Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Browse Axad's library of lessons and practice exams. With no lang, list the languages Axad teaches. With lang, list the sections and lessons at path in that language's tree (an empty path is its top). Use it when the learner wants something to study, or names a goal such as an exam: open a lesson with open_lesson.
BCP 47 code of the language to learn: "nl".
Section names from the top of the language's tree: ["Inburgering", "A2"].
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http axad https://mcp.axad.ai/mcp
Browse Axad's library of lessons and practice exams. With no lang, list the languages Axad teaches. With lang, list the sections and lessons at path in that language's tree (an empty path is its top). Use it when the learner wants something to study, or names a goal such as an exam: open a lesson with open_lesson.
| Argument | Type | Description |
|---|---|---|
| lang | string | BCP 47 code of the language to learn: "nl". |
| path | array | Section names from the top of the language's tree: ["Inburgering", "A2"]. |
{
"type": "object",
"properties": {
"lang": {
"type": "string",
"description": "BCP 47 code of the language to learn: \"nl\"."
},
"path": {
"type": "array",
"items": {
"type": "string"
},
"description": "Section names from the top of the language's tree: [\"Inburgering\", \"A2\"]."
}
}
}Close a lesson the learner finished or set aside. It keeps its place, chats and cards, and no longer counts against the plan's active lessons; open_lesson with its material id opens it again.
| Argument | Type | Description |
|---|---|---|
| material* | string |
{
"type": "object",
"required": [
"material"
],
"properties": {
"material": {
"type": "string"
}
}
}The learner's cards that are due for review now, the most overdue first, each with an example sentence. Quiz the learner on them (show_quiz, show_examples), then record each answer with review_card.
| Argument | Type | Description |
|---|---|---|
| lang | string | |
| limit | integer |
{
"type": "object",
"properties": {
"lang": {
"type": "string"
},
"limit": {
"type": "integer",
"default": 10,
"maximum": 20,
"minimum": 1
}
}
}Search the shared image library for pictures that show a word, chunk, action or scene. You need not call it before nominate_image: a nomination searches the library itself and answers 'existing' when a picture fits. Use a match only when it truly shows the concept.
| Argument | Type | Description |
|---|---|---|
| lang* | string | BCP 47 code of the concept: pictures with words in another language are left out. |
| query* | string | A literal English description of the picture you want: "a patient in a doctor's waiting room, reading a magazine". |
| concept | string | The language element it should teach, in its language: "la sala de espera". |
{
"type": "object",
"required": [
"query",
"lang"
],
"properties": {
"lang": {
"type": "string",
"description": "BCP 47 code of the concept: pictures with words in another language are left out."
},
"query": {
"type": "string",
"description": "A literal English description of the picture you want: \"a patient in a doctor's waiting room, reading a magazine\"."
},
"concept": {
"type": "string",
"description": "The language element it should teach, in its language: \"la sala de espera\"."
}
}
}The learner's Axad account: their plan, its limits and what they use of them (cards in the bag, active lessons), and their credit. Use it when the learner asks about their plan or credit, or after a limit_reached.
No arguments.
{
"type": "object",
"properties": {}
}List the learner's lessons: the active ones first, then the closed ones. Use it when the learner wants to go on with something they studied.
| Argument | Type | Description |
|---|---|---|
| lang | string | Only lessons in this language. |
{
"type": "object",
"properties": {
"lang": {
"type": "string",
"description": "Only lessons in this language."
}
}
}List Axad's learning paths (daily review, exam path, page by page, free-form chat), with the learner's place in each when signed in.
| Argument | Type | Description |
|---|---|---|
| lang | string |
{
"type": "object",
"properties": {
"lang": {
"type": "string"
}
}
}Read the learner's current step in a learning path they started.
| Argument | Type | Description |
|---|---|---|
| lang* | string | |
| runbook* | string |
{
"type": "object",
"required": [
"runbook",
"lang"
],
"properties": {
"lang": {
"type": "string"
},
"runbook": {
"type": "string"
}
}
}Get a picture for a generic concept of your reply: the server answers 'existing' when the shared library has one that stands for it, else draws it. A picture shows a concept every learner shares: a concrete thing, action, place or everyday scene ("apple", "train platform", "doctor's practice"), never the details of one sentence. Give its `canon`, the generic English concept in 1–4 words, and make the title and prompt generic: no names, clock times, dates or numbers. Put the same `concept` in the `image` field where the picture belongs: widgets show at once and the picture fills in when it comes. Send the nominations first in the message with your widgets, suggest_cards and show_next, one call per picture. The server checks each against its rules and answers what to fix; fix it and nominate again with the same concept. After 4 nominations of one picture (its fixes included), go on without it: the 5th is refused. The server adds the house style: never name a style, medium or artist.
| Argument | Type | Description |
|---|---|---|
| lang* | string | BCP 47 code of the concept. |
| text | string | Only when words in the picture are the lesson: at most 6 words in `lang`, also written in the prompt in double quotes. Omit it for a picture with no words, which serves every language. |
| canon* | string | The generic English concept the picture shows, 1–4 words, dictionary form, with no names, numbers or words of one sentence: "doctor's practice", "apple", "train platform". The library keeps one picture per canon. |
| title* | string | One literal English sentence of 8–40 words: who, what, where. Generic, with no names, numbers, clock times or dates. Search reads it. |
Open a lesson for the learner and read where they are in it: a lesson of Axad's library (lesson), one they have already (material), or a new lesson from a text they give you (text). It counts as one of the plan's active lessons until close_lesson. Returns the current page's text, or the first 40 sentences, and starter suggestions; show the learner what fits, then teach from it with Axad's widgets.
| Argument | Type | Description |
|---|---|---|
| lang | string | With text: BCP 47 code of the text's language. |
| text | string | The learner's own text to study: an article, a letter, a message. |
| title | string | With text: a short title in the text's language. |
| lesson | string | A lesson id from browse_catalog. |
| material | string | A material id from list_lessons or an earlier open_lesson. |
{
"type": "object",
"properties": {
"lang": {
"type": "string",
"description": "With text: BCP 47 code of the text's language."
