RooQuiz

add_question

Append an item to the end of a form. type is a question type, Breaker (page break — only formId + type are needed, other fields are ignored) or a display block (Statement / Swiper). Question types: SingleCheck / MultiCheck / TrueFalse; FillBlank (free text — scored in quiz via correctAnswer, an unscored data-collection field in scored_quiz); DropDown (single or multiple via multiple — prefer it over SingleCheck/MultiCheck past 20 choices); Cascade (hierarchical via choices[i].children, scored_quiz only); Ordering (quiz only, order-sensitive grading); DateField / TimeField (unscored data-collection fields, scored_quiz only, no correctAnswer/score); NumberField (quiz: optional numeric correctAnswer + score; scored_quiz: the submitted number feeds report formulas); Rate (scored_quiz only, the submitted 1..steps rating is the question score unless per-star scores are set in the web app). Display blocks carry no answer: { type: "Statement", content } renders a rich-text passage (intro, section lead-in, disclaimer) and { type: "Swiper", items } an image carousel; both work in every scene. Configure random_knowledge_quiz question banks in the web app. Not idempotent: if the call times out it may still have succeeded, so retrying blindly can create a duplicate — check first, then retry only if it is really missing.

Input Schema

TableJSON Schema

Name	Required	Description
`max`	No	NumberField only: maximum allowed input value (must be >= min). Rejected for other question types.
`min`	No	NumberField only: minimum allowed input value (respondents cannot submit a smaller number). Rejected for other question types.
`code`	No	Optional stable identifier for this question (field code). Omit it to let the server auto-generate one. Set a meaningful code (e.g. "q1") when report.formula or a dimension needs to reference this question, so you can write the formula as `{{q1}}` in the same call instead of round-tripping via get_form. Rules: start with a letter or underscore, then only letters/digits/underscores (no hyphens, spaces, or leading digit), at most 64 chars, and not a reserved math word (e, E, pi, PI, tau, phi, i, Infinity, NaN, true, false, null, undefined). Must be unique among all items in the form.
`name`	No	Question stem text. Allows plain text or restricted HTML (tag allowlist: <p> <strong>/<b> <em>/<i> <u> <s> <mark> <span> <sup> <sub> <br>; other tags are stripped and the text kept).
`type`	Yes	Question type; Breaker means a page break, no name/choices etc. needed; Statement / Swiper are display-only blocks that collect no answer
`unit`	No	NumberField only: display unit suffix shown after the input, e.g. "kg" / "$" / "min". Rejected for other question types.
`items`	No	Swiper only, and required there: the carousel slides in display order, 1-10 of them. Upload the images with prepare_image_upload / finalize_image_upload first and pass the returned media IDs. Rejected for every other type.
`score`	No	Points this question is worth, default 0 (not scored). Quiz scene: awarded when the answer matches correctAnswer, and a positive value is required once correctAnswer is set. Scored Quiz scene: pairing it with correctAnswer enables the fallback mode above, but choices[i].score is more flexible. Rejected in the outcome_quiz scene, for DateField / TimeField / Rate, and — in the scored_quiz scene — for NumberField (where the submitted number itself is the score) and FillBlank (collected only, never scored).
`steps`	No	Rate only: number of rating steps, i.e. the highest rating (3-10, default 5). In the scored_quiz scene the submitted rating value (1..steps) is the question score, unless a per-star score is configured in the web app. Rejected for other question types.
`words`	No	Rate only: optional scale labels evenly distributed under the rating control, e.g. ["Poor", "Excellent"] for the two endpoints (up to 5 labels). Rejected for other question types.
`formId`	Yes	The form ID to append the item to
`aiMatch`	No	Only for FillBlank in the knowledge_quiz scene. Enables AI grading: the AI compares the respondent answer against correctAnswer and scores by accuracy, instead of requiring an exact string match. Requires correctAnswer (the standard answer) and score > 0 (the score earned when accuracy reaches the threshold). Pass an empty object {} to enable with default settings; omit for plain exact-match grading.
`choices`	No	Choice-based questions only (SingleCheck / MultiCheck / DropDown / Ordering / Cascade), where it is required; ignored for every other type, including TrueFalse — its two options come from trueLabel / falseLabel. Per-type limits: SingleCheck / MultiCheck 2-20 items — for a longer list use DropDown (2-100 items) instead; Ordering 2-10 items; Cascade nests via choices[i].children (up to 3 levels, at most 100 nodes in total). IMPORTANT (knowledge_quiz scene): vary the position of the correct option(s) across questions — do NOT always place the correct answer first. Distribute correct answers roughly evenly over all positions so they are not predictable.
