# Graters, Peelers, Zesters & Corers: research notes

Date: 2026-10-25. Scope: hand tools that grate, zest, peel, cut strands from or core fruit, vegetables, cheese and spices: rasp graters (zesters), box graters, flat and paddle graters, rotary drum graters, ginger graters, Y and straight peelers (smooth and serrated), julienne peelers, channel knives and loop-style citrus zesters, manual spiralizers (handheld and crank), apple corers and wedgers, crank apple peeler-corer-slicers, and pineapple corers. Out of scope: electric spiralizers, food-processor shredding discs and spice mills (machines: Food Grinders, Mills & Slicers); cut-resistant gloves (no leaf query in this group; mentioned only as safety context). Also out of scope but currently unowned by any leaf, a gap this register cannot close on its own (2026-10-25 blind check): mandoline slicers (`knives`'s query list carries "mandoline slicer", but `knives`'s own notes send them back here with no facets, and they overlap the multi-drum rotary graters sold here as "Rotary Vegetable Mandoline"); push-down garlic and onion choppers (`mashers`'s notes send them to "`knives` or `graters`"; neither register claims them); cherry and olive pitters, strawberry hullers and avocado slicers (routed to `kitchen-other`, whose catalog status is `skipped`). Fixing this needs a change to the owning register (`knives`, `mashers` or `kitchen-other`), which is outside this folder. The register keeps one facet set and lets the `tool_job` split divide the four tool families the label bundles.

Quote labels: **verbatim** = the sentence in quotation marks was in raw text I read myself (curl plus tag stripping, or pdftotext on a downloaded PDF); **fetched excerpt** = the WebFetch tool summarised or reworded the page; **search excerpt** = the claim came only from a search-engine summary. No search-excerpt-only claim is in category.json. Every quote below is verbatim unless labelled otherwise; all pages were read with curl into the scratchpad and quoted from the text.

## Seed

Seed output, recorded verbatim (first run):

```
=== Example entry (from an unrelated category) ===
category: bedsheets
what this shows: the expert split is often not the number everyone advertises. Thread count is the generic trap here. Weave is what changes how the sheet feels and sleeps.

<example>
{
  "id": "weave",
  "label": "Weave",
  "definition": "The pattern the threads are crossed in, which decides how the sheet feels and how it handles heat.",
  "why_experts_care": "Percale is a simple one-over-one weave that feels crisp and cool and gets softer with washing. Sateen puts more thread on the surface so it feels silky and warmer and shows wear sooner. Thread count, the number most listings lead with, matters far less than which of these two you pick, and above about 400 it is mostly marketing.",
  "kind": "categorical",
  "values": ["percale", "sateen", "linen", "flannel", "jersey"],
  "roles": ["split", "filter"],
  "novice_answerable": true,
  "cold_start_question": "Do you sleep hot, or do you like a sheet that feels smooth and a bit warm?",
  "source": "listing",
  "listing_field_hint": "Usually in the title or the first bullet. Sometimes only in the fabric care tab.",
  "backfill_cost": "low",
  "backfill_note": "When missing, the product photo and the description words 'crisp' or 'silky' usually settle it.",
  "visual": { "chartable": false },
  "priority": { "narrows_set": 4, "cost_to_answer": 1 },
  "evidence": ["wirecutter_sheets", "textile_primer"]
}
</example>

=== Persona ===
A pastry chef, who thinks in ratios, temperatures and why a recipe fails

Ask yourself two things and write the answers under 'Seed' in notes.md:
  1. What is the graters equivalent of the example above?
  2. What would this persona notice first about graters that a typical buyer would not?
Keep what survives the data check. Drop the rest.
```

Process note: seed.py was invoked a second time by mistake while creating the output folder; its output went to /dev/null and was not read. The answers below use only the first run, as the brief requires.

1. **The graters equivalent of thread count is the number of sides on a box grater (or pieces in a grater set).** It is the advertised number ("4-sided", "6-sided", "5-in-1") and it matters less than the thing experts split on, which is how the teeth were made. ATK: "many of us only use the side with the largest teeth"; Reviewed: "Most people only use the coarse setting" and its 6-sided Utopia box: "none of those sides really exceeded expectations". The weave equivalent is tooth type, etched against stamped, which decides fluffy against thick output (ATK, Reviewed, Microplane catalogue). **Kept** as the `tooth_type` facet, the `grate_texture` split, the `blade_style` facet (sides as a list of named textures rather than a count) and the "most use only the coarse side" judgement.
2. **The pastry chef notices that grated volume depends on the tool, so a recipe's "1 tablespoon zest" or "1 cup Parmesan" is not a fixed amount.** ATK rasp review: "Differences in the design of the graters' teeth affect the texture of grated foods and the postgrating volume. For this reason we recommend measuring ingredients by weight instead of volume." Microplane's blade guide describes its zester output as "a soft, ultra- fluffy grating result". The chef would also want zest without pith, because pith is bitter (Serious Eats: "the underlying bitter, white pith"; ATK channel knives on pith bitterness). **Kept** as the weigh-it judgement and inside the rasp tooth-height trade-off. **Dropped**: a zest-yield-per-lemon derived metric (no source gave a figure, let alone two); grating frozen butter for pastry as a use case (not in any fetched source); a volume-per-ounce conversion for rasp versus box Parmesan (I recall ATK publishing one but could not fetch it, and the search budget was gone).