},
"text": {
"type": "string",
"maxLength": 20000,
"description": "The learner's own text to study: an article, a letter, a message."
},
"title": {
"type": "string",
"description": "With text: a short title in the text's language."
},
"lesson": {
"type": "string",
"description": "A lesson id from browse_catalog."
},
"material": {
"type": "string",
"description": "A material id from list_lessons or an earlier open_lesson."
}
}
}Read a page of a lesson that has pages (a practice exam: one question and its text a page), and keep the learner's place there; or, for a lesson with no pages, read sentences from..to (at most 40).
| Argument | Type | Description |
|---|---|---|
| to | integer | No pages: the last sentence number. |
| from | integer | No pages: the first sentence number. |
| page | integer | 0-based page; omit for the learner's current page. |
| material* | string |
{
"type": "object",
"required": [
"material"
],
"properties": {
"to": {
"type": "integer",
"minimum": 1,
"description": "No pages: the last sentence number."
},
"from": {
"type": "integer",
"minimum": 1,
"description": "No pages: the first sentence number."
},
"page": {
"type": "integer",
"minimum": 0,
"description": "0-based page; omit for the learner's current page."
},
"material": {
"type": "string"
}
}
}Record how the current step of a learning path went, and move the learner to the next step. Returns the next step, or that the path is finished.
| Argument | Type | Description |
|---|---|---|
| lang* | string | |
| note | string | What the learner found hard, in a few words. |
| step* | string | The current step's id. |
| outcome* | string | |
| runbook* | string |
{
"type": "object",
"required": [
"runbook",
"lang",
"step",
"outcome"
],
"properties": {
"lang": {
"type": "string"
},
"note": {
"type": "string",
"description": "What the learner found hard, in a few words."
},
"step": {
"type": "string",
"description": "The current step's id."
},
"outcome": {
"enum": [
"done",
"skipped",
"struggled"
],
"type": "string"
},
"runbook": {
"type": "string"
}
}
}Redeem a gift code the learner was given: its credit goes to their account at once.
| Argument | Type | Description |
|---|---|---|
| code* | string |
{
"type": "object",
"required": [
"code"
],
"properties": {
"code": {
"type": "string"
}
}
}Record how well the learner recalled a card: 1 forgot, 2 hard, 3 good, 4 easy. Axad schedules its next review (FSRS).
| Argument | Type | Description |
|---|---|---|
| card* | string | |
| grade* | integer |
{
"type": "object",
"required": [
"card",
"grade"
],
"properties": {
"card": {
"type": "string"
},
"grade": {
"type": "integer",
"maximum": 4,
"minimum": 1
}
}
}Save flash cards to the learner's bag: what they engaged with and will meet again, at or just above their level. Never a name, a number, a function word, or a card they have (search_bag). At most 6 a call; ask the learner first unless they asked you to save. A card already in the bag gets the new example.
| Argument | Type | Description |
|---|---|---|
| cards* | array | |
| material | string | The lesson the cards come from, when there is one. |
{
"type": "object",
"required": [
"cards"
],
"properties": {
"cards": {
"type": "array",
"items": {
"type": "object",
"required": [
"lang",
"kind",
"key",
"front",
"back",
"example"
],
"properties": {
"key": {
"type": "string",
"description": "The dictionary form, lowercase; a pattern's frame (\"om … te\")."
},
"back": {
"type": "string",
"description": "The meaning, at most 6 words, in the learner's support language."
},
"kind": {
"type": "string",
"description": "word, phrase, pattern, concept, rule, exception, tip."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the card's language."
},
"front": {
"type": "string",
"description": "The card's general form: a word's dictionary form (a noun with de/het where the language needs it), a pattern's frame with … for the open part."
},
"image": {
"type": "string",
"description": "A library picture id, or the concept of a nominate_image call."
},
"example": {
"type": "string",
"description": "The sentence where the learner met it, copied exactly."
}
}
},
"maxItems": 6,
"minItems": 1
},
"material": {
"type": "string",
"description": "The lesson the cards come from, when there is one."
}
}
}Search the learner's flash cards. Use it when the learner asks what they learned, or when you want to build on cards that the context did not list.
| Argument | Type | Description |
|---|---|---|
| lang | string | BCP 47 code: only cards in this language. |
| limit | integer | |
| query* | string | Words in the card's language, or a short meaning: "nunca", "double negatives". |
{
"type": "object",
"required": [
"query"
],
"properties": {
"lang": {
"type": "string",
"description": "BCP 47 code: only cards in this language."
},
"limit": {
"type": "integer",
"default": 10,
"maximum": 30
},
"query": {
"type": "string",
"description": "Words in the card's language, or a short meaning: \"nunca\", \"double negatives\"."
}
}
}Show the alternatives widget. The learner wants other ways to say the same thing, to sound more natural or to find a simpler form. Rules: - 2 to 5 items, each keeping the meaning of `original`. - Put a "simpler" item first when one exists. - Every item must be something a native speaker says. No word-for-word translations. - Mark forms above the learner's level with a tag such as "above A2" (the profile's level).