`content`	No	Statement only, and required there: the text respondents read — an intro, a section lead-in, instructions, a disclaimer. The block renders this and nothing else. Same rich-text rules as `description` (headings / lists / links / <img src> / math formulas). This field also accepts an inline image: put an <img src="..."> in it, where src is a direct image URL that renders in <img src> (a page URL that merely contains an image does not work). Use finalize_image_upload to host an image yourself, or a direct URL the user supplied. Never invent an image URL — omit the image instead of risking a broken one.
`explain`	No	Optional answer explanation. The frontend renders it in the question's "answer explanation" field (DescriptionEditor); the rich-text rules are identical to description. Do not stuff the answer explanation into description — that is the question's supplementary note and will not be shown as an explanation to respondents/graders.
`shuffle`	No	Ordering only: shuffle the displayed choice order for each respondent. Defaults to true for MCP-created questions — the stored choices order would otherwise leak the correct order when correctAnswer matches it. Pass false only when the initial order is intentionally meaningful. Rejected for other question types.
`multiple`	No	DropDown only: allow selecting multiple options (default false = single select). Affects the quiz-scene correctAnswer shape: an array of labels/codes when true, a single one when false. Rejected for other question types (SingleCheck/MultiCheck are inherently single/multi).
`required`	No	Whether the question is required, default false
`precision`	No	DateField / TimeField only: picker precision. DateField accepts year \| month \| day \| hour \| minute \| second (default day; e.g. "month" shows a year-month picker, "second" a full datetime picker). TimeField accepts only minute \| second (default minute). Ignored for other question types.
`trueLabel`	No	TrueFalse only: custom display text for the "true" option (e.g. "Yes" / "Agree"). Ignored for other question types. Leave empty to fall back to the built-in default for the form language ("Correct" in English forms). Does not change the stored answer value, which stays "true".
`falseLabel`	No	TrueFalse only: custom display text for the "false" option (e.g. "No" / "Disagree"). Ignored for other question types. Leave empty to fall back to the built-in default for the form language ("Incorrect" in English forms). Does not change the stored answer value, which stays "false".
`description`	No	Optional supplementary note for the question. Allows a wider HTML subset: everything the stem allows + <h1>-<h6> <ul> <ol> <li> <blockquote> <a href> <img src> <hr> <art-field> (variable placeholder, data-type / data-cid); unsafe protocols (javascript:/data:) and unknown attributes are stripped. This field also accepts an inline image: put an <img src="..."> in it, where src is a direct image URL that renders in <img src> (a page URL that merely contains an image does not work). Use finalize_image_upload to host an image yourself, or a direct URL the user supplied. Never invent an image URL — omit the image instead of risking a broken one.
`trueOutcomes`	No	Outcome scene + TrueFalse only (required there together with falseOutcomes): the outcome codes that answering "true" votes for. Use [] for a side that votes for nothing. Rejected for other question types / scenes.
`correctAnswer`	No	The "correct answer" of the knowledge_quiz scene; setting it makes the question scored, so pair it with a positive `score`. The shape follows the question type — see the anyOf branches; a choice is referenced by its label or its code, so reference it by code whenever the same label appears more than once (Ordering rejects an ambiguous label outright). Required on SingleCheck / MultiCheck / DropDown / Ordering in the knowledge_quiz scene, optional on FillBlank / NumberField there. NumberField answers must be typeable within min / max and decimalPlaces. Rejected for DateField / TimeField / Rate (data-collection and rating fields; configure date/time scoring in the web app), rejected for FillBlank in the scored_quiz scene (free text is collected only there), and rejected in the outcome_quiz scene (no right or wrong answers there). In the scored_quiz scene prefer choices[i].score per option; passing correctAnswer + score there only falls back to "the matching choice gets score, others get 0".
`decimalPlaces`	No	NumberField only: how many decimal places respondents may enter (stored as the field's numeric precision), default 0 = integers only. Rejected for other question types. Note this is different from the string `precision` of DateField / TimeField.
`falseOutcomes`	No	Outcome scene + TrueFalse only (required there together with trueOutcomes): the outcome codes that answering "false" votes for. Use [] for a side that votes for nothing. Rejected for other question types / scenes.
`trueDimensionScores`	No	scored_quiz scene + TrueFalse only: the points answering "true" adds to each dimension, as a map of dimension code → score (same semantics as choices[i].dimensionScores). Rejected for other question types / scenes.
`falseDimensionScores`	No	scored_quiz scene + TrueFalse only: the points answering "false" adds to each dimension, as a map of dimension code → score (same semantics as choices[i].dimensionScores). Rejected for other question types / scenes.