## Sources and what each supports

### Specialist reviewers who measure

**America's Test Kitchen, The Best Graters, Miye Bromberg, published 6 May 2026**
https://www.americastestkitchen.com/equipment_reviews/2131-graters
- verbatim: "The holes can be stamped—punched out of a thick, rigid sheet of metal, forming teeth that rise relatively high above the grating surface and are sharpened in a separate step by the manufacturer. Or the holes can be etched, the pattern of the holes printed photographically onto a thinner, more flexible sheet of metal, and then selectively eroded by chemicals. After the holes have been etched, the teeth are pushed up by a machine."
- verbatim: "The type of teeth doesn't necessarily dictate a grater's sharpness; we've tested stamped and etched graters that were razor-sharp and also reviewed models in both categories that were dull as rocks. Rather, the main difference between stamped and etched graters came down to how thickly they shredded food—and how easy they were to use as a result."
- verbatim: "In general, we preferred graters with stamped teeth." "More often than not, we ended up extruding the mozzarella through the etched holes, creating odd, pellet-like shreds".
- verbatim: "Box graters with etched holes do have one advantage: Their small-holed sides are much better for zesting citrus and grating hard cheeses such as Parmesan than the pinholes found on stamped graters, which clog easily. But this isn't a big issue for us, since we rely on our favorite rasp graters to perform these jobs anyway."
- verbatim: "We liked models with grating surfaces that were at least 6 inches in length and had a total area of at least 15 square inches."
- verbatim: "Models with dull teeth struggled to shred food and were more dangerous to use, too ... which made it more likely that our knuckles would scrape on the teeth."
- verbatim: collection bins "raised the graters' center of gravity too high off the counter, which made them more prone to tipping ... Unless you shred only a half-cup of food at a time".
- verbatim: "many of us only use the side with the largest teeth". One model "comes with ... a hand guard, which we found awkward to use."
- Spec table per product lists "Tooth Type: Stamped / Etched" and "Dishwasher-Safe: Yes, top rack only".
- Supports: mechanism, `tooth_type`, `grate_texture`, `grater_format`, `grating_surface_area`, bin trade-off, sides judgement, safety judgement. Confidence high.

**America's Test Kitchen, The Best Rasp Graters, Chase Brightwell, published 11 June 2025**
https://www.americastestkitchen.com/equipment_reviews/1798-rasp-style-graters
- verbatim: "Grace Manufacturing. They pioneered and patented a chemical etching method ... Grace used their etching patent to launch Microplane ... Grace's patent expired in 2011".
- verbatim: zesters "relatively long, narrow blades"; paddle models "usually grated more food per grating pass".
- verbatim: "we preferred models with grating surface areas measuring at least 7.25 square inches"; "Our favorites had more than 400 teeth, while our least favorite, slowest models had fewer than 300."
- verbatim: "Teeth that jutted upward dramatically ... carved too deeply into food, taking off excess pith"; "small teeth that lay too flat ... the slowest took nearly three times as long as our winners".
- verbatim: "we recommend measuring ingredients by weight instead of volume."
- verbatim: "models measuring 13 inches or more in total length felt too long and cumbersome".
- verbatim: "Every model we tested is technically dishwasher-safe, but the dishwasher's corrosive detergent, hot water, and heightened water pressure will certainly dull the blades. We suggest washing it by hand ... store your grater in its plastic sheath".
- Supports: `grater_format`, tooth-height trade-off, weigh judgement, dishwasher judgement, `blade_cover`, the 12-inch length point (with Serious Eats). Confidence high.

**America's Test Kitchen, The Best Vegetable Peelers, Lisa McManus, published 27 June 2025**
https://www.americastestkitchen.com/equipment_reviews/2592-the-best-vegetable-fruit-peeler
- verbatim: "There are two main types: Y-shaped and straight; we chose favorites of each." "Whether you prefer a Y-shaped or straight peeler is personal." "We slightly prefer a Y peeler because its open Y shape gives us an unobstructed view".
- verbatim: "All of the peelers we tested had swivel blades."
- verbatim: "In our testing, a few stainless-steel blades dulled, but models with carbon-steel blades stayed sharp." "Rust isn't dangerous, and it's easily scrubbed off the blade." "(Those with carbon-steel blades must be hand-washed and dried.)" "The dishwasher is convenient but will shorten the life of your peeler."
- verbatim: "Serrated peelers are designed to remove the skin of very soft fruits and vegetables, such as ripe peaches, plums, and tomatoes ... we don't prefer this style as an all-purpose peeler." The serrated "universal" peeler "peeled too deeply and left grooves".
- verbatim: "Because peelers' dual blades are set quite close together, it would be very difficult to reach a blade at the correct angle and sharpen it effectively."
- verbatim (Kuhn Rikon spec): "Weight : ½ oz", "Materials : Plastic, carbon steel", "Care : Dishwasher-safe but hand washing recommended; always hand-dry thoroughly", "Works For Lefties : Yes, for peeling, but potato-eye remover favors right-handed users". RSVP model: "Weight : 4¼ oz", "Works For Lefties : No".
- Test list includes "Remove a long, oval strip of lemon peel with as little pith as possible (as for cocktail garnish)".
- Supports: `peeler_shape`, `blade_edge`, `blade_steel`, `works_left_handed`, swivel and serrated and sharpening judgements. Confidence high.

**America's Test Kitchen, The Best Ginger Graters, updated 10 December 2024**
https://www.americastestkitchen.com/equipment_reviews/1869-ginger-graters
- verbatim: "Some had rasp-style 'etched' holes, some had puckered 'stamped' holes punched out by a machine, and others—traditional Japanese oroshigane—had either metal teeth or ceramic nubs."
- verbatim: "Our two favorites were the fastest, producing 1 tablespoon of puree in just 15 seconds; both had etched holes. The worst model, with metal teeth, took more than 4 minutes." "The best models had at least 5 square inches of surface area." Etched graters left "just 1 to 3 unusable grams of fibrous pulp"; stamped and toothed "7 to 10 grams of pulp and only 4 to 7 grams of puree".
- verbatim: "Our favorite paddle-style rasp grater grates ginger faster and more efficiently than any other option" — the Microplane Gourmet Series Fine Grater (re-verified 2026-10-25; the page no longer names a "Home Series Fine Grater").
- Supports: `tooth_type` (ginger waste), `grater_format` (ginger graters add nothing over a fine paddle rasp), oroshigane vocabulary. Confidence high for direction; the gram figures are single-source and stay here.

