Integrations

Mastra

Mastra is a TypeScript framework for building AI agents. Connect it to DEVUP AI through @ai-sdk/openai-compatible.

Prerequisites

Node.js
A DEVUP AI API key from https://devupai.com/dashboard/api-keys.

Install

bash

npm install @mastra/core @ai-sdk/openai-compatible zod

The examples use top-level await; save them in a .mts file. Node.js 24 runs .mts files directly: node app.mts.

Set your API key

export DEVUP_API_KEY="your-key"

Create an agent

typescript

import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
import { Agent } from "@mastra/core/agent";

const devupai = createOpenAICompatible({
  name: "devupai",
  baseURL: "https://api.devupai.com/v1",
  apiKey: process.env.DEVUP_API_KEY,
});

const agent = new Agent({
  id: "assistant",
  name: "Assistant",
  instructions: "Be concise.",
  model: devupai("deepseek-ai/DeepSeek-V4-Pro"),
});

const result = await agent.generate("Reply with exactly: Salam DEVUP");
console.log(result.text);

Output: Salam DEVUP

Streaming

typescript

const stream = await agent.stream("Count from 1 to 5, separated by spaces.");
for await (const chunk of stream.textStream) {
  process.stdout.write(chunk);
}
console.log();

Output: 1 2 3 4 5

Tool calling

Define a tool with createTool and a zod schema, then pass it in tools.

typescript

import { createTool } from "@mastra/core/tools";
import { z } from "zod";

const getOrderStatus = createTool({
  id: "getOrderStatus",
  description: "Look up the delivery status of an order by its ID.",
  inputSchema: z.object({ orderId: z.string() }),
  execute: async ({ orderId }) => ({ orderId, status: "shipped" }),
});

const supportAgent = new Agent({
  id: "support",
  name: "Support",
  instructions: "Answer order questions with the getOrderStatus tool.",
  model: devupai("deepseek-ai/DeepSeek-V4-Pro"),
  tools: { getOrderStatus },
});

const orderResult = await supportAgent.generate("What is the status of order A-100?");
console.log(orderResult.toolCalls.map((call) => call.payload.toolName), orderResult.text);

The agent calls getOrderStatus and answers that order A-100 has shipped.

Structured output

Pass a zod schema as structuredOutput. jsonPromptInjection: true puts the schema in the prompt so the model returns the fields you asked for.

typescript

const city = await agent.generate("Which city is the capital of Algeria?", {
  structuredOutput: {
    schema: z.object({ name: z.string(), country: z.string() }),
    jsonPromptInjection: true,
  },
});
console.log(city.object);

Output: { name: 'Algiers', country: 'Algeria' }

Choosing models

Use the exact catalog ID as the model name. Tool calling needs a model whose catalog entry supports it. See Models and Tool Calling.

Troubleshooting

Authentication error (401)Set DEVUP_API_KEY in the same terminal that runs your script, without a Bearer prefix.
Structured output fails validationSet jsonPromptInjection: true in structuredOutput. Without it the model is only asked for a JSON object and may choose its own keys.
The agent answers without calling your toolModels can answer simple questions on their own. Use tools for data the model cannot know, and a model whose catalog entry supports tool calling.

LlamaIndex

n8n