Three small tools.
Every field and code.
The reference for audit_size_chart, normalise_size_chart and export_agent_ready_chart: what each one takes, what it returns, what each problem code means, and the limits. Every tool example uses a fictional chart, and outputs marked “Output from the live server” were captured on 17 Nov 2026. The refusal replies are shown as the server code writes them.
Audit, normalize, export.
In that order.
The usual order is audit, then normalize, then export. Each call works only on what you send in it. The server builds a fresh MCP server for every request and keeps nothing between calls, so send the chart again each time.
audit_size_chart"Audit a size chart." Reports what is wrong with a chart, or still unanswered, without changing it.
normalise_size_chart"Normalise a size chart." Turns a table, or a chart with gaps, into SizeRef chart format v1 from what the table states plus your answers. Open facts come back as questions with allowed answers.
export_agent_ready_chart"Export a size chart." Writes a valid chart v1 as SizeRef JSON, schema.org JSON-LD, CSV and plain text, and lists anything left out and why.
Takes a size chart, never a person's measurements. Do not send body measurements, customer records or other personal information. See the pilot terms and the privacy details.
Tool annotations
All three tools carry the same MCP annotations.
| Annotation | Value | What it means here |
|---|---|---|
readOnlyHint | true | The tool changes nothing. It has nothing to change: no database, no store connection. |
destructiveHint | false | Nothing is deleted or overwritten. |
idempotentHint | true | The same input gives the same answer. |
openWorldHint | false | The tool reaches nothing outside the call. The server calls no AI model and makes no outbound requests. |
What every answer holds
- The result comes twice: as
structuredContent, and as the same JSON in a text item ofcontent. - A refusal or an error sets
isError: true. Questions from normalise_size_chart are a normal answer, not an error. - Every answer from a tool carries
"note": "SizeRef stored nothing from this request."A call refused before the tool runs gets a short refusal without it. - Inputs are strict: a top-level field that is not listed below makes the call fail input validation.
Check a chart.
Change nothing.
Use it before the other two. It reads a table or a chart and says what is wrong or still open. It cannot see the product page, so a pasted table is at best needs_fix until normalise_size_chart gets the answers: category and children are always asked for a table.
Inputs
| Field | Type | Notes |
|---|---|---|
chart | object | A chart in SizeRef chart format v1. Send chart or table, not both. Each rows cell is a number or a list of two numbers. |
table | object | One form only: html (text), csv (text, the top row is the header), or headers (list of text) with rows (list of lists of text). Optional caption (text) is read for unit, kind and flat. With html, it replaces the caption of every table found. |
variant_sizes | list of text | Optional. The size options the product sells, such as ["S","M","L","XXL"]. Up to 60 labels, each up to 64 characters. |
product_title | text | Optional. Read for a chart input: on a Tops chart, hips stay a sizing area only for long tops such as a kurta, tunic or dress. |
product_text | text | Optional. Read for a table input, only for a fit word. If it names one fit and no other, that fit is used. |
How a table is read
- Every table in
htmlis read as text, with colspan and rowspan expanded. Up to 20 tables. - The size column is headed Size (or a label such as US or UK size), or is the leftmost column with a blank header.
- Measurement columns are found by header words only: chest, waist, hips, inseam, sleeve, shoulders and length. Any other column is listed in
ignored_columnsand left out. - A cell must be a number or a range, such as
38,38.5,38 1/2or38-40. Anything else is aCELLproblem, never dropped silently. - Unit, kind and flat come only from plain statements in the caption or headers, such as "in inches", "(cm)", "garment", "to fit" or "pit to pit". When the text states neither unit or both, or neither kind or both, that becomes a question. "flat" or "pit to pit" means flat; "circumference" or "around" means all the way round.
Output
| Field | What it holds |
|---|---|
verdict | The worst verdict among the tables: ok, needs_fix or unusable. Read results[] for each table. |
results[] | One entry per table found (a chart gives one): index, caption, verdict, problems[], questions[], ignored_columns[] and, when variant_sizes was sent and the labels differ, label_mismatch. |
problems[] | Each with severity, code, sometimes size and area, a plain text and a fix. See problem codes. |
questions[] | Each { key, allowed }. See questions and answers. |
ignored_columns[] | Each { col, header }: columns that are not measurements. |
label_mismatch | chart_only, variants_only and suggested_aliases. Suggestions come only from a fixed list of labels that mean the same (2XL and XXL, 3XL and XXXL, 4XL and XXXXL, XS and X-Small, S and Small, M and Medium, L and Large, XL and X-Large). They are never applied. |
note | "SizeRef stored nothing from this request." |
| Verdict | Meaning |
|---|---|
ok | No errors and nothing left open. Warnings about unusual values are allowed. |
needs_fix | Some error, or some fact still open. |
unusable | No sizes, no measurements, a bra band-and-cup chart, no chest, waist or hips for every size, or the input was refused. |
Only a refused input sets isError on an audit. A chart that is merely bad is a normal answer with a verdict.