**America's Test Kitchen, The Best Rotary Graters, published 13 April 2020**
https://www.americastestkitchen.com/equipment_reviews/2153-rotary-graters
- verbatim: "because the food is completely enclosed inside the hopper, your fingers never come anywhere near the sharp teeth of the drum. This clever design also means that rotary graters are safe for even children to use."
- verbatim: "On average, the rotary graters took about half a minute to finely grate an ounce of Parmesan; the rasp-style grater took more than twice that, at about 1 minute and 20 seconds."
- verbatim: "Some models require you to install the crank on the right side of the rotating drum, a position that's awkward for lefties".
- verbatim: "Rotary graters just have more parts than the simple rasp-style grater and are thus a bit more of a pain to clean."
- Supports: rotary family, `grater_format`, `works_left_handed`, rotary trade-off. Confidence high; the article is from 2020, older than the others.

**America's Test Kitchen, The Best Julienne Peelers**
https://www.americastestkitchen.com/equipment_reviews/2636-the-best-julienne-peelers
- verbatim: julienne peelers "are very inexpensive and compact, work fast, and require no skill. They also take just seconds to clean." "Sharp blades make safer tools: They bite in exactly where you want them and don't skid off to cut you." "Peelers with more space between the pair of blades performed better ... cut thicker, more symmetrical squared-off matchsticks instead of thin, flat noodle shapes." "The OXO peeler (left) has a ⅛-inch gap between its dual blades. The GIR peeler, right, has just half that space, at 1/16 inch."
- Supports: `strand_tool`, safety judgement. Confidence high.

**America's Test Kitchen, The Best Channel Knives for Cocktails, updated 22 January 2025**
https://www.americastestkitchen.com/equipment_reviews/1736-citrus-zesters-slash-channel-knives
- verbatim: "There is one big problem with all of the channel knives we've tested: none are great for lefties. That's because the blades are positioned on the right side of the tool and will only cut if you push the knife from right to left". "any bitterness from the pith is unlikely to be soluble in a cold cocktail".
- Supports: `works_left_handed`, channel-knife judgement, `tool_job` mapping of cocktail twists to peeling. Confidence medium (single reviewer).

**America's Test Kitchen, Apple Corers/Slicers, published 24 June 2025**
https://www.americastestkitchen.com/equipment_reviews/2362-apple-corer-slicers
- verbatim: "crank style, which peels, cores, and/or slices fruit with the turn of a crank, and push style (shaped like a wagon wheel), which simply cores and slices fruit"; "Push-style devices can be used year-round on both firm and soft fruit ... but the crank style requires crisp, firm apples"; "you can peel, core, and slice an apple in about 15 seconds"; "if your countertop is made of a material with a rough texture or tiny holes in its surface, such as granite or marble, you might have trouble getting the device to attach"; Norpro wedger "cores and slices apples into 16 narrow slices", OXO "into eight chunky wedges".
- Supports: `corer_type`, crank trade-off. Confidence high for the styles; the 15-second and 8/16 figures are single-source and kept here, not in the JSON.

**America's Test Kitchen, Cut-Resistant Gloves**
https://www.americastestkitchen.com/equipment_reviews/2066-cut-resistant-gloves
- verbatim: "there's a good chance that at some point you've cut yourself while breaking down a chicken, skinned a knuckle grating cheese, or sliced a fingernail (or worse) using a mandoline." "their products are cut-resistant but not cut- proof". ANSI cut scores in the lineup "ranged from A3 ... to A7".
- Supports: safety judgement. Gloves themselves are out of scope. Confidence medium.

**Serious Eats, We Tested 12 Microplanes (Rasp-Style Graters), syndicated on Yahoo Lifestyle** (seriouseats.com itself blocked curl and WebFetch; the Yahoo copy carries the full article text)
https://www.yahoo.com/lifestyle/tested-12-microplanes-rasp-style-224738749.html
- verbatim: "In 1991 , Grace Manufacturing invented the Microplane".
- verbatim: lower performers "had too-small holes (like the OXO model, which had teeth that were 3/32-inch wide compared to our favorite Microplane's 1/8-inch width)"; others "removed the pith as well as the zest". "OXO Good Grips Etched Zester and Grater : This rasp-style grater was one of the worst-performing graters we tested".
- verbatim: "we liked rasp-style graters that were one to one-and-a-half inches wide ... And we preferred graters that were about 12 inches long in total—anything more than that and the grater felt awkward and unbalanced to use."
- verbatim: "always store the grater with the cover on to prevent injury and the grater from dulling." "No, you shouldn't put a Microplane in the dishwasher, as it'll dull the blade faster."
- Supports: tooth-height trade-off, etched-copies judgement, 12-inch point (with ATK's 13), `blade_cover`, dishwasher judgement. Confidence high.

**Reviewed (USA Today network), The Best Graters, Lindsay D. Mattison**
https://www.reviewed.com/cooking/best-right-now/the-best-graters
- verbatim: "Unlike traditional stamped-style models, the blades for this grater are etched into the metal using a chemical process that creates super-sharp edges."
- verbatim (stamped Cuisinart): "The one area where the stamped grater can't compete, though, is on the fine holes, which turned Parmesan into a powder and tended to trap more lemon zest than it let through."
- verbatim: "Etched graters ... The metal used for these types is thinner, though, so they will flex in and out when being used with heavy-duty foods." Cuisipro etched box: "hard vegetables, which caused the front to bend in slightly".
- verbatim: "Most people only use the coarse setting". Utopia 6-sided: "none of those sides really exceeded expectations ... a narrow surface area that slowed us down".
- verbatim: "Like most blades, the holes on a grater can wear down over time, and unfortunately, they can't be sharpened."
- verbatim: OXO box with container "rocking dangerously side to side"; KitchenAid container "did cause the grater to swing around a bit".
- Supports: second source for etched versus stamped output, flexing, sides judgement, sharpening judgement, bin trade-off, surface area. Confidence high.

