Mathpix connector for AI agents

Tools your research agent reaches for on Mathpix, scoped per user.

CALL ANY TOOL

15 tools for document OCR with Mathpix: read math, text, tables, and handwriting from images and ink strokes, process PDF and Word files, convert Mathpix Markdown to DOCX, LaTeX, and HTML, and check usage.

mathpix_convert_markdown

Convert markdown

Convert a Mathpix Markdown document into DOCX, Markdown, HTML, PDF, LaTeX, PowerPoint or Excel. Returns conversion_id, the handle for status checks and downloads. Use this for Markdown you already have. Use the conversion_formats input of mathpix_submit_pdf to convert a document while it is processed. Poll mathpix_get_conversion_status until each format is completed, then read text formats with mathpix_get_conversion_result.

Parameters

Name

Type

Required

Description

formats

array

Required

Output formats to produce. md and html are text and can be read with mathpix_get_conversion_result; docx, xlsx (tables only), pptx, pdf, latex.pdf and the zip formats are binary files. Example: ["docx", "tex.zip"]

mmd

string

Required

The Mathpix Markdown (MMD) document to convert, as one string. MMD is Markdown with LaTeX math in  and \[ \] delimiters, as returned by mathpix_process_image or mathpix_get_pdf_result. The request body can be up to 10 MB. Example: # Notes\n\nThe area is $ \pi r^2 $.

conversion_options

object

Optional

Settings for specific formats, keyed by format name. For example docx takes font, fontSize (half-points), orientation and margins; md takes math delimiters and escape flags; pdf takes fontSize, margin and custom_css; html takes math_output_format (svg, mathml or latex). Example: {"docx": {"font": "Arial", "fontSize": 24}}

improve_mathpix

boolean

Optional

Set to false to stop Mathpix from keeping this request's image and result to improve its models (sent as metadata.improve_mathpix). Results sent with false do not appear in mathpix_list_ocr_results.

mathpix_delete_conversion

Delete conversion

mathpix_delete_pdf

Delete pdf

mathpix_get_conversion_result

Get conversion result

mathpix_get_conversion_status

Get conversion status

mathpix_get_ocr_usage

Get ocr usage

mathpix_get_pdf_lines

Get pdf lines

mathpix_get_pdf_result

Get pdf result

mathpix_get_pdf_status

Get pdf status

mathpix_list_ocr_results

List ocr results

mathpix_list_pdf_results

List pdf results

mathpix_process_equation

Process equation

Read one equation image with Mathpix's legacy equation endpoint and return it in the LaTeX and math notations you choose. Returns the requested formats (for example latex_styled, latex_simplified, mathml, asciimath, wolfram), latex_confidence and, with n_best, alternative candidates. Use this when you need MathML, AsciiMath, Wolfram or simplified LaTeX for a single equation. Use mathpix_process_image for everything else; Mathpix recommends it for new work.

Parameters

Name

Type

Required

Description

formats

array

Required

LaTeX and math formats to return. text and text_display give text with math in delimiters; latex_normal, latex_styled and latex_simplified are LaTeX variants (styled adds \left/\right for appearance, simplified shortens operators for symbolic processing); latex_list splits several equations into a list; mathml, asciimath and wolfram give those notations (wolfram is ready for Wolfram Alpha). Example: ["latex_styled", "asciimath"]

src

string

Required

The equation image to read, as a public image URL (http or https) or a base64 data URL such as data:image/png;base64,iVBORw0KGgo... Supported image types include JPEG, PNG, WebP, BMP and TIFF. Mathpix downloads URLs itself (up to 10 MB, 15 second timeout); a base64 image must be 2 MB or less. Example: https://mathpix-ocr-examples.s3.amazonaws.com/cases_hw.jpg

beam_size

integer

Optional

Number of candidate results (1 to 10) Mathpix considers during recognition. Needed when n_best is set. Example: 5

confidence_threshold

number

Optional

Minimum confidence between 0 and 1. When the result's confidence is below this value, Mathpix returns a math_confidence error instead of a result. Example: 0.5

displaymath_delims

array

Optional

Begin and end delimiters for display math in the text formats (sent as format_options.displaymath_delims), as a two-item list. Example: ["$$", "$$"]

improve_mathpix

boolean

Optional

Set to false to stop Mathpix from keeping this request's image and result to improve its models (sent as metadata.improve_mathpix). Results sent with false do not appear in mathpix_list_ocr_results.