Example
A fictional tee chart with one bad cell, a color column and a 2XL label the product calls XXL.
{
"table": {
"caption": "Fictional Example Tee, garment measurements in inches, measured flat",
"headers": ["Size", "Chest", "Length", "Colour"],
"rows": [
["S", "20", "27", "Grey"],
["M", "21", "n/a", "Grey"],
["L", "22", "29", "Grey"],
["2XL", "24", "30", "Grey"]
]
},
"variant_sizes": ["S", "M", "L", "XXL"]
}{
"verdict": "needs_fix",
"results": [
{
"index": 0,
"caption": "Fictional Example Tee, garment measurements in inches, measured flat",
"verdict": "needs_fix",
"problems": [
{
"severity": "error",
"code": "CELL",
"size": "M",
"area": "length",
"text": "length for size M is not a number or a range.",
"fix": "Write the cell as a number such as 38, a range such as 38-40, or a fraction such as 38 1/2."
}
],
"questions": [
{
"key": "category",
"allowed": ["Tops", "Bottoms", "Leggings", "Outerwear", "Dresses", "Other"]
},
{
"key": "fit",
"allowed": ["compression", "bodyfit", "fitted", "regular", "relaxed", "oversized"]
},
{ "key": "stretch", "allowed": ["low", "medium", "high"] },
{ "key": "children", "allowed": [true, false] }
],
"ignored_columns": [
{ "col": 3, "header": "Colour" }
],
"label_mismatch": {
"chart_only": ["2XL"],
"variants_only": ["XXL"],
"suggested_aliases": { "2XL": "XXL" }
}
}
],
"note": "SizeRef stored nothing from this request."
}The caption states the unit, the kind and that it was measured flat, so those are not asked. Category and children are always asked for a table; fit and stretch are asked because it is a garment chart.
Stated facts only.
Ask for the rest.
Send the same table (or a chart with gaps) plus answers to the open questions. Nothing is inferred from the numbers. A value the table does not state comes only from answers.
Inputs
| Field | Type | Notes |
|---|---|---|
chart | object | A chart in format v1, perhaps with gaps. Send chart or table, not both. Fields that are not chart v1 fields, such as source, are dropped. |
table | object | As for audit: html, csv, or headers with rows, plus optional caption. |
table_index | whole number, 0 or more | Needed when the input holds more than one table. Without it the call returns MANY_TABLES. |
answers | object | Question keys with allowed answers, plus optional aliases ({ "2XL": "XXL" }). See questions and answers. |
product_title | text | Optional. On a Tops chart, hips stay a sizing area only for long tops such as a kurta or a dress. |
product_text | text | Optional. Gives the fit when the table does not state it. |
Output: one of three shapes
| Shape | Fields | isError |
|---|---|---|
| A chart | ok: true, chart_v1, sizing_chart_cm (the areas used for sizing, in cm, each as [min, max]; a flat chest, waist or hips is doubled), kept (extra columns, or null), warnings when there are any, note | no |
| Questions | ok: false, questions[], note | no |
| Problems | ok: false, problems[], note. No partial chart is returned. | yes |
Rules
- Facts the table already states win. An answer that differs is refused with
BAD_ANSWER, never applied. - An answer for a key that is not a question, or a value not on its list, is also
BAD_ANSWER. - For a body chart,
flatis set to false without asking. - Labels read from the table and
answers.aliasesmeet the same size caps as a chart you send.