### Brand spec sheets

**Microplane Product Catalog 2026 (PDF, read with pdftotext)**
https://central.microplane.com/wp-content/uploads/2021/12/Microplane-2022-catalog.pdf
- verbatim: "Microplane® puts a photosensitive resist film onto first-rate stainless steel sheets ... A special solution is sprayed on the sheets and dissolves the exposed parts ... In the final manufacturing process, the teeth are pushed up to create a perfect cutting angle".
- verbatim: "Microplane® - the original manufacturer of photo- etched culinary tools (patent protected in USA from 1990 to 2010)".
- verbatim (blade guide): Zester "a soft, ultra- fluffy grating result"; Spice "our smallest blade style ... a super fine powder-like grating result"; Fine "a very delicate, fine texture"; Coarse "small strands"; Ultra Coarse "large strands"; Ribbon "designed to grate in both directions"; Large Shaver "wide but extremely fine shavings"; Extra Coarse "large strands ... very similar to 'julienne-strips'".
- Product blocks print "Photo-etched, long-lasting, ultra-sharp, stainless steel blade - Made in USA" and "Dishwasher Safe / Hand Wash Cover".
- Supports: mechanism, blade-style scale (the Grate scale grammar), patent dates, fluffy volume. The PDF's URL says 2022 but its pages say "Microplane Catalog 2026". Confidence high as the brand's own account.

**Microplane, About Us** https://microplane.com/pages/about-us
- verbatim: "Most graters on the market use a stamped metal process. In this process, the teeth are made the same thickness as the rest of the blade. As a result, stamped blades are dull."
- Supports: the brand side of the etched-versus-stamped disagreement.

**Microplane product pages** (read from the store's product JSON, the same text as the HTML pages)
- Premium Classic https://microplane.com/products/premium-classic-series-lemon-zester: "Every Microplane Premium Classic Zester/Grater comes with a reusable protective cover, ensuring safe storage and maintaining blade sharpness."
- Professional Y Serrated https://microplane.com/products/professional-y-serrated-blade-stainless-steel-peeler: "a double ground, serrated blade, making it the ideal tool for tackling the skins of potatoes, carrots, parsnips"; spec fields "Overall Dimensions", "Blade Length: Approximately 2.17 inches", "Dishwasher Safe: Yes".
- Jar Top Spiralizer https://microplane.com/products/jar-top-spiralizer-black-html: a handheld spiralizer that "Screws securely onto any standard wide-mouth Mason jar".
- Fields Microplane lists, in order: blade style, blade material and origin, dishwasher, cover, dimensions, blade length. No tooth count, no surface area.

### Primers

**Wikipedia, Peeler** https://en.wikipedia.org/wiki/Peeler
- verbatim: "Most handheld peelers are either straight or Y-type"; "Swivel peelers have the blade mounted on a pivot"; Jonas peeler "The blade has two edges to enable use in either direction, and by either hand"; Lancashire "often single edged"; Rex Y peeler "a pivoting carbon steel blade with dual edges"; "A mechanical apple peeler is a crank operated device that peels apples and optionally cores and slices them".

**Wikipedia, Grater** https://en.wikipedia.org/wiki/Grater: "Other styles include paddles, microplane/rasp graters, and rotary drum graters."

**Wikipedia, Spiral vegetable slicer** https://en.wikipedia.org/wiki/Spiral_vegetable_slicer: "Spiralizers usually contain three blades: a round blade for spaghetti , a small flat blade for ribbons, and a large wide blade for spiral strands." "Vegetables are clamped between the blade and crank." Single source for "three", so the JSON says "interchangeable blades" without the count.

**Wikipedia, Apple corer** https://en.wikipedia.org/wiki/Apple_corer: tube corer "removes the core to the diameter of the circular cutting device"; another type "both cores and slices the apple".

**Wikipedia, Zester** https://en.wikipedia.org/wiki/Zester: loop zester "perforated with a row of round holes with sharpened rims"; Microplane "marketed them as 'zester / graters'". Read for vocabulary; not cited in the JSON.

**Wikipedia, Microplane** https://en.wikipedia.org/wiki/Microplane: "in 1991 would come to be called Microplane". Read; not cited in the JSON.

**Tasting Table, Alton Brown's Preferred Type Of Vegetable Peeler** https://www.tastingtable.com/1141693/alton-browns-preferred-type-of-vegetable-peeler/
- verbatim (quoting Brown on Good Eats): "harp or 'y' peelers. They usually have a deeper, wider bite so they're really adept at handling heavy peels and large surface areas like a really big eggplant."
- Supports: Y-peeler bite on heavy peels. Kind "other" (a food site reporting a chef's view). Confidence medium.

### Regulators

**Regulation (EC) No 1935/2004** (EUR-Lex returned an empty 202 bot challenge to curl; read via Internet Archive snapshot https://web.archive.org/web/2024/https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32004R1935)
- verbatim, Article 3: materials "shall be manufactured in compliance with good manufacturing practice so that, under normal or foreseeable conditions of use, they do not transfer their constituents to food in quantities which could: (a) endanger human health; or (b) bring about an unacceptable change in the composition of the food; or (c) bring about a deterioration in the organoleptic characteristics thereof."
- verbatim, Article 15(1)(a): "the words 'for food contact', or a specific indication as to their use, such as coffee machine, wine bottle, soup spoon, or the symbol reproduced in Annex II".
- Supports: the food-contact judgement, including that an obvious use indication substitutes for the symbol. Confidence high.

**21 U.S.C. 321(s), via Cornell LII** https://www.law.cornell.edu/uscode/text/21/321
- verbatim: "The term 'food additive' means any substance the intended use of which results or may reasonably be expected to result, directly or indirectly, in its becoming a component or otherwise affecting the characteristics of any food (including any substance intended for use in producing, manufacturing, packing, processing, preparing, treating, packaging, transporting, or holding food ...)".
- Also read 21 U.S.C. 348(h)(6) (food contact substance: "any substance intended for use as a component of materials used in manufacturing, packing, packaging, transporting, or holding food"), which omits "preparing"; replaced by 321(s) in the JSON because a grater prepares food. FDA's own FCS page returned no definition through WebFetch and an empty body to curl.