include_detected_alphabets

boolean

Optional

Return detected_alphabets, which flags each alphabet (en, ru, zh and so on) found in the image.

math_delims

array

Optional

Begin and end delimiters for inline math in the text formats (sent as format_options.math_delims), as a two-item list. Example: ["$", "$"]

n_best

integer

Optional

Number of highest-confidence alternatives to return in candidates, from 1 up to beam_size. Requires beam_size. Example: 3

ocr

array

Optional

What to read: ["math"] (the default) reads math only; ["math", "text"] also reads the text around it.

region

object

Optional

Restrict recognition to one rectangle of the image, in pixels. Give all four keys: top_left_x, top_left_y, width and height. Example: {"top_left_x": 0, "top_left_y": 0, "width": 400, "height": 120}

skip_recrop

boolean

Optional

Read the whole image instead of letting Mathpix crop to the area it thinks holds the equation.

tags

array

Optional

Labels stored with this result so it can be found later with mathpix_list_ocr_results. Example: ["homework", "batch-42"]

mathpix_submit_pdf

Submit pdf

Start asynchronous Mathpix OCR of a PDF, Word, PowerPoint or ebook file from a public URL. Returns pdf_id, the handle for status checks and result downloads. Use this for multi-page documents. Use mathpix_process_image for a single image. Poll mathpix_get_pdf_status until status is completed, then read the output with mathpix_get_pdf_result or mathpix_get_pdf_lines.

Parameters

Name

Type

Required

Description

url

string

Required

Public http(s) URL Mathpix downloads the document from. Accepts PDF, DOCX, DOC, PPTX, ODT, WPD, EPUB, MOBI, AZW/AZW3/KFX and DJVU, up to 1 GB. Example: https://cdn.mathpix.com/examples/cs229-notes1.pdf

alphabets_allowed

object

Optional

conversion_formats

array

Optional

Extra output formats to produce once processing finishes, on top of Mathpix Markdown (mmd) and lines.json, which are always produced. md and html are text and can be read with mathpix_get_pdf_result; docx, xlsx, pptx, pdf, latex.pdf, mmd.overlay.pdf and the zip formats are binary files. Example: ["md", "docx"]

conversion_options

object

Optional

Settings for specific conversion formats, keyed by format name. For example docx takes font, fontSize (half-points), orientation and margins; md takes math delimiters and escape flags; pdf takes fontSize, margin and custom_css; html takes math_output_format (svg, mathml or latex). Example: {"docx": {"font": "Arial", "orientation": "landscape"}}

disable_itemize

boolean

Optional

Emit list items as plain lines with their markers (1., -, and so on) instead of \begin{itemize} environments. Mathpix default: false.

disable_lstlisting

boolean

Optional

Emit code and pseudocode as Markdown triple-backtick fences instead of \begin{lstlisting} blocks. Mathpix default: false.

idiomatic_eqn_arrays

boolean

Optional

Use aligned, gathered or cases environments instead of array for a list of equations. Mathpix default: false.

improve_mathpix

boolean

Optional

Set to false to stop Mathpix from keeping this request's image and result to improve its models (sent as metadata.improve_mathpix). Results sent with false do not appear in mathpix_list_ocr_results.

include_chemistry_as_image

boolean

Optional

Return each chemistry diagram as an image crop, with its SMILES in the image alt text. Mathpix default: false.

include_equation_tags

boolean

Optional

Keep equation numbers inside the LaTeX as \tag{1.12}. Setting this to true also turns on idiomatic equation arrays.

