Public API is available. Convert Markdown to images with 50 watermarked requests/month.
View API Docs

Automation

Give your AI agent a Markdown-to-image tool

One REST endpoint, one bearer token. Any framework that can call a function or import an OpenAPI schema can let a model render its own Markdown into an image.

Why

What people build with it

01

Let an assistant answer with a rendered table or diagram instead of raw Markdown the user has to squint at.

02

Give a reporting agent the ability to produce a PDF at the end of a research run.

03

Let a content agent generate and publish social cards without a human formatting step in between.

Setup

Connecting AI agents

  1. 01

    Create an API token under Profile → API tokens. It starts with mti_ and is shown only once.

  2. 02

    For frameworks that read a tool descriptor, this endpoint returns the tool name, parameters, and endpoint in a compact JSON format.

    GET https://markdowntoimage.com/api/v1/schema
  3. 03

    For anything that imports OpenAPI 3.x — most agent platforms — point it at this URL instead.

    https://markdowntoimage.com/api/v1/openapi.json
  4. 04

    Underneath it is a plain REST call, so any framework that can make an HTTP request can use it without an adapter.

    curl -X POST https://markdowntoimage.com/api/v1/images/generate \
  -H "Authorization: Bearer mti_your_token_here" \
  -H "Content-Type: application/json" \
  -d '{"markdown": "# Hello", "format": "png", "width": 800}'

Every parameter — themes, fonts, code styles, page size for PDF — is listed in the API reference.

Other integrations

n8nGenerate images from Markdown in n8nDifyAdd a Markdown-to-image tool to DifyMakeConvert Markdown to images in MakeZapierTurn Markdown into images with Zapier

Troubleshooting

Support
Q1

Which frameworks does this work with?

Any that can call an HTTP endpoint — function calling on the major model APIs, OpenAPI tool imports, and code-first agent frameworks. There is no SDK to install because there is nothing framework-specific about a POST request.
Q2

Should the model choose the format and theme?

It can. format accepts png, jpeg, webp, and pdf, and theme and codeStyle are plain strings, so you can expose them in the tool definition and let the model decide. Omit them and sensible defaults apply.
Q3

How do I keep an agent from burning through quota?

Call GET /api/v1/account/usage to read remaining quota, and gate the tool on it. The generate endpoint also returns quota state in meta on every successful call, so an agent can self-limit without a second request.
Q4

Does the model need to handle the image bytes?

The response returns a hosted URL. If you need the file in your own storage, either download it in the following step of your workflow, or set "mode" to "binary" to get the bytes back directly and write them wherever you want.
Markdown to Image for AI Agents | Tool & Function Calling | MarkdownToImage