### Retail listings (what data is present)

**Amazon, Microplane 4-Sided Box Grater (B000W8AMOO)**: title "Microplane Four Sided Stainless Steel Ultra-Sharp Multi-Purpose Box Grater - Slicer, Fine, Ribbon, and Extra Coarse Blade Styles"; rows "Material Stainless Steel", "Style Box Grater", "Product Dimensions 10.25"L x 5.12"W x 3.12"H", "Item Weight 5.92 ounces", "Manufacturer Warranty Description 1 year"; bullet "original, ultra-sharp, photo-etched made in USA blades"; description "Safety cover included". Brand panel: "Graters (excluding Master Series) are dishwasher safe. However, to ensure that the blades stay sharp as long as possible, we recommend washing them by hand." Related titles on the page: "OXO Good Grips Etched Box Grater with Removable Zester | 4-Sided Grater", "Utopia ... 6-Sided".

**Amazon, Kuhn Rikon Original Swiss Peeler (B0000DE824)**: bullets "Super-sharp carbon steel blade", "Ergonomic design works in right or left hand", "Hand wash and towel dry promptly after use; Do not soak in water; Do not clean in dishwasher"; rows "Blade Edge swiss", "Blade Material Type Carbon Steel", "Item Weight 0.01 kg". Related titles: "KitchenAid Y Peeler with Ergonomic Handle and Single Edge Blade".

**Amazon, OXO Good Grips Swivel Peeler (B00004OCIP)**: rows "Blade Edge swivel", "Blade Material Type Stainless Steel", "Item Weight 87 g", "Style Name Swivel Peeler"; related titles "OXO Good Grips Y-Peeler | Sharp Stainless Steel Blade ... Dishwasher Safe".

What this tells the scraper: titles carry the tool name, "Y-Peeler" or "Swivel", "Etched" when etched, blade styles (Microplane), "Dishwasher Safe" and "serrated". Amazon's structured rows carry Material, Blade Material Type, Style, dimensions and weight. Amazon's "Blade Edge" row is free text ("swiss", "swivel") and useless. Spiralizer and corer listings were not fetched, so `strand_tool` and `corer_type` are marked derived from title keywords, with the keyword rule stated, rather than listing.

### Could not reach

- seriouseats.com (people.inc access block for curl; WebFetch refused the domain; no Wayback snapshot for the guessed URLs). The rasp review was read via its Yahoo syndication; Serious Eats peeler, box-grater and spiralizer reviews were not read.
- nytimes.com/wirecutter (JS wall for curl; WebFetch refused). Wirecutter's peeler pick appears only as a search excerpt and in Tasting Table's second-hand mention; not cited.
- kuhnrikon.com (Anubis bot challenge), oxo.com (Cloudflare challenge), saveur.com and gearpatrol.com (empty bodies).
- ATK's spiralizer review no longer appears in ATK's reviews sitemap; the `1824-handheld-spiralizers` URL redirects to small food processors.
- Search excerpts seen but not used: CPSC/NEISS kitchen-injury and mandoline-injury figures (third-party blogs summarising NEISS; mandolines are a sibling's product and no figure was fetched from CPSC).
- WebSearch: the session's search budget (200 calls, shared with the other concurrent runs) was exhausted after 10 searches of mine, so later sources were found by reading links inside fetched pages and the ATK sitemap.

## Data points used for derived metrics

The register has no derived metric. The candidates (zest yield per lemon, teeth per square inch, grated volume per ounce) had no second source, so none went in. Measured points kept for reference:

| Point | Value | Source | Second source |
|---|---|---|---|
| Rasp surface floor | 7.25 sq in | ATK rasp | none |
| Coarse grater floor | 6 in long, 15 sq in | ATK graters | none |
| Ginger grater floor | 5 sq in | ATK ginger | none |
| Microplane Gourmet Fine surface | 8.5 sq in | ATK ginger spec table | none |
| Rasp width preferred | 1 to 1.5 in | Serious Eats | none |
| Rasp overall length ceiling | about 12 in preferred; 13 in or more cumbersome | Serious Eats; ATK rasp | yes, two sources (used in JSON as "about 12 inches") |
| Teeth on best rasps | more than 400 (worst under 300) | ATK rasp | none |
| Rotary Parmesan time | about 30 s per oz against 1 min 20 s for a rasp | ATK rotary | none |
| Ginger waste from 14 g | etched 1 to 3 g; stamped and toothed 7 to 10 g | ATK ginger | none |
| Peeler weight | Kuhn Rikon ½ oz (ATK) and 0.01 kg (Amazon); RSVP 4¼ oz (ATK); OXO swivel 87 g (Amazon) | ATK peelers, Amazon | Kuhn weight in two sources; not used in JSON |
| Julienne blade gap | ⅛ in (OXO) against 1/16 in (GIR) | ATK julienne | none |
| Crank apple speed | about 15 s per apple | ATK apple | none |
| Wedger slices | 8 or 16 | ATK apple | none |
| Rasp tooth width | 1/8 in (Microplane) against 3/32 in (OXO) | Serious Eats | none |

## Numbers in category.json and how many sources stand behind each

- "about 12 inches" overall rasp length (grating_surface_area): Serious Eats 12 in, ATK 13 in or more too long. Two sources.
- Patent dates 1990 to 2010 / 2011 (judgement 2): Microplane catalogue and ATK, recorded as a disagreement.
- "6-sided" and "four- to six-sided": product descriptors (ATK "Most of these have four sides"; Reviewed's 6-sided Utopia). Two sources for the span.
- EU Regulation 1935/2004 and 21 U.S.C. 321: identifiers, not measurements.
- "top 50" in backfill notes: a pipeline choice, ours.
- Removed during the self-check because each had one source: "1 to 1.5 inches wide", "up to three times as long", "8 chunky or 16 slim slices", "about 15 seconds", "about half a cup", "three blades", "worst of 12".