include_hyperlinks

boolean

Optional

Keep the http, https and mailto links stored in the PDF as Markdown links in the output. Mathpix default: false.

include_page_breaks

boolean

Optional

Append a \pagebreak marker after each page in the Mathpix Markdown output. Mathpix default: false.

include_smiles

boolean

Optional

Recognize chemistry structure diagrams as SMILES inside <smiles>...</smiles>. Mathpix default for documents: true.

math_display_delimiters

array

Optional

Begin and end delimiters for display (block) math in the text output, as a two-item list. Default is ["\\[", "\\]"]. Example: ["$$", "$$"]

math_inline_delimiters

array

Optional

Begin and end delimiters for inline math in the text output, as a two-item list. Default is ["\$", "\$"]. Example: ["$", "$"]

numbers_default_to_math

boolean

Optional

Always treat numbers as math, so "Answer: 17" becomes "Answer: $ 17 $". Mathpix default: false.

page_ranges

string

Optional

Pages to process, as a comma-separated list of pages and ranges. Negative numbers count from the end. Example: 2,4-6 for pages 2, 4, 5 and 6; 2 - -2 for the second through the next-to-last page.

rm_fonts

boolean

Optional

Remove font commands such as \mathbf and \mathrm from equations in the latex_styled and text outputs. Mathpix default: false.

rm_spaces

boolean

Optional

Remove extra white space from equations in the latex_styled and text outputs. Mathpix default: true.

section_numbering

string

Optional

How to handle section and subsection numbers: preserve keeps existing numbering as is (the Mathpix default), auto numbers sections automatically, remove strips existing numbering. Sent as the matching auto_number_sections, remove_section_numbering or preserve_section_numbering flag. One of: `preserve`, `auto`, `remove`.

Build your Agent

Same auth pattern across LangChain, OpenAI, Anthropic, and Google ADK.

import { ScalekitClient } from "@scalekit-sdk/node";
import { createReactAgent } from "@langchain/langgraph/prebuilt";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

// Mathpix tools scoped to this user
const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["mathpix"], toolNames: [
    "mathpix_process_image",
    "mathpix_submit_pdf",
    "mathpix_get_pdf_result"] },
  pageSize: 100,
});

const agent = createReactAgent({ llm, tools });
await agent.invoke({ messages: [{ role: "user", content: "Convert this lecture notes PDF to Markdown" }] });

import OpenAI from "openai";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const openai = new OpenAI();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["mathpix"] }, pageSize: 100,
});

const res = await openai.chat.completions.create({
  model: "gpt-5",
  messages: [{ role: "user", content: "Convert this lecture notes PDF to Markdown" }],
  tools,
});

// Execute the tool call with the user's vaulted Mathpix API key
await sk.tools.executeTool(res.choices[0].message.tool_calls[0], "user_123");

import Anthropic from "@anthropic-ai/sdk";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const anthropic = new Anthropic();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["mathpix"] }, pageSize: 100,
});

const msg = await anthropic.messages.create({
  model: "claude-sonnet-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Convert this lecture notes PDF to Markdown" }],
  tools,
});

// Tool call runs with the user's vaulted Mathpix API key
await sk.tools.executeTool(msg.content, "user_123");

import { Agent } from "@google/adk/agents";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["mathpix"] }, pageSize: 100,
});

const agent = new Agent({
  name: "research_agent",
  model: "gemini-2.5-pro",
  instruction: "Mathpix tools scoped to this user",
  tools,
});

await agent.run("Convert this lecture notes PDF to Markdown");

Try these prompts

Copy any prompt into your agent. Each maps directly to a Mathpix tool. Click to copy, paste into your agent, done.

Images

Copy the prompt

Copied

Read the equation in this image and return LaTeX.

Copy the prompt

Copied

Extract the table from this photo of a lab notebook.

Copy the prompt

Copied

Recognize this handwritten math from its stroke data.

