mathgraphs-mcp

mathgraphs-mcp

An agentic MCP for math and statistics graphing, computation, visualization, and validation.

MathTalking computes and renders interactive math visualizations. Plot functions, geometry, histograms, regression curves, and more. Results are graph-verified — roots, extrema, and intersections are computed from actual plotted curves, not generated by AI.

Saves ~410 tokens per render vs LLM-generated canvas code. Zero LLM inference needed — the engine computes, not generates.

Endpoint

https://mathtalking.com/api/mcp

Tools

plot_graph — Math Visualization

Plot functions, points, segments, labels, and shapes on an interactive graph. Auto-computes roots, extrema, and intersections for plotted functions.

Input: Array of elements, each with a type and type-specific fields:

Type	Fields	Example
function	expression (required), color	{"type":"function", "expression":"x^2-4", "color":"#4A90D9"}
points	points [{x,y}] (required), color, label	{"type":"points", "points":[{"x":2,"y":0}], "label":"root"}
segment	x1,y1,x2,y2 (required), color, label, arrow, dashed	{"type":"segment", "x1":0, "y1":0, "x2":3, "y2":4, "label":"hypotenuse"}
label	text (required), x,y (required), color, fontSize	{"type":"label", "text":"vertex", "x":0, "y":-4}
triangle	x1,y1,x2,y2,x3,y3 (required), color, opacity, border	{"type":"triangle", "x1":0, "y1":0, "x2":3, "y2":0, "x3":3, "y3":4}
box	x1,y1,x2,y2 (required), height (required), color, opacity	{"type":"box", "x1":0, "y1":0, "x2":1, "y2":0, "height":5}

Supports: Explicit y=f(x), implicit x²+y²=1, parametric (cos(t),sin(t)), piecewise, domain restrictions.

Output formats: output param accepts "url" (default, interactive page), "embed" (iframe URL), or "svg" (vector image as text).

Returns: Interactive URL with zoom, pan, and sliders. Or SVG/embed URL depending on output param.

create_show — Slideshow

Bundle multiple plot_graph results into a slideshow. First create each slide with plot_graph, then pass the render IDs here.

Input: title (string), slide_ids (array of render IDs from plot_graph results).

Returns: Interactive show URL with prev/next navigation, keyboard controls, and auto-play.

compute_stats — Descriptive Statistics

Compute mean, median, standard deviation, min, max, range, quartiles from raw data.

add_histogram — Histogram

Create auto-binned histogram from raw numerical data.

add_regression — Regression

Fit linear, quadratic, exponential, or power regression to data points. Returns slope, intercept, R².

fit_distribution — Distribution Fitting

Fit normal, uniform, or exponential distribution to data. Returns best-fit parameters.

test_hypothesis — Hypothesis Testing

Run t-test, paired t-test, or one-sample t-test. Returns test statistic, p-value, and significance.

When to Use

• User asks to graph, plot, or visualize any math
• You need to verify a mathematical result visually (roots, intersections, extrema)
• You computed an answer and want to show it, not just describe it
• You need a multi-step explanation — create slides with plot_graph, bundle with create_show
• Data needs statistical visualization (histogram, regression, distribution fit)
• Geometry needs precise rendering (triangles, circles, constructions)

Why Use This

• Computed, not hallucinated — roots from actual zero-crossings, regression from least squares
• Interactive — user can zoom, pan, adjust sliders, explore
• Shareable — each graph gets a permanent URL
• 9 languages — en, zh, zh-TW, ja, ko, es, fr, de, pt-BR

Quick Test

curl -X POST https://mathtalking.com/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"plot_graph","arguments":{"elements":[{"type":"function","expression":"sin(x)","color":"#4A90D9"}]}}}'

Links

• Website: mathtalking.com
• Interactive demo: mathtalking.com/math/quadratic-en
• API endpoint: POST https://mathtalking.com/api/render (REST) or POST https://mathtalking.com/api/mcp (MCP)