## Against the existing add-on schema

Add-on read: `supplies/kitchen/add-ons/grater.jsonc`, which is empty (`{}`), composed with `kitchen-standard.jsonc` (materials with roles including `blade` and `working_head`, `max_heat_temp`, `dishwasher_safe` with `top_rack_only`) and `supplies-standard.jsonc` (high_signal_markdown, sizing with dimensions, weight, pack_count, pieces_included_text). Kinds follow AGENTS.md: vertical anchor, category quality currency, hard filter, sizing, or markdown.

| Facet in this doc | In schema today | Proposed kind under AGENTS.md | Note |
|---|---|---|---|
| tool_job | no | hard filter (enum) | Proposal: `tool_type` enum [rasp_grater, box_grater, flat_grater, rotary_grater, ginger_grater, peeler, julienne_peeler, channel_knife, spiralizer, apple_corer_wedger, apple_peeler_corer, pineapple_corer, other]. `tool_job` is then derived from it at query time. Text fact: the title names it. |
| grate_texture | no | derived at query time from blade_styles | Not a stored field. |
| peeler_shape | no | none (image pipeline) | Y against straight is answered by the photo, and titles mislead ("Swivel Peeler" for straight ones). AGENTS.md rule 1 keeps it out of the schema. |
| tooth_type | no | hard filter (enum) | Proposal: `grating_tooth_type` enum [etched, stamped, raised_teeth, not_applicable, unspecified]. Not photo-answerable at thumbnail size; the text says "etched". The category's one true quality currency. |
| grater_format | no; partly via tool_type above | folded into the `tool_type` enum | rasp zester against rasp paddle is photo-answerable; leave that split to images. |
| blade_style | no | hard filter (multi-enum) | Proposal: `blade_styles` array of enum [spice, zester, fine, medium, coarse, ribbon, extra_coarse, ultra_coarse, shaver, slicer, julienne, serrated]. Covers peeler edges too, so `blade_edge` needs no field of its own. |
| grating_surface_area | no | none (researched, outside schema) | Only reviewers measure it. Overall dimensions in `sizing.dimensions` overstate it. Top-N backfill from reviews if at all. |
| blade_edge | no | folded into `blade_styles` (serrated, julienne) | |
| blade_steel | yes, `materials[]` with `role: blade` and `material: carbon_steel / stainless_steel` | already covered | Make sure extraction fills the blade role, not just body; Amazon's "Blade Material Type" row maps straight to it. |
| works_left_handed | no | high_signal_markdown | Sparse on listings and not worth augmenting for every row; a markdown line ("crank fits either side", "right-handed only") is enough. |
| strand_tool | no; folded into `tool_type` | | |
| corer_type | no; folded into `tool_type` | | Slice count for wedgers ("16 Slices") belongs in high_signal_markdown. |
| blade_cover | no | high_signal_markdown | AGENTS.md cuts "curated feature vocabularies"; a cover is a what's-included line. |
| dishwasher_safe (kitchen-standard) | yes | keep, with the reviewers' caveat in the register | Reviewers and brands disagree on whether it is wise; the field records the label, the judgement records the advice. |

Fields this register considers noise for this leaf: `max_heat_temp` (no tool here meets heat; leave null); `sizing.capacity` (only box-grater catch bins have one, and reviewers advise discarding them); `sizing.pack_count` is fine for peeler multipacks but carries no expertise. Warranty text on listings (Microplane "1 year") is already dead under rule 5.

## Branches

Each leg is traced to a sentence in category.json or in the sources above. Split 1's `when` values are its facet's values; the question offers exactly those four answers. Split 2's question offers "Fluffy zest and Parmesan", "thick cheddar shreds" and "both", which map to the values fluffy, thick shreds and both by the rule in `grate_texture`'s definition.

**Split 1, tool_job ("Grating, peeling, veg noodles or matchsticks, or coring?")**

- when `grating`: then_ask grate_texture; makes_irrelevant peeler_shape, blade_edge, blade_steel.
  - grate_texture: "Within graters this is the expert's split, because it decides the tooth type" (split 2 rationale).
  - peeler_shape, blade_edge and blade_steel irrelevant: all three are defined as properties of a peeler ("Whether the blade sits across the top of a Y-shaped frame ...", "Whether a peeler's blade is smooth or serrated", "Whether a peeler's blade is carbon steel ... or stainless"); a grater has none of them. [2026-10-25: blade_steel was left live on this branch in the pre-check register; added to makes_irrelevant, since it is a peeler facet by definition.]
- when `peeling`: then_ask peeler_shape, blade_edge, blade_steel; makes_irrelevant grate_texture, tooth_type, grater_format.
  - The three asks are the peeler facets; each definition names a peeler. blade_steel: "ATK's long-time winner has a carbon-steel blade" (blade_steel, why experts care).
  - The three skips are defined on graters ("What the grater produces", "How the grating teeth were made", "The body the grating surface sits on").
  - Caveat, unresolved: `peeling` also holds channel knives and loop-zester-plus-channel-knife combos (tool_job definition, 2026-10-25 revision). None of peeler_shape, blade_edge or blade_steel describes a channel knife's single notched blade. The schema has no way to branch on a value within `peeling`, so a channel-knife buyer still sees these three asks; this is recorded as a known gap, not fixed, because fixing it would need a new, sourced sub-facet.
- when `veg noodles or matchsticks`: then_ask strand_tool; makes_irrelevant grate_texture, tooth_type, grater_format, blade_edge, blade_steel.
  - strand_tool is defined as "The tool type for cutting vegetables into strands". The four skips are grater and peeler facets that no strand tool has a value for.
  - [2026-10-25: `peeler_shape` was dropped from this branch's makes_irrelevant. The earlier chain rested on ATK julienne's "Most models are Y-shaped," which is "most," not "all"; retail sells straight ("I-shaped") and I-plus-Y julienne peelers, so shape stays a live property of the strand tool rather than an irrelevant one. It is not added to `then_ask` either, since the split's own question already names the tool family and nothing in the sources says to ask shape again for it.]
- when `coring`: then_ask corer_type; makes_irrelevant grate_texture, peeler_shape, tooth_type, grater_format, blade_edge, blade_steel.
  - corer_type is defined as "The tool type for removing cores". The skips are the grater and peeler facets, which no corer has a value for. The crank peeler-corer peels, but its blade is fixed in the machine (ATK apple corers), so peeler shape does not apply. [2026-10-25: tooth_type, grater_format, blade_edge and blade_steel added; these were left live by omission in the pre-check register.]

