The Tintly API: generate palettes and check contrast from your own code

The same engine that powers Tintly's color tools is available over HTTPS. You send JSON, you get JSON back, and every answer comes from one of 13 tools: generate a palette, measure contrast, describe a color, preview color blindness, build a Tailwind scale and more. It is free, read-only and limited to 60 requests a minute per key.

1. 1
   Create a key

   On the developer page, sign in and create a key. Keys start with tnt_live_ and are shown once.

2. 2
   Send the key as a Bearer token

   Put it in the Authorization header on every request. An X-API-Key header also works.

3. 3
   Call a tool

   POST to /api/v1/tools/<name> with the tool's inputs as a JSON body. The answer arrives under result.

Your first request

curl -X POST https://www.trycolorkit.com/api/v1/tools/check_contrast \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"foreground":"#8A8A8A","background":"#FFFFFF"}'

The response gives the ratio, the strongest WCAG level reached, a pass or fail for AA and AAA at normal and large sizes, and, when the pair fails, the closest foreground color of the same hue that does pass. A mid-gray on white comes back at about 3.45:1: fine for large headings, not for body text, with a slightly darker gray suggested.

The same call from JavaScript

const res = await fetch("https://www.trycolorkit.com/api/v1/tools/generate_palette", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.TINTLY_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ base_color: "#2463eb", harmony: "triadic" }),
});
if (!res.ok) throw new Error((await res.json()).error.message);
const { result } = await res.json();
console.log(result.colors, result.css);

What the 13 tools do

GET /api/v1/tools lists every tool with its description and JSON Schema, so you can generate a client or validate inputs before you send them.

Errors and limits

Designing around the limit

Sixty requests a minute is plenty for a build step, a design-system script or an internal tool, and it is deliberately modest for a public website. If you find yourself calling the API on every page view, move the call into your build, or cache the answer on your server. Calls that return the same result for the same input are the easiest to cache, and they are most of what the API does. When you do hit the limit, the Retry-After header tells you exactly how long to wait, and the X-RateLimit-Remaining header on every response lets you slow down before you get there.

A small example: tokens from a brand color

A short script can turn one brand color into a full set of design tokens: call tailwind_scale for the 50 to 950 steps, then find_accessible_pairs on the steps you plan to use for backgrounds and text, and write only the pairs that reach AA. Because the tool measures contrast, the script can fail your build when a token pair regresses, which is a far better time to find out than after launch.

1. Call tailwind_scale with your brand color and a scale name.
2. Pick the steps for background, surface, text and accent.
3. Call find_accessible_pairs with those colors at level AA.
4. Write the CSS variables, and exit with an error if a required pair is missing.

Prefer to talk to the tools in chat instead of code? The same 13 tools are available to AI assistants through the MCP server, with the same key.