> ## Documentation Index > Fetch the complete documentation index at: https://dev.polltheroom.com/llms.txt > Use this file to discover all available pages before exploring further. # Screenshot a step > A PNG of one step as the room will see it: the slide's page with its cards, text, shapes and images on it, or the full-screen poll the way the audience screen draws it. Beside the picture come the numbers: every element's rect, the pairs that overlap, `warnings` for a card too small to read or over another card or text box, `coversPageText`, the PDF page's own words a card or text box hides, and `measured`, each card and text box as the drawing measured it: cut off by its box, and how high its biggest answer text is drawn. Send `annotate` to have every element's box and name drawn on the picture, the clashing ones in red and the hidden words tinted; `element` to crop to one element, the way Paper screenshots one node; `scale` 2 for twice the pixels. Rendered on the server in a real browser, so the first look takes a few seconds; the same step at the same revision comes straight back. Through MCP the image arrives as an image; over HTTP, read `png` as a data URL or send `Accept: image/png` for the bytes. Needs the `decks:read` scope. ## OpenAPI ````yaml https://test.polltheroom.com/openapi.json get /api/v1/decks/{id}/steps/{index}/screenshot openapi: 3.1.0 info: title: Poll the Room API version: '1' description: >- Live polls on a presenter's own slides. A presenter (or an agent acting for one) builds a deck; the audience joins on their phones with a six-digit code and never signs up. Keys are made under Settings › API and sent as `Authorization: Bearer ptr_…`. A key belongs to a team, carries scopes, and is rate limited per hour. Plans: Free may decks:read, decks:write, rooms:read, results:read at 60 requests an hour; Pro may decks:read, decks:write, rooms:read, rooms:start, results:read at 1000 requests an hour. The same verbs are MCP tools at https://test.polltheroom.com/api/mcp. The full reference is at https://test.polltheroom.com/llms-full.txt. servers: - url: https://test.polltheroom.com security: - apiKey: [] tags: - name: decks description: A deck is a document the editor and the API both read and write. - name: rooms description: A room is one live session of a deck. paths: /api/v1/decks/{id}/steps/{index}/screenshot: get: tags: - decks summary: Screenshot a step description: >- A PNG of one step as the room will see it: the slide's page with its cards, text, shapes and images on it, or the full-screen poll the way the audience screen draws it. Beside the picture come the numbers: every element's rect, the pairs that overlap, `warnings` for a card too small to read or over another card or text box, `coversPageText`, the PDF page's own words a card or text box hides, and `measured`, each card and text box as the drawing measured it: cut off by its box, and how high its biggest answer text is drawn. Send `annotate` to have every element's box and name drawn on the picture, the clashing ones in red and the hidden words tinted; `element` to crop to one element, the way Paper screenshots one node; `scale` 2 for twice the pixels. Rendered on the server in a real browser, so the first look takes a few seconds; the same step at the same revision comes straight back. Through MCP the image arrives as an image; over HTTP, read `png` as a data URL or send `Accept: image/png` for the bytes. Needs the `decks:read` scope. operationId: get_screenshot parameters: - name: id in: path required: true schema: type: string minLength: 1 description: The deck's id - name: index in: path required: true schema: type: integer minimum: 1 maximum: 9007199254740991 description: The step's index; the join screen (0) has no picture - name: scale in: query required: false schema: description: 1 (the default) for 1600 pixels wide, 2 for 3200 type: integer minimum: 1 maximum: 2 - name: element in: query required: false schema: description: 'Crop to this element, as get_step names it: "overlay 0", "text 1"' type: string pattern: ^(overlay|text|shape|image) \d+$ - name: annotate in: query required: false schema: description: Draw every element's box and name on the picture, clashes in red anyOf: - type: boolean - type: string enum: - 'true' - 'false' responses: '200': description: OK content: application/json: schema: type: object properties: png: type: string description: The image, as a data URL width: type: integer minimum: -9007199254740991 maximum: 9007199254740991 description: Pixels height: type: integer minimum: -9007199254740991 maximum: 9007199254740991 description: Pixels revision: type: string description: The deck's revision this picture shows elements: type: array items: type: object properties: id: type: string description: >- "overlay 0", "text 1": the list and the index, as the step verbs address it kind: type: string enum: - overlay - text - shape - image index: type: integer minimum: -9007199254740991 maximum: 9007199254740991 rect: type: object properties: x: type: number minimum: 0 maximum: 1 'y': type: number minimum: 0 maximum: 1 w: type: number minimum: 0.02 maximum: 1 h: type: number minimum: 0.02 maximum: 1 required: - x - 'y' - w - h additionalProperties: false z: type: number description: Stacking order; higher is on top label: type: string description: >- The card's question, the text box's first words, the shape's kind, or image required: - id - kind - index - rect - z - label additionalProperties: false overlaps: type: array items: type: object properties: a: type: string b: type: string covered: type: number description: >- How much of the smaller element the other covers, 0 to 1 required: - a - b - covered additionalProperties: false description: >- Pairs of elements that sit on top of each other; empty when nothing overlaps warnings: type: array items: type: object properties: element: type: string description: The element it is about, "overlay 0", "text 1" with: description: The other element, for an overlap type: string message: type: string description: What is wrong and what to do required: - element - message additionalProperties: false description: >- What on this slide will not read well: a card too small to read, a card or text box over another coversPageText: type: array items: type: object properties: element: type: string text: type: string description: The first of the page's words it hides hidden: type: number description: How much of those lines it hides, 0 to 1 required: - element - text - hidden additionalProperties: false description: >- Cards and text boxes over the PDF page's own text. Words drawn inside a picture on the page are pixels, not text, so look at the picture for those measured: type: array items: type: object properties: element: type: string clipped: type: boolean description: Something inside is cut off by its box overflowPx: type: number description: >- By how many pixels the content outgrows its box; a few are padding textPx: anyOf: - type: number - type: 'null' description: >- How high a card's biggest answer text is drawn, in pixels on the 1600-pixel slide; null for a text box or a card with no answer text required: - element - clipped - overflowPx - textPx additionalProperties: false description: >- Each card and text box as the drawing measured it after paint. How small a card can go depends on its look, so this, not a fraction, says whether it fits required: - png - width - height - revision - elements - overlaps - warnings - coversPageText - measured additionalProperties: false '401': description: No usable key. content: application/json: schema: $ref: '#/components/schemas/Error' '403': description: The key lacks the scope, or the plan does not sell it. content: application/json: schema: $ref: '#/components/schemas/Error' '404': description: Nothing by that id within this key's reach. content: application/json: schema: $ref: '#/components/schemas/Error' '429': description: Over the key's hourly limit. content: application/json: schema: $ref: '#/components/schemas/Error' components: schemas: Error: type: object properties: error: type: string description: What went wrong, in a sentence fix: description: What to do about it, when we can tell type: string required: - error additionalProperties: false securitySchemes: apiKey: type: http scheme: bearer description: >- An API key from Settings › API, sent as `Authorization: Bearer ptr_…`. Shown once when made; only a hash is stored. ````