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| gloss | string | |
| items* | array | |
| original* | string |
{
"type": "object",
"required": [
"original",
"items",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"gloss": {
"type": "string"
},
"items": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"tag"
],
"properties": {
"say": {
"type": "string"
},
"tag": {
"type": "string"
},
"mean": {
"type": "string"
},
"note": {
"type": "string"
}
},
"additionalProperties": false
}
},
"original": {
"type": "string"
}
},
"additionalProperties": false
}Show the contrast widget. Two or three forms look alike or get confused, and the difference is the lesson. Rules: - 2 or 3 sides, the same shape each, so the eye compares row by row. - Examples are minimal pairs where possible: the same sentence with each form. - `key` is a test the learner can apply, not a definition.
| Argument | Type | Description |
|---|---|---|
| key* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| sides* | array |
{
"type": "object",
"required": [
"sides",
"key",
"lang"
],
"properties": {
"key": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"sides": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"mean",
"use",
"example"
],
"properties": {
"say": {
"type": "string"
},
"use": {
"type": "string"
},
"mean": {
"type": "string"
},
"example": {
"type": "object",
"required": [
"say",
"mean"
],
"properties": {
"say": {
"type": "string"
},
"mean": {
"type": "string"
}
},
"additionalProperties": false
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}Show the correction widget. The learner asks to have their own target-language writing checked ("correct me", "check my sentence") or is in a talk they asked for: check it and fix it with the fewest changes. Never for text they pasted or typed to understand. Rules: - Fix errors only. Style goes to `natural`. - Each `why` names the rule in at most 12 words. - When `ok` is true, `fixes` is empty: say what is good in the prose line, in 1 short sentence. - The app draws the word diff from `wrote` and `fixed`; `fixes` give the reasons.
| Argument | Type | Description |
|---|---|---|
| ok* | boolean | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| fixed* | string | |
| fixes* | array | |
| gloss | string | |
| wrote* | string |
Show the custom widget. The learner asks for something no other widget shows well, such as a grid of times or prices, a mini dictionary of one theme, a scene with labels, or a plan of steps. Build it from blocks. Rules: - Use a prebuilt widget when one fits the ask; build a custom widget only for what none of them shows. - Shape it to the learner's own question: their words decide what the blocks hold and in what order. - Every word of a learned language goes in a `Say`: a `say` block, a list item, or a plain table cell. Never put it inside a `text` block or a `note` except in «guillemets». - Every `Say` must be correct and natural, at the learner's level. - Keep it small: at most 12 blocks, tables of at most 6 columns and 12 rows, lists of at most 12 items. - An `image` block names its picture as every `image` field does: the concept of a nominate_image call in this turn, or a library id you have.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| title* | string | |
| blocks* | array |
{
"type": "object",
"required": [
"title",
"blocks",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"title": {
"type": "string"
},
"blocks": {
"type": "array",
"items": {
"anyOf": [
{
"type": "object",
"required": [
"kind",
"say"
],
"properties": {
"say": {
"type": "string"
},
"kind": {
"type": "string",
"const": "say"
},
"means": {
"type": "string"
},
"words": {
"type": "array",
"items": {
"type": "array",
"maxItems": 2,
"minItems": 2,
"prefixItems": [
{
"type": "string"
},
{
"type": "string"
}
]
}
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"text"
],
"properties": {
"kind": {
"type": "string",
"const": "text"
},
"text": {
"type": "string"
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"head",
"rows"
],
"properties": {
"head": {
"type": "array",
"items": {
"type": "string"
}
},
"kind": {
"type": "string",
"const": "table"
},
"rows": {
"type": "array",
"items": {
"type": "array",
"items": {
"anyOf": [
{
"type": "string"
},
{
"type": "object",
"required": [
"note"
],
"properties": {
"note": {
"type": "string"
}
},
"additionalProperties": false
}
]
}
}
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"items"
],
"properties": {
"kind": {
"type": "string",
"const": "list"
},
"items": {
"type": "array",
"items": {
"type": "object",
"required": [
"say"
],
"properties": {
"say": {
"type": "string"
},
"note": {
"type": "string"
},
"means": {
"type": "string"
}
},
"additionalProperties": false
}
},
"ordered": {
"type": "boolean"
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"image"
],
"properties": {
"say": {
"type": "string"
},
"kind": {
"type": "string",
"const": "image"
},
"image": {
"type": "string"
},
"means": {
"type": "string"
}
},
"additionalProperties": false
}
]
}
}
},
"additionalProperties": false
}Show the dialogue widget. The learner wants to practise a conversation of a real situation (calling the GP, at the counter, asking the way) by playing one side of it. Rules: - 4 to 8 lines. A and B take turns; A opens, unless the situation has the learner speak first. - The learner plays B. The app hides each B line behind "Show" and shows its `mean` as the cue: the learner says the line, then shows it. - Each line is what a native speaker says there, at the learner's level, at most 12 words. Use the set phrases of the situation: greetings, asking, thanking, closing. - A B line answers the A line before it and needs nothing that the cue does not say: no invented names, dates or numbers that the learner cannot know. Give such facts in `situation`. - `mean` is natural, in the support language; a B cue says what to say, not how: "Say you'd like an appointment." - Take the chunks of this chat's material and the learner's bag when they fit the situation.
| Argument | Type | Description |
|---|---|---|
| a* | string | |
| b* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| lines* | array |
Show the exam-task widget. The learner asks to practise for their exam, or for an exam part (reading, listening, a test item): one item in that exam's own format. Rules: - Copy the exam's format: its kind of text, its question style, its number of options. A reading part: a short real text, a question in the exam's language, 3 options A–C. - The text is new, at the exam's level, 40–120 words, from the exam's daily settings: the town hall, the doctor, school, work, the house. Use the learner's material and bag when they fit. - One option is right by the text alone. The wrong ones are what a hurried reader picks: a word from the text in a wrong claim, true but not asked, the opposite. - The props carry the answer: this is practice. The app checks the pick against `answer` and shows the feedback at once; the answer is recorded with no turn. You run only when the learner taps "Explain more" (`explain-answer`) or "Dig deeper" (`dig-deeper`) under it: then show the words in the text that decide it, and on a wrong pick call `suggest_cards` with a `hard` card. - `seconds`: the exam's own pace for the item (a reading part: about 90 per item); the app counts it down and does not stop the learner. - One exam-task per reply.
