Quickstart

Prepare an API Key. Save the code below as quickstart.mjs and run it with Node.js. It creates a library, uploads an image, searches by tag, and produces delivery URLs.

npm install cloudinary@2
# Set PRAXIS_API_KEY in your environment.
node quickstart.mjs

import { v2 as cloudinary } from "cloudinary";

const base = "https://api.deepraxis.ai";
const key = process.env.PRAXIS_API_KEY;
if (!key) throw new Error("Set PRAXIS_API_KEY");
async function media(path, method = "GET", body) {
  const response = await fetch(base + "/media/v1" + path, {
    method,
    headers: { Authorization: `Bearer ${key}`, "Content-Type": "application/json" },
    ...(body === undefined ? {} : { body: JSON.stringify(body) }),
  });
  if (!response.ok) throw new Error(`Media HTTP ${response.status}`);
  return response.json();
}
const library = await media("/libraries", "POST", { name: "Agent media" });
cloudinary.config({
  cloud_name: library.id,
  oauth_token: key,
  upload_prefix: base + "/cloudinary",
});
const asset = await cloudinary.uploader.upload(
  "data:image/png;base64,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",
  { public_id: "quickstart/pixel", tags: ["docs"] },
);
const found = await cloudinary.search.expression("resource_type:image AND tags:docs").execute();
const delivery = asset.secure_url;
const thumbnail = delivery.replace("/upload/", "/upload/w_160,c_fit/");
const understandings = await media(`/libraries/${library.id}/assets/${asset.asset_id}/understandings`);
const current = await media(`/libraries/${library.id}/assets/${asset.asset_id}`);
const matches = current.semantic === "ready"
  ? await media(`/libraries/${library.id}/search/semantic`, "POST", { query: "a small image", top_k: 5 })
  : null;
console.log({ library_id: library.id, asset_id: asset.asset_id, found, delivery, thumbnail, understandings, matches });

Understanding completes asynchronously. Upload responses carry no processing state; GET the asset to read processing and semantic. Measured scale: about half a minute for an image, about 50 seconds for a 3-second video, and immediately for raw files (no understanding). Initial results may be empty. When matches is null, read the asset again later and search once semantic is ready. Understanding and semantic search are available. Uploading, expression search, and original access do not need to wait for understanding.

The example keeps the library. When finished, set LIBRARY_ID to the returned library_id and delete it:

curl -X DELETE "https://api.deepraxis.ai/media/v1/libraries/${LIBRARY_ID}" \
  -H "Authorization: Bearer $PRAXIS_API_KEY"

API reference →

Quickstart

On this page