Example
The same fictional tee, with the bad cell fixed and the color column removed. One call with no answers:
{
"table": {
"caption": "Fictional Example Tee, garment measurements in inches, measured flat",
"headers": ["Size", "Chest", "Length"],
"rows": [
["S", "20", "27"],
["M", "21", "28"],
["L", "22", "29"],
["2XL", "24", "30"]
]
}
}{
"ok": false,
"questions": [
{
"key": "category",
"allowed": ["Tops", "Bottoms", "Leggings", "Outerwear", "Dresses", "Other"]
},
{
"key": "fit",
"allowed": ["compression", "bodyfit", "fitted", "regular", "relaxed", "oversized"]
},
{ "key": "stretch", "allowed": ["low", "medium", "high"] },
{ "key": "children", "allowed": [true, false] }
],
"note": "SizeRef stored nothing from this request."
}The next call, with the answers and an alias for the 2XL label:
{
"table": { … the same fictional table … },
"answers": {
"category": "Tops",
"fit": "regular",
"stretch": "low",
"children": false,
"aliases": { "2XL": "XXL" }
}
}{
"ok": true,
"chart_v1": {
"v": 1,
"sizes": ["S", "M", "L", "2XL"],
"rows": {
"S": { "chest": 20, "length": 27 },
"M": { "chest": 21, "length": 28 },
"L": { "chest": 22, "length": 29 },
"2XL": { "chest": 24, "length": 30 }
},
"kind": "garment",
"unit": "in",
"flat": true,
"category": "Tops",
"fit": "regular",
"stretch": "low",
"children": false,
"aliases": { "2XL": "XXL" }
},
"sizing_chart_cm": {
"type": "garment",
"areas": ["chest"],
"sizes": ["S", "M", "L", "2XL"],
"rows": {
"S": { "chest": [101.6, 101.6] },
"M": { "chest": [106.68, 106.68] },
"L": { "chest": [111.76, 111.76] },
"2XL": { "chest": [121.92, 121.92] }
}
},
"kept": null,
"note": "SizeRef stored nothing from this request."
}The chest was measured flat, so sizing_chart_cm holds the doubled chest in cm. Length is carried in the chart but is not a sizing area.
An answer that contradicts the caption is refused:
{
"table": { … the same fictional table … },
"answers": {
"unit": "cm",
"category": "Tops",
"fit": "regular",
"stretch": "low",
"children": false
}
}{
"ok": false,
"problems": [
{
"severity": "error",
"code": "BAD_ANSWER",
"text": "The chart already states unit \"in\", so the answer \"cm\" was not applied.",
"fix": "Answer only what the chart leaves open, using one of the allowed answers."
}
],
"note": "SizeRef stored nothing from this request."
}Every question.
Every allowed answer.
When a table does not state a fact, the tools ask for it instead of guessing. Each question is { key, allowed }. Put the answers in answers on the next normalise_size_chart call. Each answer must be one of the allowed values, with the same type: true, not "true".
| Key | What it asks | Allowed answers | When it is asked (table input) |
|---|---|---|---|
kind | Do the numbers measure a body or the garment? | "body", "garment" | When the headers and caption do not say |
sizeBy | The table has body and garment columns: size by which? | "body", "garment" | When both kinds are present |
flat | Was the garment measured laid flat, or all the way round? | true, false | Once kind is garment, flat is not stated, and there is a chest, waist or hips column |
unit | Inches or centimeters? | "in", "cm" | When the unit is not stated plainly |
category | The product category | "Tops", "Bottoms", "Leggings", "Outerwear", "Dresses", "Other" | Always |
fit | The cut's fit | "compression", "bodyfit", "fitted", "regular", "relaxed", "oversized" | Once kind is garment and no fit was found |
stretch | Fabric stretch | "low", "medium", "high" | Whenever kind is garment |
children | Is this a children's chart? | true, false | Always |
- For a table, questions come back in the order above.
- Answering
kindwith"garment"can bringflat,fitandstretchinto the reply. - One more key can go in
answers:aliases, an object from a chart size to the product's option, such as{ "2XL": "XXL" }. Aliases are shown and exported, never used for sizing. - For a chart input (not a table), only the missing required fields are asked:
kind,unit,flat,categoryandchildren.
One chart.
Four formats.
Send a complete, valid chart v1, usually the chart_v1 from normalise_size_chart. A chart with a missing field or an error is refused; nothing is exported from it.