| Argument | Type | Description |
|---|---|---|
| ask* | string | |
| exam* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
Show the examples widget. The learner wants to see a word, chunk or pattern at work in other sentences. Rules: - 3 to 6 items. The first is the simplest. - Each sentence is natural, at the learner's level, at most 10 words. - Vary person, tense and setting. Prefer daily-life and exam settings: work, school, the town hall, the doctor, the house. - `mark` is an exact substring of `say`, in the form it has there (Spanish «fuimos» for «ir»). When the parts of the target stand apart in the sentence, mark the first part. - No example repeats the source sentence. - Give `image` to each line whose scene is concrete (a place, an action, a thing): nominate a picture that truly shows that scene, and write its concept here.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| note | string | |
| gloss | string | |
| items* | array | |
| target* | string |
Show the explain widget. One rule, exception or tip explains what the learner met. Keep it to what this case needs. Rules: - One idea per component. Two ideas: two components, or a `contrast`. - Start from the learner's sentence, then the general rule. - No grammar words the learner has not met; if one is needed, explain it in the same line. - Never state a rule you are not sure of. - Tips that help the learner remember it, use it or pass their exam with it go in `tips`, not here.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| kind* | string | |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| gloss | string | |
| table | object | |
| title* | string | |
| points* | array |
Show the forms widget. The learner asks about the forms of one word: verb conjugation, gender or article, plural, comparative. Rules: - Only the forms the learner's level needs. For verbs at A2: present, perfect, past of common verbs. - Show a word whose form changes as it appears in sentences, and one whose parts split in the language: Spanish «vuelvo», «he vuelto». - Nouns: the forms the target language marks, such as article or gender, plural, diminutive.
| Argument | Type | Description |
|---|---|---|
| pos* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| note | string | |
| gloss | string | |
| lemma* | string | |
| table* | object |
Show the gloss-chunks widget. The learner wants to see how a target-language span is built: what each word means, which words move together, and which parts belong together across the sentence. Rules: - Copy `text` exactly from the source. Leave out punctuation from `words`; the app puts it back from `text`. Only with `fix`, `text` is the corrected span and `fix.wrote` the span as the source has it. - `fix`: only for a clear mistake: spelling, grammar, word order or a wrong word. Never for style, punctuation spacing, or a span that is already correct; never for official or published material. Most spans have none. With a fix, the chunks, the glosses and the translation are of the corrected `text`. - Every word of `text` appears once, in order, in exactly one chunk. A chunk holds only words that stand side by side in `text`. - Gloss: the literal meaning of the word alone, 1–3 lowercase words. Broken English is right: «No olvides nunca tus medicinas» → not · forget · never · your · medicines. - A chunk is what a native speaker says as one piece: 1–4 words. Its `means` is short and natural. - When parts of one unit stand apart in the sentence, keep each part in its own chunk where it stands, and link them with the same LinkId. Never pull them into one chunk: Dutch «Ik heb je gisteren gebeld» is [Ik] [heb¹] [je] [gisteren] [gebeld¹], not [heb gebeld] [je] [gisteren]. Name the whole form in `links` (Spanish: `no … nunca`, Dutch: `heb … gebeld`) and say what kind of unit it is: a reference to its concept card, `[double negative](card:doble negación)`, the KEY its name in the text's language. Offer that concept in suggest_cards (`kind: concept`) unless <bag> has it. No reference in `text` or `words`. - If the selection cuts a link (the learner selected «olvides nunca» without «no»), widen `text` to the smallest span that holds both parts and set `widened`. - `translation` uses the sentence around the span for its meaning, but translates only `text`. - Give `image` to every chunk that names a concrete thing, action or place: nominate a picture that truly shows it, and write its concept here. Never on a function word or a grammar chunk.
| Argument | Type | Description |
|---|---|---|
Show the next widget. The end of most replies: offer 1–6 one-click next steps, the asks that would help but were not made in this reply, and the next text to digest in this material or this card. Rules: - Each item is one click: it sends the learner's next message. Give exactly one of `action` and `ask`. - `action` is an id from the list, never a label or a question. A question the learner may want to ask goes in `ask`, with no `on`. - An `action` item needs `on` when this turn had no selection (the learner typed with no context bubble): copy `on` exactly from the material or from your own widgets. - Point to text the learner has not touched yet when the material has some (the context lists untouched sentences). - Offer the tertiary asks: more of what you showed, or a widget you did not make, each as one item: "More examples", "More quiz questions", "Synonyms", "Its forms", "Say it formally". An `ask` holds the whole request in the learner's words, about this turn's selection: "Give me 3 more examples of «a tiempo»". - Never more than 6; the first is the most useful. Skip `next` when the learner asked a closed question.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| items* | array |
{
"type": "object",
"required": [
"items",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"items": {
"type": "array",
"items": {
"type": "object",
"required": [
"label"
],
"properties": {
"on": {
"type": "string"
},
"ask": {
"type": "string"
},
"label": {
"type": "string"
},
"action": {
"enum": [
"play",
"dig-in",
"examples",
"other-ways",
"quiz",
"check",
"make-questions",
"tour"
],
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}Show the pattern widget. The learner met a frame they can reuse with other words. Rules: - The number of `slots` equals the number of "___" in `frame`. - Fillers fit every other filler: any combination is a correct sentence. - Take fillers from the learner's bag first when the context lists cards that fit.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| mean* | string | |
| built* | array | |
| frame* | string | |
| slots* | array |
{
"type": "object",
"required": [
"frame",
"mean",
"slots",
"built",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"mean": {
"type": "string"
},
"built": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"mean"
],
"properties": {
"say": {
"type": "string"
},
"mean": {
"type": "string"
}
},
"additionalProperties": false
}
},
"frame": {
"type": "string"
},
"slots": {
"type": "array",
"items": {
"type": "object",
"required": [
"hint",
"fill"
],
"properties": {
"fill": {
"type": "array",
"items": {
"type": "string"
}
},
"hint": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}Show the picture widget. A concrete word, chunk, action or scene is easier to remember with a picture than with words alone. Rules: - Nominate the picture of `say` in the same message, and write its concept here: the widget shows at once, and the picture fills in when it is drawn. - One picture per concrete word, chunk or action worth seeing. When the turn teaches several, show the most useful one here and put the others on the chunks and example lines that teach them (`image`). None for grammar that no scene shows. - `say` is the language element the picture teaches, copied as the learner met it.
