. |
| drying_profile | no | high_signal_markdown | "slow dried", "lenta essiccazione", hours and °C when given. Not dense enough on listings to be a field; it is exactly what the Jev pass should keep. |
| wheat_origin | no | high_signal_markdown now; candidate hard filter later | Italian packs state it by law; export packs mostly do not. Promote to an enum only if fill rate on the catalog's Italian-market listings justifies it. |
| geographic_designation (Gragnano IGP) | no | high_signal_markdown | AGENTS.md rule 5 cuts certifications from the schema; the IGP is a provenance grade and belongs in the markdown. |
| glycemic_band | no | derived from base_family at display time | Not a stored field. |
| cooking_loss | no | out of schema | Reviewer and lab data only; lives in the expertise layer, backfilled for the top N. |

Add-on fields considered noise: none, the add-on is empty. Fields the standards carry that this category does not need as filters: `allergens.may_contain` is still useful (Garofalo declares soy and mustard; legume and gluten-free lines declare wheat cross-contact) so keep it.

## Blind-check corrections (2026-10-19)

A checker who had not seen these notes reported eight problems. What changed in category.json:

1. Protein basis made consistent. The law's 10.5% and the IGP's 13% are dry-matter figures; converted to about 9 g and 11.4 g per 100 g as sold and stated once as such. Protein facet, protein scale, IGP facet and the threshold judgement rewritten so no dry-basis number sits on the as-sold axis.
2. Cooking loss: the "6 to 6.5 g" figure rested on a paper we could not fetch (ResearchGate 403, Semantic Scholar 429), so it is out of the facet. The facet now gives the two thresholds we hold (≤12 g in PMC6722931, <3 g in Bresciani 2022) and a judgement records the disagreement and the uncited third figure.
3. Glycaemic band: the "10 to 15 points higher" delta was our arithmetic; replaced with the sourced bands (al dente 42 to 54 across sources; overcooked 55 to 65 from GlycemicSnap). GlycemicSnap re-fetched: its table gives whole wheat 40 to 45; supports text corrected.
4. Drying: fast band now "roughly 4 to 10 hours" (ATK 7 to 10 h; IGP minimum 4 h); slow band "roughly 9 to 60 hours" so De Cecco's 9 to 18 h fits. Furosine figures attributed to De Cecco's own tests and to the 2013 survey separately.
5. Modern Retail re-fetched: it carries the price comparison and the bronze-die claim only. Protein and cook-time claims dropped; supports list corrected; reference removed from the protein facet's evidence.
6. wheat_origin and die_surface reclassified from listing to researched. Hints now say aggregator listings drop the text (Whole Foods De Cecco, Instacart Rustichella) and that the absence-means-smooth rule applies only after the producer page has been checked.
7. Die-surface cold-start question now asks about the surface (rough and matte vs smooth and glossy), not the price tier; price_sensitivity removed from skip_if_known.
8. Shape question and values gained a bucket for buyers with no sauce in mind (salads, bakes, mac and cheese).

Not changed: the Giannetti 2013 numbers stay, labelled as a search excerpt, because the direction is confirmed by the academic review and De Cecco's data; the Cook's Illustrated result stays as a search excerpt because the page redirects.

## Branches (2026-10-19)

Each branch is a claim about how an expert narrows the choice; the sentence it follows from is quoted so it can be checked. Values are verbatim from the facet's values list.

**base_family**

1. `whole durum wheat` → then ask `fiber_per_100g`. From `fiber_per_100g.why_experts_care`: "this is the number that separates the fibre families from the rest and, within whole wheat, separates genuine wholegrain from blends."
2. `legume (chickpea, lentil or pea)` → then ask `protein_per_100g`, `fiber_per_100g`, `glycemic_band`; drop `wheat_origin`, `geographic_designation`, `drying_profile`. Ask: derived metric `price_per_10g_protein`: "The legume-pasta buyer is paying for protein"; `protein_per_100g` derived-metric why: "Experts read protein ... as the point of the product on legume pasta"; protein trade-off: "Legume pasta adds about half again the protein and three times the fibre of white durum"; `glycemic_band.why_experts_care`: "Legume pasta sits lower still. This matters for anyone managing blood sugar". Drop: `wheat_origin` is defined as "Where the durum wheat was grown" and the origin rule covers "pasta di semola di grano duro" (lab-to source above); `geographic_designation.why_experts_care`: the mark bundles "semolina with at least 13% protein", so it is a semolina-pasta grade; `drying_profile.backfill_note`: "Backfill the top 200 ranked wheat pastas from producer pages", and its why: slow drying "only holds together if the semolina is good".
3. `gluten-free grain blend` → drop `wheat_origin`, `geographic_designation`, `drying_profile`, `glycemic_band`. The first three for the same sentences as branch 2; `glycemic_band` because its own values list carries "not assessed (gluten-free grain blends)".