Inputs
| Field | Type | Notes |
|---|---|---|
chart_v1 | object | Required. A complete chart in format v1. |
formats | list | Optional. Any of sizeref_json, jsonld, csv, text. All four when left out. An empty list is refused. |
product | object | Optional. name and brand (text, up to 255 characters each) and url (starts with http:// or https://, up to 2,048 characters). |
Output
| Field | What it holds |
|---|---|
sizeref_json | The chart in format v1, with $schema at the top. Fields that are not chart v1 fields are dropped. |
jsonld | A whole <script type="application/ld+json"> block for the product page: one schema.org SizeSpecification per size, and no Product node. |
jsonld_note | Says the theme can link each variant's size to the matching @id. That step is optional. |
csv | One row per size, columns such as chest (in). A flat chest, waist or hips is doubled; ranges are written 38-40. |
text | One readable line per size, with the flat number after a doubled one. |
omitted[] | Each { area, reason }: a measurement left out of the JSON-LD because schema.org has no type for it. It still appears in the CSV and text. |
warnings | Only when a value is outside the usual range (IMPLAUSIBLE). |
note | "SizeRef stored nothing from this request." |
How the JSON-LD is built
- Each size gets an
@id:<url>#size-<size>(any # part of the URL is dropped and the size is URL-encoded) with a product URL, orurn:sizeref:<product name as a slug>:<size>without one (chartwhen there is no name either). nameis the brand and product name with the size, or the size alone. An alias becomesalternateName.- Garment charts use
hasMeasurement; body charts usesuggestedMeasurement. Each value has aunitCode(INHfor inches,CMTfor cm) and a schema.org measurement type invalueReference. - A range is written as
minValueandmaxValue; a single number asvalue. - A flat garment's chest, waist and hips are written as the full measurement, with a description saying so.
- Kept columns of the other kind are written with a
namesuch as "body chest" instead of a measurement type. <,>,&and U+2028 and U+2029 are written as\uescapes, so no size label or product name can end the script block.
| Measurement | Garment chart | Body chart |
|---|---|---|
chest | WearableMeasurementChestOrBust | BodyMeasurementChest |
waist | WearableMeasurementWaist | BodyMeasurementWaist |
hips | WearableMeasurementHips | BodyMeasurementHips |
inseam | WearableMeasurementInseam | BodyMeasurementInsideLeg |
sleeve | WearableMeasurementSleeve | left out, listed in omitted |
length | WearableMeasurementLength | left out, listed in omitted |
shoulders | left out, listed in omitted | left out, listed in omitted |
CSV fields that a spreadsheet could read as a formula (starting with =, +, -, @, a tab or a carriage return) get a leading apostrophe. With kept columns, every CSV and text column is named with its kind, such as "garment chest".
Example
The fictional chart from the normalize example, for a fictional product on a reserved example domain.
{
"chart_v1": {
"v": 1,
"sizes": ["S", "M", "L", "2XL"],
"rows": {
"S": { "chest": 20, "length": 27 },
"M": { "chest": 21, "length": 28 },
"L": { "chest": 22, "length": 29 },
"2XL": { "chest": 24, "length": 30 }
},
"kind": "garment",
"unit": "in",
"flat": true,
"category": "Tops",
"fit": "regular",
"stretch": "low",
"children": false,
"aliases": { "2XL": "XXL" }
},
"product": {
"name": "Example Tee",
"brand": "Fictional Co",
"url": "https://shop.example.com/products/example-tee"
}
}{
"omitted": [],
"note": "SizeRef stored nothing from this request.",
"sizeref_json": { "$schema": "https://mcp.sizeref.online/schema/chart-v1.json", "v": 1, … the other chart v1 fields, in schema order … },
"jsonld": "<script type=\"application/ld+json\">\n{ … }\n</script>", (shown below, trimmed)
"jsonld_note": "Optional: the theme can link each variant's size to the matching @id in this block. This step is not required.",
"csv": "size,chest (in),length (in)\nS,40,27\nM,42,28\nL,44,29\n2XL,48,30",
"text": "Example Tee: size chart, garment measurements in inches\nS: chest 40 in (20 in flat), length 27 in\nM: chest 42 in (21 in flat), length 28 in\nL: chest 44 in (22 in flat), length 29 in\n2XL (product option XXL): chest 48 in (24 in flat), length 30 in"
}<script type="application/ld+json">
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "SizeSpecification",
"@id": "https://shop.example.com/products/example-tee#size-S",
"name": "Fictional Co Example Tee - S",
"hasMeasurement": [
{
"@type": "QuantitativeValue",
"value": 40,
"unitCode": "INH",
"valueReference": "https://schema.org/WearableMeasurementChestOrBust",
"description": "chest measured flat, shown as full circumference"
},
{
"@type": "QuantitativeValue",
"value": 27,
"unitCode": "INH",
"valueReference": "https://schema.org/WearableMeasurementLength"
}
],
"description": "Fictional Co Example Tee, size S. chest measured flat, shown as full circumference."