| Argument | Type | Description |
|---|---|---|
| say* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| note | string | |
| image* | string | |
| means* | string |
Show the quiz widget. The learner should use what was just shown, once, right away. Rules: - Test only what this chat just taught. - One quiz per reply. Distractors are real mistakes a learner makes, not nonsense. - The app checks choice, cloze and order from these props (`answer`) and shows the result at once, in place: the right option, box or tiles in green, a wrong pick or answer in red with the right one beside it, then `explain`. No model call, right or wrong. - The `answer` of a choice quiz is the 0-based index of its right option, a number: `0` is the first option, `1` the second. Send `0`, not `"0"`, `"A"` or the option's text. (The app also reads those three, as a safety net; do not rely on it.) - Every answer, of any type, is recorded with no turn: the app sends `POST /chats/:id/turns {action:"answer", target, given, correct}` (`given` of a choice is the picked option's index), and the server stores it; no tutor reply follows. A quiz is answered once. - The tutor runs only when the learner taps an `on: quiz` action under the feedback (harness/actions.md): "Explain more" (`explain-answer`) or "Dig deeper" (`dig-deeper`). It comes with the quiz and the learner's answer, as an "answer" turn did. - Translate is free text: `correct` is null, and `answer` is one model answer, which the app shows with `explain` after the answer. "Check my answer" (`explain-answer`) asks you to check the learner's text against it.
| Argument | Type | Description |
|---|---|---|
| ask* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| kind* | string | |
Show the register widget. The learner asks how to say a message politely or casually, to whom to say «u» or «je», or must write or say it to an official, a stranger or a friend: one message, formal and informal. Rules: - Both sides say the same thing; change only what the register changes: pronouns, verb forms, greetings, closings, set phrases. - 1 to 3 sentences per side, natural, at the learner's level. - `marks` are exact substrings of their `say`, in the order they appear, each only once; leave out the words both sides share. - `formal` is what you write to a town hall, a school or a doctor; `informal`, to a friend, a neighbour, a colleague you know. - `note` says who gets which, with a word from the message: «u» to someone you do not know.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| mean* | string | |
| note | string | |
| formal* | object | |
| informal* | object |
Show the scene widget. The learner wants the words of a place or situation (what is in a waiting room, at the pharmacy, in a kitchen): one library picture with its 3–6 key words. Rules: - Nominate the scene in the same message, with the scene as `concept` and every word of `words` drawn in its `prompt`, and write that concept here. A library picture is used only when it truly shows the scene and every word. - 3 to 6 words, each one the picture shows, most useful first. Nouns with their article; an action as the learner would say it: «wachten». - The app lists the words beside the picture; it puts nothing on the picture. Choose words the learner finds without a pointer. - One scene per place or situation the turn teaches; a scene and pictures of single words can share a reply. Fold a scene that is not the reply's main widget.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| image* | string | |
| means* | string | |
| title* | string | |
| words |
Show the sentence-frame widget. The learner asks where the words of one sentence go, or why a word stands where it does (a verb at the end, a subject after the verb): word order, not meaning. Rules: - Slots in sentence order; their `words`, joined, are `say` without its punctuation. Every word is in exactly one slot. - Use the slots the language's grammar teaches, as its school grammar names them: for example subject · verb 1 · middle · verb 2, or for a language with a free order the slots that this sentence fills. - Keep a slot that is empty in this sentence when its place is the lesson (verb 2 with no second verb); else leave it out. - `verb` marks only the verb places: the finite verb, the infinitive or participle, a split verb's particle. - `note` names the rule in plain words: «Quiero» says «I», so the subject place stays empty.
| Argument | Type | Description |
|---|---|---|
| say* | string | |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| mean* | string | |
| note | string | |
| slots |
Show the shadow widget. The learner should hear lines and say them after the voice: hear it, repeat it. Rules: - Lines of at most 8 words, natural rhythm, from this chat's material and examples. - Order from short to long. - The app plays each line, waits about 1.5 × its length for the learner to repeat, then plays the next.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| lines* | array | |
| speed | number |
{
"type": "object",
"required": [
"lines",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"lines": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"mean"
],
"properties": {
"say": {
"type": "string"
},
"mean": {
"type": "string"
}
},
"additionalProperties": false
}
},
"speed": {
"type": "number"
}
},
"additionalProperties": false
}Show the tips widget. Tips help the learner keep or use what they are learning: a tip for their goal exam, how people typically use it, a memory trick. Rules: - Each tip is specific to what the learner met: the word, chunk, pattern or exam item of this chat. Never generic study advice: no "practise every day", no "review often". - An exam tip names the learner's goal exam from <goals>, and says how this item shows up there or how to handle that kind of question. Without a goal exam, leave the exam tip out. - Never invent an exam fact. Say only what you are sure of about the exam's format, parts and question types; when unsure, leave the exam tip out. - A use tip says what the item goes with and where people say it: collocations, set phrases, register (a letter, a shop, with friends, with an official). - A memory tip gives one concrete trick: a picture to see, a sound-alike in the support language, a short story, or a contrast with a look-alike. One trick, not a list. - At most one tip per kind. Give only the kinds you have something good for; 1 good tip beats 3 weak ones. - `example` is one natural sentence at the learner's level that shows the tip at work; leave it out when the points already hold a «…» line. - Put each word of a learned language in a point in «guillemets».