**Split 2, grate_texture ("Fluffy zest and Parmesan, thick cheddar shreds, or both?")**

- when `fluffy`: then_ask tooth_type, grater_format.
  - tooth_type: ATK found stamped fine holes turn Parmesan to powder and trap zest, while etched teeth shave it into fluffy wisps (grate_texture, why experts care, restated 2026-10-25 as ATK's finding rather than "testers" generally); Reviewed is the contested source on the soft-cheese half of this, not the zest/Parmesan half.
  - grater_format: fluffy work is done by rasp zesters, rasp paddles and rotary drums, which differ in grip and safety ("Narrow rasp zesters are nimble on citrus; wide paddles grate more per stroke ... Rotary graters keep fingers away from the teeth", grater_format, why experts care).
  - Removed in self-check: a `makes_irrelevant grating_surface_area` leg. Surface area matters for rasps too (ATK's rasp floor, Serious Eats on width), so the leg could not be traced.
  - [2026-10-25: multi-drum rotary graters are not automatically fluffy; `grate_texture`'s backfill_note now routes titles naming several drums, or sold as a mandoline, to thick shreds instead (27 of 45 rotary titles on the shelf per the blind check).]
- when `thick shreds`: then_ask tooth_type, grating_surface_area.
  - tooth_type: "high stamped teeth shred it" (grate_texture) and ATK "In general, we preferred graters with stamped teeth" for shredding. Contested by Reviewed's etched box graters on mozzarella and potato (new judgement, 2026-10-25); not contested for Parmesan/zest.
  - grating_surface_area: "A bigger surface grates more per stroke, so it is faster" (grating_surface_area, why experts care); ATK's 15 sq in floor is for this format, single-sourced, so it stays out of the JSON as a hard number (backfill_note corrected 2026-10-25 to say so plainly).
- when `both`: then_ask tooth_type, blade_style.
  - tooth_type: on a box, the fine side only works if etched (Reviewed's stamped Cuisinart fine side "turned Parmesan into a powder"; ATK "Box graters with etched holes do have one advantage").
  - blade_style: "It tells the buyer what the grater will actually produce" (blade_style, why experts care); a box carries several.
  - [2026-10-25: `grater_format` dropped from this branch's then_ask. The split rationale already states box graters are the "both" answer, so asking format again only repeats that answer (blind check, WEAK). No source gives a further format question for the "both" answer, so nothing replaces it.]

**Split 3, peeler_shape**: no branches. Neither answer changes what else to ask; blade edge and steel matter for both shapes. [2026-10-25: `skip_if_known` and the facet's `listing_field_hint` now also cover "I peeler", "I and Y", "Euro peeler" and "Swiss peeler" — retail words the blind check found unmapped (KitchenAid's Euro Peeler is straight; the Kuhn Rikon Original Swiss Peeler is a Y peeler despite the name).]

Renderer note: the flowchart draws split 2's legs continuing to question 3 (peeler shape) even though split 1's grating leg marks peeler shape irrelevant, because tree.py's `next_question` only consults the current branch's `makes_irrelevant`. The timeline and mindmap are unaffected. The scripts are out of scope, so this is recorded, not fixed.

## Trimmed detail

Nothing hit a validator cap on the first validation; every prose field was written to its cap. Facts kept out of the JSON because they had one source are in the tables above. Wording changes made during the self-check that removed a fact:

- grating_surface_area.why_experts_care: "Serious Eats liked rasps 1 to 1.5 inches wide" became "Serious Eats preferred wider rasps" (single source).
- strand_tool.why_experts_care: "usually comes with three blades" became "usually with interchangeable blades" (Wikipedia only).
- corer_type.why_experts_care: "8 chunky or 16 slim slices" and "in about 15 seconds" removed (ATK only).
- tradeoffs: "take up to three times as long" and "fills after about half a cup" removed (ATK only).
- judgement 2 note: "of 12" removed from "among its worst".
- peeler_shape.why_experts_care: "and squash" removed from the Y-peeler bite sentence (Alton Brown named eggplant only).
- grater_format.why_experts_care: "Ginger graters added nothing over a good fine rasp" restated as "no dedicated ginger grater beat a fine rasp paddle" in ATK's test, which is what the source says.

## Corrections after the blind check (2026-10-25)

Re-verified against the live pages with WebFetch (no WebSearch budget was needed; all facts below were already quoted in check.md from pages already cited in this register, except the pineapple-corer source, which was newly found): the ATK ginger model name, and the four reference dates.