},
… the M and L entries, built the same way, are left out here …
{
"@type": "SizeSpecification",
"@id": "https://shop.example.com/products/example-tee#size-2XL",
"name": "Fictional Co Example Tee - 2XL",
"alternateName": "XXL",
"hasMeasurement": [ … chest 48 and length 30, as above … ],
"description": "Fictional Co Example Tee, size 2XL. chest measured flat, shown as full circumference."
}
]
}
</script>A fictional body chart with a shoulders column shows omitted:
{
"chart_v1": {
"v": 1,
"unit": "cm",
"kind": "body",
"flat": false,
"children": false,
"category": "Tops",
"sizes": ["S", "M"],
"rows": {
"S": { "chest": 92, "shoulders": 42 },
"M": { "chest": 98, "shoulders": 44 }
}
},
"formats": ["jsonld"]
}{
"omitted": [
{
"area": "shoulders",
"reason": "schema.org has no body measurement type for shoulders."
}
],
"note": "SizeRef stored nothing from this request."
}Where a brand can put each format: the jsonld block in the product page's HTML, the CSV in a spreadsheet, the text wherever a plain chart is needed, and the SizeRef JSON in tools that read format v1. More in chart format.
The chart format,
as a resource.
The server offers one MCP resource: the JSON Schema of SizeRef chart format v1.
- URI
sizeref://schema/chart-v1- Title
- SizeRef size chart format v1
- MIME type
application/schema+json- Also on the web
- https://mcp.sizeref.online/schema/chart-v1.json, which is the schema's
$idand the$schemavalue insizeref_json.
The fields, required ones and caps are explained on chart format.
What each code means.
What to do next.
Problems sit in problems[]. Audit lists warnings there too; normalise_size_chart and export_agent_ready_chart list them in warnings. Each one says what is wrong and how to fix it.
{
"severity": "error",
"code": "CELL",
"size": "M",
"area": "length",
"text": "length for size M is not a number or a range.",
"fix": "Write the cell as a number such as 38, a range such as 38-40, or a fraction such as 38 1/2."
}- severity
error, orwarningforIMPLAUSIBLE.- code
- One of the codes below.
- size, area
- Present when the problem is about one size or one measurement. A size label read from a table is cut to 40 characters, followed by "…".
- text, fix
- What is wrong, in plain words, and how to fix it.
Refusals
| Code | Tools | Meaning | Effect |
|---|---|---|---|
TOO_LARGE | all | Over a size cap: sizes, columns, measurements a size, names, aliases, tables, cells, HTML or CSV length, or a product field. | Refused. Audit verdict unusable |
BAD_INPUT | all | Malformed: neither or both of chart and table, a table not in exactly one form, html or csv not text, table_index out of range, an empty formats list, or a product url that does not start with http:// or https://. Wrong types and unknown formats usually fail input validation first. | Refused. Audit verdict unusable |
NOT_A_CHART | audit, normalize | The call's own fields include one named for a person (body, measurements, person, user, shopper). The call is refused. Because these are not tool inputs, the server's input validation usually refuses it first, so you may see an input validation error instead of this code. Nothing looks inside a chart for such fields, so do not send a person's measurements. | Refused |
NO_TABLE | audit, normalize | No table was found in the HTML, or the CSV is empty. | Refused. Audit verdict unusable |
NOT_V1 | all | The chart is not a JSON object, or v is not 1. | Refused. Audit verdict unusable |
MANY_TABLES | normalize | The input holds several tables. Run audit to see them, then send table_index. | Error |
BAD_ANSWER | normalize | An answer is for something that is not a question, is not an allowed value, or contradicts what the table states. aliases is not an object. | Error, not applied |
MISSING | export | A required field (area names it) is not stated, so the chart cannot be exported. | Error |
Chart and table problems
| Code | Meaning | Audit verdict |
|---|---|---|
NO_SIZES | No size column or no size labels. | unusable |
NO_MEASURES | No chest, waist, hips, inseam, sleeve, shoulders or length column. | unusable |
BRA | Bra band-and-cup sizes, which SizeRef does not size. | unusable |
NO_SIZING_AREAS | No chest, waist or hips is given for every size. Length and sleeve alone cannot pick a size. | unusable |