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| tips* | array | |
Show the visual widget. The learner needs to see how things relate: a verb's forms across tenses in a grid, tenses and aspect on a time line, the parts of one sentence as a tree, or a topic (a KNM theme) with what belongs to it as a map. Rules: - Choose the `kind` by what must be seen. `verbs`: the forms of one verb in 1–4 tenses. `timeline`: when things happen against now, with tense and aspect. `tree`: how the parts of one sentence fit. `map`: a topic with the words and ideas that belong to it. - A simple table of one word's forms is `forms`; the places of words in a sentence are `sentence-frame`. Use `visual` when the picture teaches more than the table or the line would. - Every word of a learned language is in a `Say`: `lemma`, a form, `say`, `topic`, an item. A `label`, `role`, `link` or `head` is in the support language; learned words in them go in «guillemets». - `verbs`: persons as the language lists them; one form per column, every row complete. Name in `irregular` the forms that break the pattern, so the grid draws them in colour. - `timeline`: left is past, right is future, `now` between. A stretch (`to`) is something that lasted or kept happening; a point is something done. At most 5 events, ordered from `from`, each with its own tense or aspect and one natural sentence. - `tree`: the `say` of the parts, joined, are the words of the sentence; split only the part the lesson is about, and name roles as the language's school grammar does. At most 4 parts, each with at most 3 smaller parts. - `map`: for a KNM topic, branches are its subtopics and items the words the learner must know for the exam; at most 6 branches of at most 4 items. - Every `Say` must be correct and natural, at the learner's level. Show only what the learner has met or asked about. - One visual per reply.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
Show the voice widget. The learner should hear a line or a short dialogue and work from the sound: listening practice, an exam's listening part, or lines to say after the voice. Rules: - Write what is said, as it is said: natural, at the learner's level, at most 6 lines of at most 20 words. For a dialogue give each line its `by` and let the speakers take turns; one voice reads every line. - The transcript stays hidden until the learner shows it or answers the question. Leave `transcript` out for listening practice; set `"shown"` only when there is nothing to find out by ear, or the learner is to say the lines after the voice. - Exam style: set `plays` to what the exam gives (its listening part: 2), leave the transcript hidden and add one `question` whose answer is in the sound, not in `title` or `ask`. The limit holds while the transcript is hidden: showing it or answering ends it. - The app checks the pick against `answer` and shows the feedback at once, then the transcript; no call goes to the host, right or wrong. Run again only for the learner's next ask: a harder one, a slower one, or why an answer was wrong. - Shadowing: set `shadow: true`, `speed: 0.75` for new lines, lines of at most 8 words ordered short to long, `transcript: "shown"`. It plays each line and waits for the learner to say it. Without `plays`, nothing limits it. - For lines to say after the voice and nothing else, use `shadow`; for a role-play the learner takes a part in, `dialogue`. - Give `mean` on every line, so each can be saved once the transcript shows. - One voice widget per reply.
| Argument | Type | Description |
|---|---|---|
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
Make a natural voice recording of a sentence in the language the learner learns, with the time of each word, for the voice widget (show_voice) or a link. Spends a little credit; Axad keeps each recording, so the same sentence again is free.
| Argument | Type | Description |
|---|---|---|
| say* | string | |
| lang* | string | |
| voice | string | A speaker label of the language ("Vrouw", "Man"); omit for the narrator. |
{
"type": "object",
"required": [
"say",
"lang"
],
"properties": {
"say": {
"type": "string",
"maxLength": 600
},
"lang": {
"type": "string"
},
"voice": {
"type": "string",
"description": "A speaker label of the language (\"Vrouw\", \"Man\"); omit for the narrator."
}
}
}Start a learning path for the learner in a language, or go back to their place in it. Returns the current step: what to do, with which tools and widgets, and when it is done. Follow it, then call record_step.
| Argument | Type | Description |
|---|---|---|
| lang* | string | |
| runbook* | string |
{
"type": "object",
"required": [
"runbook",
"lang"
],
"properties": {
"lang": {
"type": "string"
},
"runbook": {
"type": "string"
}
}
}A link for the learner to pick a plan, add a card for more free use, or buy credit on axad.ai. Give it when the learner asks, or after limit_reached or allowance_spent; never push it otherwise.
| Argument | Type | Description |
|---|---|---|
| want | string |
{
"type": "object",
"properties": {
"want": {
"enum": [
"plans",
"card",
"credit"
],
"type": "string"
}
}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 21, 2026.
| prompt* | string | What to draw, 40–200 English words: the scene, each object and where it is, what the people do. Concrete, drawable and generic: no names, clock times, dates or numbers (except quoted `text`); no style words. |
| concept* | string | The language element it teaches, 1–8 words in its language: "cancelar", "la sala de espera". The `image` fields that show this picture hold these words. |
{
"type": "object",
"required": [
"title",
"concept",
"canon",
"lang",
"prompt"
],
"properties": {
"lang": {
"type": "string",
"description": "BCP 47 code of the concept."
},
"text": {
"type": "string",
"description": "Only when words in the picture are the lesson: at most 6 words in `lang`, also written in the prompt in double quotes. Omit it for a picture with no words, which serves every language."
},
"canon": {
"type": "string",
"description": "The generic English concept the picture shows, 1–4 words, dictionary form, with no names, numbers or words of one sentence: \"doctor's practice\", \"apple\", \"train platform\". The library keeps one picture per canon."
},
"title": {
"type": "string",
"description": "One literal English sentence of 8–40 words: who, what, where. Generic, with no names, numbers, clock times or dates. Search reads it."
},
"prompt": {
"type": "string",
"description": "What to draw, 40–200 English words: the scene, each object and where it is, what the people do. Concrete, drawable and generic: no names, clock times, dates or numbers (except quoted `text`); no style words."