1. Restated the etched/stamped mozzarella claim as ATK's finding, contested by Reviewed's two etched box graters (best mozzarella and potato shreds): `grate_texture.why_experts_care`, split 2's `rationale`, trade-off 1's `plain`, and added a new judgement recording the disagreement. Removed `reviewed_graters` from trade-off 1's `evidence`, since Reviewed does not support the clogging claim it was cited for.
2. Logged the mandoline, push-down chopper, and pitter/huller/avocado-slicer scope orphans honestly in notes.md's scope paragraph: no leaf currently owns them, including this one. Rejected building a facet for them here: they are explicitly out of this leaf's scope by design, and the actual fix (claiming them, or routing them correctly) belongs to `knives`, `mashers` or `kitchen-other`, which this task may not edit.
3. Resolved the `tool_job` rule-by-use vs rule-by-product conflict for zesters: `definition` and `backfill_note` now agree that a plain zester grates and a channel knife (alone or combined with a zester) adds peeling. Removed the bare word "zester" from `skip_if_known`, since it does not disambiguate a rasp zester from a loop zester or channel-knife combo; added "zesting lemons for baking" to make the surviving task-based skip explicit about the job it implies. Left as a documented, unfixed gap (notes.md, Branches): the `peeling` branch's three asks (peeler_shape, blade_edge, blade_steel) still describe channel knives and combos poorly, since the schema has no way to branch within one split value and no sourced sub-facet was built for it.
4. Added a `drum (rotary)` value to `tooth_type` (evidence: `atk_rotary`, which calls it "a rotating drum studded with sharp teeth") and a rotary rule to `grate_texture.backfill_note`, so multi-drum rotary slicers default to thick shreds, not fluffy.
5. Removed the unsourced "absence of the word on a coarse grater usually means stamped" rule from `tooth_type.listing_field_hint` (it mislabels the etched Cuisipro box, which Reviewed documents as etched with no such word on its listing). Raised `tooth_type.backfill_cost` from medium to high to reflect that the keyword rule now only catches explicit "etched" mentions.
6. Changed `blade_style.source` from `listing` to `derived` and `backfill_cost` from low to medium (only 2 of 12 cited box titles name hole sizes); added a normalisation rule for the Microplane box's "Extra Coarse" (title) vs "ultra-coarse" (bullet) naming the same blade.
7. Raised `blade_edge.backfill_cost` from none to low; replaced "absence of the word means smooth" with a rule that also catches tomato/kiwi peelers and notes that multi-blade sets carry both values.
8. Dropped `peeler_shape` from the `veg noodles or matchsticks` branch's `makes_irrelevant` (WEAK: ATK says "most," not all, julienne peelers are Y-shaped, and retail sells straight ones). Reworked the `both` branch's `then_ask` by dropping `grater_format`, since the split rationale already states box graters are the "both" answer and asking format again only repeats it; no sourced replacement question was found, so nothing was added in its place.
9. Added `blade_steel` to the `grating` branch's `makes_irrelevant` (it is defined as a peeler facet) and `tooth_type`, `grater_format`, `blade_edge`, `blade_steel` to both the `veg noodles or matchsticks` and `coring` branches' `makes_irrelevant` (none has a defined value for a strand tool or corer).
10. Softened judgement 3 (box-grater sides): dropped "independently" and confidence high → medium, since ATK's own wording is "many of us" (staff), not a general claim about buyers; Reviewed is the source for "most people." Mirrored the wording in `blade_style.why_experts_care`.
11. Restated the US food-contact note (judgement) with 21 U.S.C. 321(s)'s actual conditions (expected migration into food, not GRAS or prior-sanctioned) instead of dropping them, and added EU Art. 15(2)'s exemption for obviously food-contact articles, with a matching `supports` line on the `eu_1935_2004` reference.
12. Sourced the pineapple corer value: added a new reference, `wiki_pineapple_cutter` (Wikipedia, "Pineapple cutter" — re-fetched 2026-10-25, confirmed live), to `corer_type.evidence`, and added one sentence to `corer_type.why_experts_care` on how it works.
13. Corrected `peeler_shape.why_experts_care`'s "nearly every peeler" (contradicted by Wikipedia's fixed-blade Lancashire/French Econome peelers) to "every peeler ATK tested."
14. Corrected `grating_surface_area.backfill_note`: ATK's box-and-paddle review does carry the surface-area field (the earlier note said it did not); named ATK's single-source minimums by format instead of claiming a "top 50" that reviewer tables cannot reach, and kept them out of the JSON as hard numbers since each rests on one source only.
15. Added "I peeler", "I and Y", "Euro peeler" and "Swiss peeler" to `peeler_shape`'s split `skip_if_known` and the facet's `listing_field_hint`.
16. Minor fixes: judgement 4's note now states ATK's own self-contradiction on carbon-steel dishwashing (its spec row vs its body text), not just a disagreement with Amazon. Judgement 10's claim dropped the unsourced "main hazard" framing; its note clarifies atk_gloves lists a skinned knuckle as one of several injuries, not the worst. Re-fetched and corrected notes.md's ATK ginger quote to name the Microplane Gourmet Series Fine Grater (the page no longer names a "Home Series Fine Grater"). Added publication dates to four references after re-fetching each page: `atk_julienne` (9 July 2025), `atk_gloves` (19 August 2019), `se_rasp` (9 June 2022), `reviewed_graters` (21 April 2022).

Rejected / not actioned, with reasons:
- The Microplane catalogue's "2026" vs "2023" footer discrepancy (Part A row 2 of check.md): flagged there as a "small error," not a numbered required fix. Left as is; nothing in the fetched PDF resolves which footer is current, and relabelling the reference risks guessing.
- The Zyliss NSF certification lead (check.md, "Other sourcing findings"): the checker itself says "this is a lead, not a fix" because the applicable NSF/ANSI standard could not be verified without search budget. Not added; adding an unverified standard claim would violate the house rule against invented claims.

## Corrections

1. `vertical` is "other", not "supplies". The brief asked for "supplies", but schema.json's enum is [formulations, machines, fashion, furnishing, other], and validate.py rejected "supplies" ("'supplies' is not one of [...]"). The scripts and schema are out of scope, so the register uses "other" to validate clean. Every Supplies leaf running concurrently will hit the same error; the schema needs "supplies" added, then this field can change back.
2. US food-contact citation moved from 21 U.S.C. 348(h)(6) to 21 U.S.C. 321(s): the FCS definition lists "manufacturing, packing, packaging, transporting, or holding" and not preparing; 321(s) includes "preparing".
3. Five single-source numbers removed from prose (listed under Trimmed detail).
4. Branch skip lists shortened to the facets that would otherwise be asked (the later split questions and their own family's facets); a fluffy-branch leg was removed as untraceable.
5. Validator, renderer and viewer all ran clean after these changes. The skill's Playwright check (`tests/viewer-browser.py`) passed on a scratchpad copy at 1380, 768 and 390 px in light and dark.