Documents

Copy the prompt

Copied

Process this textbook chapter PDF from its public URL.

Copy the prompt

Copied

Is the thesis PDF done processing?

Copy the prompt

Copied

Download the Markdown for the thesis PDF.

Conversion and usage

Copy the prompt

Copied

Convert this Mathpix Markdown to a DOCX file.

Copy the prompt

Copied

List the documents we submitted this month.

Copy the prompt

Copied

How many OCR requests did we make last week?

SEE HOW AUTH WORKS

Each user connects their own Mathpix API key once, and Scalekit sends it with every call. Keys stay vaulted, every call is scope checked, and every action is logged.

Authorize

Your user connects

Mathpix

once. We tie it to their identity and the meetings they approved — no shared bot account, no org-wide access

Who:

user ‘A’

when:

Once per user

access:

Limited to user

Store

Their

Mathpix

token lives in a vault scoped to them. User A's meetings are never reachable by an agent acting for user B, even on the same connection

vault:

encrypted

scope:

per-user

tokens:

auto-refreshed

Resolve

When your agent calls a

Mathpix

tool, we fetch the right token server-side. It never touches your agent, never appears in the LLM context, never shows up in your logs

speed:

~40ms

check:

before every call

seen by:

nobody

Audit

Every

Mathpix

tool call is logged — who triggered it, which meeting was fetched, what came back. 90 days of history, tied to the user who authorized it

history:

90 days

export:

SIEM-ready

logged:

every call

Test other agents

See the same per-user auth pattern across other documents and AI connectors.

GTM and RevOps Teams

Competitive intelligence briefing agent

Scans Gong calls for competitor mentions, matches each one to its Notion battlecard, and DMs every affected rep a single Slack digest per cycle. Every call runs as the PMM who owns the briefing, never a shared bot.

Why Scalekit

Secure your agent's access. Connectors ship in minutes

01.

Shared tokens break per-user analytics

A shared Mathpix API key looks fine in a demo. In production every OCR request looks like one service account, and you cannot tell which user triggered it. Scalekit resolves the credential of the actual user who triggered the agent, never a shared bot.

// shared key
audit → bot_service_account

// scalekit
audit → user_abc ✓

02.

Authentication is not authorization

03.

Multi-tenancy is architectural

04.

Mathpix today. Ten connectors tomorrow.

“Our agents act across Salesforce, Gong, Google Drive, and more, on behalf of every customer. Scalekit behind the scenes meant we can keep adding tools without ever rebuilding how credentials or tool calling work.”

Venu Madhav Kattagoni

Head of Engineering / Von

FAQs

Frequently Asked Questions

Does the agent access Mathpix as the user or through a shared API key?

As the user. Each user connects their own Mathpix API key once, and Scalekit sends it with every call. Every OCR request and conversion the agent runs is attributed to that user in your audit trail, not a shared service account.

Where is the Mathpix API key stored?

In Scalekit's managed AES-256 token vault, namespaced per tenant. To rotate it, the user reconnects with a new API key. Revocation is a single dashboard action. Credentials never appear in prompts, logs, or LLM context.

Can I limit what the agent does in Mathpix?

Yes. Filter by tool name in listScopedTools to expose only what you want, for example OCR and read tools without mathpix_delete_pdf and mathpix_delete_conversion, which permanently delete outputs. Scalekit also enforces scope checks before every API call.

What happens when a user revokes Mathpix access?

The connection is invalidated on the next tool call. Subsequent requests for that user fail closed with a clear error. Other users in the tenant remain unaffected. The event is logged for audit.

What can the agent send to Mathpix?

Images of printed or handwritten math, text, tables, and chemistry diagrams through mathpix_process_image, and PDF, Word, PowerPoint, or ebook files from a public URL through mathpix_submit_pdf. mathpix_convert_markdown turns Mathpix Markdown into DOCX, Markdown, HTML, PDF, LaTeX, PowerPoint, or Excel.