Output Schema

TableJSON Schema

Name	Required	Description
`field`	No	The created question, including its generated code
`formId`	No	The form that was edited
`itemCount`	No	Question / page-break count after the append

Schema Changelog

Changes observed during successful MCP inspections. Dates show when Glama detected each change.

Changed5 schema fields changedSep 5, 2026

added

Input schema / properties / choices / items / properties / children / items / properties / children / items / properties / dimensionScores

Added value: +{
+  "additionalProperties": {
+    "maximum": 100,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "scored_quiz scene only: the points this choice adds to each dimension when selected, as a map of dimension code → score, e.g. { \"dim_d\": 3, \"dim_i\": 1 }. Any choice carrying it switches the question to per-dimension scoring: for each dimension the question contributes the selected choices' scores for that dimension (a dimension not listed on the chosen choice counts 0), and a dimension without a formula automatically sums these contributions across questions. Codes must exist in report.dimensionAnalysis.dimensions — give dimensions[].code and reference them in the same create_form call. `score` remains the overall score fed to report.formula. Rejected in the knowledge_quiz / outcome_quiz scenes.",
+  "type": "object"
+}

addedInput schema / properties / choices / items / properties / children / items / properties / dimensionScores

Added value: +{
+  "additionalProperties": {
+    "maximum": 100,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "scored_quiz scene only: the points this choice adds to each dimension when selected, as a map of dimension code → score, e.g. { \"dim_d\": 3, \"dim_i\": 1 }. Any choice carrying it switches the question to per-dimension scoring: for each dimension the question contributes the selected choices' scores for that dimension (a dimension not listed on the chosen choice counts 0), and a dimension without a formula automatically sums these contributions across questions. Codes must exist in report.dimensionAnalysis.dimensions — give dimensions[].code and reference them in the same create_form call. `score` remains the overall score fed to report.formula. Rejected in the knowledge_quiz / outcome_quiz scenes.",
+  "type": "object"
+}

addedInput schema / properties / choices / items / properties / dimensionScores

Added value: +{
+  "additionalProperties": {
+    "maximum": 100,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "scored_quiz scene only: the points this choice adds to each dimension when selected, as a map of dimension code → score, e.g. { \"dim_d\": 3, \"dim_i\": 1 }. Any choice carrying it switches the question to per-dimension scoring: for each dimension the question contributes the selected choices' scores for that dimension (a dimension not listed on the chosen choice counts 0), and a dimension without a formula automatically sums these contributions across questions. Codes must exist in report.dimensionAnalysis.dimensions — give dimensions[].code and reference them in the same create_form call. `score` remains the overall score fed to report.formula. Rejected in the knowledge_quiz / outcome_quiz scenes.",
+  "type": "object"
+}

addedInput schema / properties / falseDimensionScores

Added value: +{
+  "additionalProperties": {
+    "maximum": 100,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "scored_quiz scene + TrueFalse only: the points answering \"false\" adds to each dimension, as a map of dimension code → score (same semantics as choices[i].dimensionScores). Rejected for other question types / scenes.",
+  "type": "object"
+}

addedInput schema / properties / trueDimensionScores

Added value: +{
+  "additionalProperties": {
+    "maximum": 100,
+    "minimum": 0,
+    "type": "number"
+  },
+  "description": "scored_quiz scene + TrueFalse only: the points answering \"true\" adds to each dimension, as a map of dimension code → score (same semantics as choices[i].dimensionScores). Rejected for other question types / scenes.",
+  "type": "object"
+}