Not branched: `durum wheat semolina` and `egg pasta` are the default path, and the split rationale already says so: "the next two questions do the real narrowing". `other` is a catch-all.

**shape_class**: no branches. The register treats the two facets that follow as independent of shape: the die-surface rationale says "Within one shape this is the difference a buyer can taste", and the shape rationale says "this split only asks for the family". No facet's `why_experts_care` says it matters more or less for one shape family.

**die_surface**

4. `bronze-cut (rough, matte)` → then ask `drying_profile`, `protein_per_100g`, `geographic_designation`. From the split rationale: "Bronze-cut pasta is stated on the pack and usually comes with slow drying and higher protein, so one answer about the surface also sorts the premium tier from the mass market"; `drying_profile.why_experts_care`: slow drying "travels with high protein and bronze dies"; `geographic_designation.why_experts_care`: the mark bundles "bronze dies only, semolina with at least 13% protein ... and drying between 40 and 85 °C".
5. `smooth (Teflon-lined die)` → drop `drying_profile`, `geographic_designation`. From `gragnano_consorzio`: "IGP pasta is extruded only through bronze dies"; `drying_profile.backfill_note`: "when nothing is stated anywhere, record fast, since that is the industry default"; `noodledex_dececco_barilla`: "Barilla extrudes most products through Teflon and dries fast at high temperature".

Not branched: `not stated` is a data value, not an answer a buyer gives; the backfill note says to check the producer page.

## Trimmed detail (2026-10-19)

Sentences cut from capped fields to meet the validator. Nothing here changed a number, name or claim; the field now leads with the sentence a reviewer needs and the rest is kept below.