},
"concept": {
"type": "string",
"description": "The language element it teaches, 1–8 words in its language: \"cancelar\", \"la sala de espera\". The `image` fields that show this picture hold these words."
}
}
}| natural | string |
{
"type": "object",
"required": [
"wrote",
"fixed",
"ok",
"fixes",
"lang"
],
"properties": {
"ok": {
"type": "boolean"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"fixed": {
"type": "string"
},
"fixes": {
"type": "array",
"items": {
"type": "object",
"required": [
"from",
"to",
"why"
],
"properties": {
"to": {
"type": "string"
},
"why": {
"type": "string"
},
"from": {
"type": "string"
}
},
"additionalProperties": false
}
},
"gloss": {
"type": "string"
},
"wrote": {
"type": "string"
},
"natural": {
"type": "string"
}
},
"additionalProperties": false
}| situation* | string |
{
"type": "object",
"required": [
"situation",
"a",
"b",
"lines",
"lang"
],
"properties": {
"a": {
"type": "string"
},
"b": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"lines": {
"type": "array",
"items": {
"type": "object",
"required": [
"by",
"say",
"mean"
],
"properties": {
"by": {
"enum": [
"a",
"b"
],
"type": "string"
},
"say": {
"type": "string"
},
"mean": {
"type": "string"
}
},
"additionalProperties": false
}
},
"situation": {
"type": "string"
}
},
"additionalProperties": false
}| text* | string |
| gloss | string |
| answer* | number |
| explain* | string |
| options* | array |
| seconds | number |
{
"type": "object",
"required": [
"exam",
"text",
"ask",
"options",
"answer",
"explain",
"lang"
],
"properties": {
"ask": {
"type": "string"
},
"exam": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"text": {
"type": "string"
},
"gloss": {
"type": "string"
},
"answer": {
"type": "number"
},
"explain": {
"type": "string"
},
"options": {
"type": "array",
"items": {
"type": "string"
}
},
"seconds": {
"type": "number"
}
},
"additionalProperties": false
}{
"type": "object",
"required": [
"target",
"items",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"note": {
"type": "string"
},
"gloss": {
"type": "string"
},
"items": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"mean",
"mark"
],
"properties": {
"say": {
"type": "string"
},
"mark": {
"type": "string"
},
"mean": {
"type": "string"
},
"image": {
"type": "string"
}
},
"additionalProperties": false
}
},
"target": {
"type": "string"
}
},
"additionalProperties": false
}{
"type": "object",
"required": [
"title",
"kind",
"points",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"kind": {
"type": "string"
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"gloss": {
"type": "string"
},
"table": {
"type": "object",
"required": [
"head",
"rows"
],
"properties": {
"head": {
"type": "array",
"items": {
"type": "string"
}
},
"rows": {
"type": "array",
"items": {
"type": "array",
"items": {
"type": "string"
}
}
}
},
"additionalProperties": false
},
"title": {
"type": "string"
},
"points": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
}{
"type": "object",
"required": [
"lemma",
"pos",
"table",
"lang"
],
"properties": {
"pos": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"note": {
"type": "string"
},
"gloss": {
"type": "string"
},
"lemma": {
"type": "string"
},
"table": {
"type": "object",
"required": [
"head",
"rows"
],
"properties": {
"head": {
"type": "array",
"items": {
"type": "string"
}
},
"rows": {
"type": "array",
"items": {
"type": "array",
"items": {
"type": "string"
},
"minItems": 1,
"prefixItems": [
{
"type": "string"
}
]
}
}
},
"additionalProperties": false
}
},
"additionalProperties": false
}| fix |
| object |
| fold | boolean | true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open. |
| lang* | string | BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| text* | string |
| links | object |
| chunks* | array |
| widened | boolean |
| translation* | string |
{
"type": "object",
"required": [
"text",
"chunks",
"translation",
"lang"
],
"properties": {
"fix": {
"type": "object",
"required": [
"wrote",
"changes"
],
"properties": {
"wrote": {
"type": "string"
},
"changes": {
"type": "array",
"items": {
"type": "object",
"required": [
"from",
"to",
"why"
],
"properties": {
"to": {
"type": "string"
},
"why": {
"type": "string"
},
"from": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"text": {
"type": "string"
},
"links": {
"type": "object",
"additionalProperties": {
"type": "object",
"required": [
"lemma",
"means",
"kind"
],
"properties": {
"kind": {
"type": "string"
},
"lemma": {
"type": "string"
},
"means": {
"type": "string"
}
},
"additionalProperties": false
}
},
"chunks": {
"type": "array",
"items": {
"type": "object",
"required": [
"words"
],
"properties": {
"image": {
"type": "string"
},
"means": {
"type": "string"
},
"words": {
"type": "array",
"items": {
"anyOf": [
{
"type": "array",
"maxItems": 2,
"minItems": 2,
"prefixItems": [
{
"type": "string"
},
{
"type": "string"
}
]
},
{
"type": "array",
"maxItems": 3,
"minItems": 3,
"prefixItems": [
{
"type": "string"
},
{
"type": "string"
},
{
"type": "string"
}
]
}
]
}
}
},
"additionalProperties": false
}
},
"widened": {
"type": "boolean"
},
"translation": {
"type": "string"
}
},
"additionalProperties": false
}{
"type": "object",
"required": [
"image",
"say",
"means",
"lang"
],
"properties": {
"say": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"note": {
"type": "string"
},
"image": {
"type": "string"
},
"means": {
"type": "string"
}
},
"additionalProperties": false
}| lang* |
| string |
| BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| gloss | string |
| answer* | any |
| prompt | string |
| explain* | string |
| options | array |
{
"type": "object",
"required": [
"kind",
"ask",
"answer",
"explain",
"lang"
],
"properties": {
"ask": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"kind": {
"enum": [
"choice",
"cloze",
"order",
"translate"
],
"type": "string"
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"gloss": {
"type": "string"