CELL | A cell is not a number or a range (in a chart: not a positive number or a [min, max] pair with min at most max). In a table, a blank or repeated size label is also CELL. | needs_fix |
FALLING | A measurement falls from one size to the next. It should rise or stay level. In a table every measurement column is checked; in a chart, the chest, waist or hips used for sizing. | needs_fix |
UNIT, KIND, FLAT, CHILDREN, CATEGORY, FIT, STRETCH | A chart field holds a value that is not on its list. | needs_fix |
ROW_MISSING | No rows object, or no row for a listed size. | needs_fix |
BLANK_SIZE, DUP_SIZE | A chart's size label is blank or not text, or appears more than once. | needs_fix |
BAD_ALIASES | aliases does not map each chart size to a different, non-blank option. | needs_fix |
BAD_KEPT | kept is not a valid set of extra columns. | needs_fix |
IMPLAUSIBLE | Warning only: a value is outside a wide range SizeRef set for that measurement, in cm. Check the unit and the digits. | no change |
Service problems
| Code | Meaning | What to do |
|---|---|---|
RATE_LIMITED | Too many calls to this tool just now, counted for all users together. | Wait a minute, then call again. |
INTERNAL | The tool failed on this input. | Send a smaller or simpler chart. |
If normalise_size_chart cannot turn an otherwise valid chart into a sizing chart, it returns that step's code (such as BAD_CHART) with the advice to run audit_size_chart on it.
Two kinds of no.
Both say why.
A tool's own refusal is a normal tool answer with isError: true, its problems[] (for audit_size_chart, inside results[] with verdict unusable) and the note. A request stopped before any tool runs gets a short HTTP reply instead, without the note. No refusal repeats the request.
Refused before a tool runs
| Situation | Reply |
|---|---|
| Request body over 262,144 bytes | HTTP 413, {"error":"TOO_LARGE",…} |
| Several calls in one request (a JSON-RPC batch) | HTTP 400, JSON-RPC error -32600 |
| Too many requests from one address | HTTP 429, JSON-RPC error -32000, Retry-After: 60 |
| A browser call from another website, or a wrong host | HTTP 403 |
| Fields that do not match the tool's inputs | An input validation error from the MCP layer |
{ "error": "TOO_LARGE", "text": "Send at most 262144 bytes in one request." }{
"jsonrpc": "2.0",
"id": null,
"error": {
"code": -32600,
"message": "Send one JSON-RPC message per request; batches are not supported."
}
}{
"jsonrpc": "2.0",
"id": null,
"error": {
"code": -32000,
"message": "Too many requests from this address just now. Wait a minute, then call again."
}
}Refused by a tool
{
"problems": [
{
"severity": "error",
"code": "RATE_LIMITED",
"text": "Too many requests just now.",
"fix": "Wait a minute, then call again."
}
],
"note": "SizeRef stored nothing from this request."
}Sized for size charts.
Not for more.
Anything over a size cap is refused with TOO_LARGE or an HTTP reply; the rate limits return HTTP 429 or RATE_LIMITED. The exception is cell text, which is cut, not refused. Pilot limits may change; the terms cover fair use.
| Limit | Value |
|---|---|
| One request | 262,144 bytes, and one JSON-RPC message (batches refused) |
| HTML or CSV in a table | 262,144 characters |
| Tables in one input | 20 found; 40 <table> tags as written; 2 tables inside other tables |
| Table cells | 50,000 after colspan and rowspan; 20,000 <td> and <th> tags as written |
| Columns in a table | 40 |
| Text read from one cell, header or caption | 200 characters (the rest is cut) |
| Sizes in a chart or table | 60 |
| Measurements a size; kept columns | 40 each |
| Aliases | 60 |
| Size label, measurement name, alias or option | 64 characters |
variant_sizes | 60 labels of up to 64 characters |
| Product name and brand (export) | 255 characters each |
| Product URL (export) | 2,048 characters |
| Requests from one internet address | 120 a minute (an IPv6 address counts by its /64 network) |
| Calls to one tool, all users together | 600 a minute |
Rate counts are kept by Cloudflare's rate limiter, separately at each Cloudflare location, and are approximate. The address is used only for that count; SizeRef never logs or stores it. Read the privacy details for what is logged.