- `summary`: "This document covers dried pasta only. The taxonomy leaf Pasta & Noodles also holds fresh pasta and Asian noodles, which are not covered here."
- `mechanism.plain`: the dough is "kneaded into a stiff dough"; the die is "a metal plate with holes"; lysine is "an essential amino acid"; strong gluten gives "clear cooking water"; the die sentence read "bare bronze scrapes the dough and leaves it rough and matte, a Teflon-lined die leaves it smooth and shiny"; fast drying "hides a weak semolina".
- `splits[die_surface].rationale`: "the price gap is about a dollar a box" (Modern Retail: Al Bronzo $2.99 against about $1.99 for the blue box; still in the die-surface trade-off).
- `facets.base_family.why_experts_care`: "In Italy, dried pasta sold as pasta must be durum, which is why the back of an Italian pack reads "semola di grano duro, acqua" and nothing else."
- `facets.shape_class.why_experts_care`: long strands take sauces "that coat"; tubes and twists trap chunky sauces "in their ridges and pockets"; "Producers draw their whole range as a grid grouped exactly this way" (Barilla poster, kept in the visual grammar); the exact-shape example "(penne versus rigatoni)"; "whether it is ridged is visible in the photo".
- `facets.die_surface.why_experts_care`: the starch in the water is "what thickens a carbonara or cacio e pepe"; the test kitchen compared bronze-cut "than smooth ones"; fragility means it "arrives with more broken strands and clouds the water"; "bronze-cut brands say so on their own packs and sites, though third-party listings often drop the line" (still in the listing hint).
- `facets.protein_per_100g.why_experts_care`: "so the pasta stays firm past its cooking time, loses less starch to the water and holds al dente (firm to the tooth) instead of collapsing"; "All figures here are as sold, the way the nutrition panel prints them"; "The two official floors are written on dry matter and sit below the whole shelf: Italian law demands 10.5%, about 9 g as sold, and the Gragnano IGP rules (the protected designation for pasta from Gragnano) demand 13%, about 11.4 g as sold, so neither floor separates one pack from another" (the as-sold conversions stay in the protein scale and in judgement 2).
- `facets.drying_profile.why_experts_care`: "This is the facet experts rank with the die and that almost no listing quantifies"; fast drying is "standard since the 1980s"; furosine is "a marker of heat damage"; "De Cecco's own tests put market pasta above 300 mg of furosine per 100 g of protein, its slow-dried semolina pasta at 50 to 200 and its slow-dried wholegrain at 200 to 250; a 2013 survey found 107 to 186 for low-temperature pasta against 345 to 506 for very-high-temperature pasta"; "Fast drying also firms the protein net, which is why a fast-dried pasta from weak semolina still cooks acceptably and why one test kitchen found little textural difference"; slow drying "keeps the flavour and the amino acids but only holds together if the semolina is good"; "Producers who do it say so: De Cecco prints 9 to 18 hours by shape, Martelli 33 to 36 °C for about 50 hours, Rustichella 40 to 60 hours, and the Gragnano IGP rules allow 40 to 85 °C for 4 to 60 hours" (all still in the references and the drying-hours scale).
- `facets.wheat_origin.why_experts_care`: "Italian packs have had to state the growing country and the milling country since 2018 under a rule renewed each year, most recently to the end of 2025, so a pack sold in Italy reads Italia, UE, non UE or UE e non UE."
- `facets.glycemic_band.why_experts_care`: "Pasta is unusual among refined wheat foods"; overcooking undoes the effect "because fully swollen starch digests fast"; "This matters for anyone managing blood sugar, and it is the one place where the cook time on the box is a health number, not a convenience."
- `facets.cooking_loss.why_experts_care`: "Published thresholds disagree on where good sits"; "the 2022 process review"; "Cook's Illustrated's lab found its winning spaghetti was both the strongest strand and the one with the lowest cooking loss."
- `judgements[1].note` (slow drying as annotation): "A novice cannot answer a question about it" shortened; no fact cut.
- `judgements[4].note` (glycaemic index): "both agree cooking time matters" and "we show legume as a band" shortened; no fact cut.
- `judgements[8].note` (cooking-loss thresholds): "a search excerpt of a third paper put typical spaghetti at 6 to 6.5 g but we could not fetch it, so it is not cited" (the paper is the ResearchGate "Cooking Properties and Heat Damage of Dried Pasta" excerpt under PMC6722931 above).
- `references[giannetti_2013_furosine].supports[0]`: "furosine in market pasta ranged 107 to 506 mg per 100 g protein ... (search excerpt, abstract not fetched)"; the abstract was refused by Wiley and Semantic Scholar, as recorded under the source above.
- `references[cooks_illustrated_spaghetti].supports[0]`: "the lab found its strands the strongest with the lowest cook loss (search excerpt; page now redirects)".
- `references[barilla_protein_plus].supports[0]`: "from a semolina, lentil, chickpea and pea blend".

## Corrections (2026-10-19, second pass)

Validator run after the blind check listed 24 problems across 22 fields. Changes to category.json:

- Branches added: 5 (base_family 3, die_surface 2, shape_class 0, see "Branches" above). The viewer now draws a decision tree instead of a straight sequence.
- Questions rewritten: 3 split questions and the 3 matching cold_start_questions (6 fields), each at most 12 words, no preamble, ending in a question mark, answers offered in the question. base_family was already under the cap and now also names egg pasta.
- Fields trimmed to cap: 18 (summary, mechanism, 2 split rationales, 8 why_experts_care, 3 judgement notes, 3 reference supports). Every cut sentence carrying a fact, number or source is under "Trimmed detail" above.
- Not changed: generated_at, facet ids and count, judgements, references and evidence ids, priority scores, key order.

### Correction, 2026-10-19 (blind check of branches)
- die_surface = smooth: `drying_profile` removed from `makes_irrelevant`. The checker graded the leg WEAK: the register states that bronze dies travel with slow drying, not the converse, so a smooth die does not make drying profile irrelevant. `geographic_designation` stays, since IGP pasta is bronze-cut by rule.