},
"answer": {
"anyOf": [
{
"type": "number"
},
{
"type": "string"
},
{
"type": "array",
"items": {
"type": "string"
}
}
]
},
"prompt": {
"type": "string"
},
"explain": {
"type": "string"
},
"options": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
}{
"type": "object",
"required": [
"mean",
"formal",
"informal",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"mean": {
"type": "string"
},
"note": {
"type": "string"
},
"formal": {
"type": "object",
"required": [
"label",
"say",
"marks"
],
"properties": {
"say": {
"type": "string"
},
"label": {
"type": "string"
},
"marks": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
},
"informal": {
"type": "object",
"required": [
"label",
"say",
"marks"
],
"properties": {
"say": {
"type": "string"
},
"label": {
"type": "string"
},
"marks": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
}
},
"additionalProperties": false
}| array |
{
"type": "object",
"required": [
"image",
"title",
"means",
"words",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"image": {
"type": "string"
},
"means": {
"type": "string"
},
"title": {
"type": "string"
},
"words": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"means"
],
"properties": {
"say": {
"type": "string"
},
"means": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}| array |
{
"type": "object",
"required": [
"say",
"mean",
"slots",
"lang"
],
"properties": {
"say": {
"type": "string"
},
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"mean": {
"type": "string"
},
"note": {
"type": "string"
},
"slots": {
"type": "array",
"items": {
"type": "object",
"required": [
"name",
"words"
],
"properties": {
"name": {
"type": "string"
},
"verb": {
"type": "boolean"
},
"words": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}| gloss |
| string |
| title* | string |
{
"type": "object",
"required": [
"title",
"tips",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"tips": {
"type": "array",
"items": {
"type": "object",
"required": [
"kind",
"title",
"points"
],
"properties": {
"kind": {
"enum": [
"exam",
"use",
"memory"
],
"type": "string"
},
"title": {
"type": "string"
},
"points": {
"type": "array",
"items": {
"type": "string"
}
},
"example": {
"type": "string"
}
},
"additionalProperties": false
}
},
"gloss": {
"type": "string"
},
"title": {
"type": "string"
}
},
"additionalProperties": false
}| lang* |
| string |
| BCP 47 code of the language the widget teaches: "nl". The view speaks and saves its words in it. |
| note | string |
| chart* | any |
| title* | string |
{
"type": "object",
"required": [
"title",
"chart",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"note": {
"type": "string"
},
"chart": {
"anyOf": [
{
"type": "object",
"required": [
"kind",
"lemma",
"head",
"rows"
],
"properties": {
"head": {
"type": "array",
"items": {
"type": "string"
}
},
"kind": {
"type": "string",
"const": "verbs"
},
"rows": {
"type": "array",
"items": {
"type": "array",
"items": {
"type": "string"
},
"minItems": 1,
"prefixItems": [
{
"type": "string"
}
]
}
},
"gloss": {
"type": "string"
},
"lemma": {
"type": "string"
},
"irregular": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"now",
"events"
],
"properties": {
"now": {
"type": "number"
},
"kind": {
"type": "string",
"const": "timeline"
},
"events": {
"type": "array",
"items": {
"type": "object",
"required": [
"label",
"say",
"from"
],
"properties": {
"to": {
"type": "number"
},
"say": {
"type": "string"
},
"from": {
"type": "number"
},
"mean": {
"type": "string"
},
"label": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"say",
"mean",
"parts"
],
"properties": {
"say": {
"type": "string"
},
"kind": {
"type": "string",
"const": "tree"
},
"mean": {
"type": "string"
},
"parts": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"role"
],
"properties": {
"say": {
"type": "string"
},
"role": {
"type": "string"
},
"parts": {
"type": "array",
"items": {
"type": "object",
"required": [
"say",
"role"
],
"properties": {
"say": {
"type": "string"
},
"role": {
"type": "string"
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
},
{
"type": "object",
"required": [
"kind",
"topic",
"branches"
],
"properties": {
"kind": {
"type": "string",
"const": "map"
},
"mean": {
"type": "string"
},
"topic": {
"type": "string"
},
"branches": {
"type": "array",
"items": {
"type": "object",
"required": [
"say"
],
"properties": {
"say": {
"type": "string"
},
"link": {
"type": "string"
},
"mean": {
"type": "string"
},
"items": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
}
}
},
"additionalProperties": false
}
]
},
"title": {
"type": "string"
}
},
"additionalProperties": false
}| lines* | array |
| plays | number |
| speed | number |
| title | string |
| shadow | boolean |
| question | object |
| transcript | string |
{
"type": "object",
"required": [
"lines",
"lang"
],
"properties": {
"fold": {
"type": "boolean",
"description": "true: a secondary widget of this reply, made now and shown folded: only its title row, until one click opens it. Omit it on the reply's main widgets, which show open."
},
"lang": {
"type": "string",
"description": "BCP 47 code of the language the widget teaches: \"nl\". The view speaks and saves its words in it."
},
"lines": {
"type": "array",
"items": {
"type": "object",
"required": [
"say"
],
"properties": {
"by": {
"type": "string"
},
"say": {
"type": "string"
},
"mean": {
"type": "string"
}
},
"additionalProperties": false
}
},
"plays": {
"type": "number"
},
"speed": {
"type": "number"
},
"title": {
"type": "string"
},
"shadow": {
"type": "boolean"
},
"question": {
"type": "object",
"required": [
"ask",
"options",
"answer",
"explain"
],
"properties": {
"ask": {
"type": "string"
},
"answer": {
"type": "number"
},
"explain": {
"type": "string"
},
"options": {
"type": "array",
"items": {
"type": "string"
}
}
},
"additionalProperties": false
},
"transcript": {
"enum": [
"hidden",
"shown"
],
"type": "string"
}
},
"additionalProperties": false
}