Changed5 schema fields changedAug 26, 2026

changedInput schema / description

Previous value: -"FIELD APPLICABILITY — `type` is the only field the schema always requires; `type` (plus the form scene) decides which of the remaining fields apply, and one that does not apply is either rejected with an explanatory error or ignored, so never pass it. Breaker (page break) takes nothing but `type`. Every question type takes `name` (required) plus the optional `code` / `description` / `explain` / `required`. Type-specific fields: SingleCheck / MultiCheck / DropDown / Ordering / Cascade need `choices` (Cascade nests via choices[i].children; DropDown also takes `multiple`; Ordering also takes `shuffle`); TrueFalse has a fixed pair of options and takes `trueLabel` / `falseLabel` instead of `choices`; DateField / TimeField take `precision`; NumberField takes `min` / `max` / `unit` / `decimalPlaces`; Rate takes `steps` / `words`. Scene rules for scoring: quiz — `correctAnswer` plus `score` > 0 are REQUIRED on SingleCheck / MultiCheck / DropDown / Ordering, optional on FillBlank (add `aiMatch` to have AI grade it instead of exact string match) and NumberField (omit both and the question is just a data-collection field), and rejected on DateField / TimeField; scored_quiz — score each option via choices[i].score; `correctAnswer` / `score` are rejected on NumberField and Rate (the submitted value is the score) and on FillBlank (free text is collected only, never scored); outcome — nothing is scored: `correctAnswer` / `score` / `aiMatch` are all rejected, every choice needs `outcomes` (TrueFalse votes via `trueOutcomes` / `falseOutcomes`). Type availability per scene: Ordering is knowledge_quiz-only; FillBlank works in knowledge_quiz (optionally scored) and in scored_quiz (data-collection only — its answer never feeds a score or a report formula); Cascade, Rate, DateField and TimeField are scored_quiz-only; and the outcome_quiz scene accepts only SingleCheck / MultiCheck / DropDown / TrueFalse / Breaker."New value: +"FIELD APPLICABILITY — `type` is the only field the schema always requires; `type` (plus the form scene) decides which of the remaining fields apply, and one that does not apply is either rejected with an explanatory error or ignored, so never pass it. Breaker (page break) takes nothing but `type`. Display blocks collect no answer, are never scored and are allowed in every scene: Statement takes `content` (the rich text IS the block) and Swiper takes `items` (an image carousel); both also accept an optional `name`, a label used by the web editor only, and neither takes any question field. Every question type takes `name` (required) plus the optional `code` / `description` / `explain` / `required`. Type-specific fields: SingleCheck / MultiCheck / DropDown / Ordering / Cascade need `choices` (Cascade nests via choices[i].children; DropDown also takes `multiple`; Ordering also takes `shuffle`); TrueFalse has a fixed pair of options and takes `trueLabel` / `falseLabel` instead of `choices`; DateField / TimeField take `precision`; NumberField takes `min` / `max` / `unit` / `decimalPlaces`; Rate takes `steps` / `words`. Scene rules for scoring: quiz — `correctAnswer` plus `score` > 0 are REQUIRED on SingleCheck / MultiCheck / DropDown / Ordering, optional on FillBlank (add `aiMatch` to have AI grade it instead of exact string match) and NumberField (omit both and the question is just a data-collection field), and rejected on DateField / TimeField; scored_quiz — score each option via choices[i].score; `correctAnswer` / `score` are rejected on NumberField and Rate (the submitted value is the score) and on FillBlank (free text is collected only, never scored); outcome — nothing is scored: `correctAnswer` / `score` / `aiMatch` are all rejected, every choice needs `outcomes` (TrueFalse votes via `trueOutcomes` / `falseOutcomes`). Type availability per scene: Ordering is knowledge_quiz-only; FillBlank works in knowledge_quiz (optionally scored) and in scored_quiz (data-collection only — its answer never feeds a score or a report formula); Cascade, Rate, DateField and TimeField are scored_quiz-only; and the outcome_quiz scene accepts only SingleCheck / MultiCheck / DropDown / TrueFalse / Breaker."

addedInput schema / properties / content

Added value: +{
+  "description": "Statement only, and required there: the text respondents read — an intro, a section lead-in, instructions, a disclaimer. The block renders this and nothing else. Same rich-text rules as `description` (headings / lists / links / <img src> / math formulas). This field also accepts an inline image: put an <img src=\"...\"> in it, where src is a direct image URL that renders in <img src> (a page URL that merely contains an image does not work). Use finalize_image_upload to host an image yourself, or a direct URL the user supplied. Never invent an image URL — omit the image instead of risking a broken one.",
+  "maxLength": 5000,
+  "type": "string"
+}

addedInput schema / properties / items

Added value: +{
+  "description": "Swiper only, and required there: the carousel slides in display order, 1-10 of them. Upload the images with prepare_image_upload / finalize_image_upload first and pass the returned media IDs. Rejected for every other type.",
+  "items": {
+    "additionalProperties": false,
+    "properties": {
+      "description": {
+        "description": "Optional slide note. Web editor only, same as title.",
+        "maxLength": 1000,
+        "type": "string"
+      },
+      "image": {
+        "description": "The slide image: a media ID returned by finalize_image_upload, or the media URL of an image already in this team library. Required — the answer page renders the images and nothing else.",
+        "type": "string"
+      },
+      "title": {
+        "description": "Optional slide label. Shown in the web editor only (the answer page renders the image), and translatable per language.",
+        "maxLength": 200,
+        "type": "string"
+      }
+    },
+    "required": [
+      "image"
+    ],
+    "type": "object"
+  },
+  "maxItems": 10,
+  "minItems": 1,
+  "type": "array"
+}

changedInput schema / properties / type / description

Previous value: -"Question type; Breaker means a page break, no name/choices etc. needed"New value: +"Question type; Breaker means a page break, no name/choices etc. needed; Statement / Swiper are display-only blocks that collect no answer"

changedInput schema / properties / type / enum

Previous value: -[
-  "SingleCheck",
-  "MultiCheck",
-  "TrueFalse",
-  "FillBlank",
-  "DateField",
-  "TimeField",
-  "NumberField",
-  "Rate",
-  "DropDown",
-  "Cascade",
-  "Ordering",
-  "Breaker"
-]New value: +[
+  "SingleCheck",
+  "MultiCheck",
+  "TrueFalse",
+  "FillBlank",
+  "DateField",
+  "TimeField",
+  "NumberField",
+  "Rate",
+  "DropDown",
+  "Cascade",
+  "Ordering",
+  "Breaker",
+  "Statement",
+  "Swiper"
+]

First observedAug 24, 2026

TDQS

A4.6/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations say only that the tool is not read-only and not destructive, so the description carries the full burden — and it delivers. The non-idempotency warning ('if the call times out it may still have succeeded... retrying blindly can create a duplicate') is exactly the behavioral disclosure an agent cannot infer. It also clarifies ignore-vs-reject semantics ('other fields are ignored', 'rejected for other question types'), preventing misuse.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The first sentence names the core operation, and the long body earns its length because the tool genuinely spans 14 enum values and three scoring scenes. The nested parentheticals ('scored via correctAnswer, an unscored data-collection field in scored_quiz') reduce scanability, so a structured per-type summary would be clearer — but nothing in the prose is wasted.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 28-parameter tool with nested objects and complex scene/type interactions, the description plus the schema's own field-applicability summary cover every call-shaping decision: per-type required fields, scene restrictions, rejection behavior, and retry safety. An output schema exists, so return-value documentation is appropriately left out of the description's job.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema already documents every one of the 28 parameters, but the description adds cross-field semantics the schema cannot: which fields to omit per type, the correctAnswer-shape dependency on `multiple`, and the scene-level interplay (e.g., 'the submitted number feeds report formulas' for NumberField). This meaningfully raises the value above the baseline 3 that full schema coverage alone earns.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Opens with a specific verb plus resource — 'Append an item to the end of a form' — and pins the position ('end'), which sets it apart from the sibling insert_question without needing to name it. The body then enumerates the 14 item types and contrasts them with display-only blocks, so an agent knows exactly what this tool creates versus create_form.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives strong situational guidance: prefer DropDown over SingleCheck/MultiCheck past 20 choices, maps each question type and field to its scene (quiz / scored_quiz / outcome_quiz), and routes random_knowledge_quiz bank setup to the web app as an explicit exclusion. It does not name sibling insertion/update tools as alternatives, so the append-vs-insert contrast is implied rather than stated — the only gap from a 5.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All RooQuiz tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A4.1/5.0

Disambiguation4/5

Most tools map to a distinct resource and action, and the descriptions actively disambiguate similar operations (e.g., get_form vs get_form_share_info vs get_form_stats). The main potential confusion is between list_records and list_leads and between get_record and get_lead, since both describe leads from slightly different angles.

Naming Consistency5/5

All tool names follow a consistent snake_case verb_noun convention, with resources like form, question, lead, booking, examinee, translation, and tenant parallel across actions. The few non-CRUD verbs like prepare_/finalize_, duplicate_, and reschedule_ still fit the same uniform pattern.

Tool Count2/5

48 tools is well beyond the 3–15 ideal and even past the 25+ threshold, making the tool surface heavy for an agent to navigate. The broad platform scope explains some of the size, but the count still risks overwhelming context and increasing misselection.

Completeness4/5

The server covers form lifecycle, question editing, translations, CRM leads, examinees, records, bookings, team/tenant operations, image uploads, templates, and analytics extremely well. Minor gaps remain—such as no lead/record deletion, no member removal or role updates, and limited booking-settings management—but most workflows can be completed with the existing tools.

Resources

Need Help?