Opper Node SDK

Build AI-powered applications in TypeScript with declarative task completion, structured outputs, knowledge bases, streaming, and full observability.

Installation

npm add opperai
# or: pnpm add opperai / yarn add opperai / bun add opperai

Set your API key:

export OPPER_HTTP_BEARER="your-api-key"

Get your API key from platform.opper.ai.

Core Pattern: Task Completion

The opper.call() method is the primary interface. Describe a task and get structured results:

import { Opper } from "opperai";

const opper = new Opper({
  httpBearer: process.env["OPPER_HTTP_BEARER"] ?? "",
});

// Simple call with structured output
const result = await opper.call({
  name: "extractRoom",
  instructions: "Extract details about the room from the provided text",
  input: "The Grand Hotel offers a luxurious suite with 3 rooms and an ocean view.",
  outputSchema: {
    type: "object",
    properties: {
      room_count: { type: "number" },
      view: { type: "string" },
      hotel_name: { type: "string" },
    },
    required: ["room_count", "view", "hotel_name"],
  },
});

console.log(result.jsonPayload);
// { room_count: 3, view: "ocean", hotel_name: "The Grand Hotel" }

Structured Output with JSON Schema

Define input and output schemas for type-safe interactions:

const result = await opper.call({
  name: "analyze_sentiment",
  instructions: "Analyze the sentiment of the given text",
  inputSchema: {
    type: "object",
    properties: {
      text: { type: "string", description: "Text to analyze" },
    },
    required: ["text"],
  },
  outputSchema: {
    type: "object",
    properties: {
      label: { type: "string", enum: ["positive", "negative", "neutral"] },
      confidence: { type: "number", description: "0.0 to 1.0" },
      reasoning: { type: "string" },
    },
    required: ["label", "confidence"],
  },
  input: { text: "I love this product!" },
});

console.log(result.jsonPayload);
// { label: "positive", confidence: 0.95, reasoning: "..." }

Model Selection

Control which LLM to use with the model parameter:

const result = await opper.call({
  name: "generate",
  instructions: "Write a haiku about the given topic",
  input: "autumn leaves",
  model: "anthropic/claude-4-sonnet",
});

Available models include providers like openai/, anthropic/, google/, and more.

Few-Shot Examples

Guide model behavior with input/output examples:

const result = await opper.call({
  name: "classify_ticket",
  instructions: "Classify the support ticket category",
  input: { text: "My payment was declined" },
  examples: [
    { input: { text: "I can't log in" }, output: "authentication" },
    { input: { text: "Charged twice" }, output: "billing" },
    { input: { text: "App crashes" }, output: "bug", comment: "Technical issues" },
  ],
});

Streaming

Stream responses token-by-token using opper.stream():

const outer = await opper.stream({
  name: "write_story",
  instructions: "Write a short story about the given topic",
  input: "a robot learning to paint",
  outputSchema: {
    type: "object",
    properties: {
      title: { type: "string" },
      story: { type: "string" },
    },
    required: ["title", "story"],
  },
});

// Access the result stream directly
const stream = outer.result;

for await (const event of stream) {
  const delta = event.data?.delta;
  if (delta) {
    process.stdout.write(delta);
  }
}

Tracing with Spans

Track operations by passing parentSpanId:

// Create a span for grouping related calls
const result = await opper.call({
  name: "step_one",
  instructions: "First step of the pipeline",
  input: "data",
  parentSpanId: "123e4567-e89b-12d3-a456-426614174000",
  tags: {
    project: "my_project",
    user: "user_123",
  },
});

Tags and Metadata

Add metadata for filtering and cost attribution:

const result = await opper.call({
  name: "translate",
  instructions: "Translate to French",
  input: "Hello world",
  tags: {
    project: "website",
    environment: "production",
    user_id: "usr_123",
  },
});

Error Handling

import { Opper } from "opperai";
import * as errors from "opperai/models/errors";

try {
  const result = await opper.call({ name: "task", instructions: "...", input: "..." });
} catch (error) {
  if (error instanceof errors.OpperError) {
    console.error(`Status ${error.statusCode}: ${error.message}`);
    if (error instanceof errors.BadRequestError) {
      console.error("Details:", error.data$.detail);
    }
  }
}

Common Mistakes

Missing httpBearer: The SDK won't authenticate without it. Use environment variable OPPER_HTTP_BEARER.
Forgetting required in schemas: JSON Schema fields are optional by default. Always specify required.
Not awaiting calls: All SDK methods are async. Always await them.

Additional Resources

For function CRUD operations and versioning, see references/FUNCTIONS.md
For knowledge base operations and RAG, see references/KNOWLEDGE.md
For tracing and span operations, see references/TRACING.md
For advanced streaming patterns, see references/STREAMING.md
For migrating from OpenRouter, see references/MIGRATION.md

Related Skills

opper-node-agents: Use when you need autonomous agents with tool use, reasoning loops, and multi-step execution rather than single-shot task completion.
opper-python-sdk: Use when building with Python instead of TypeScript.

Upstream Sources

When this skill's content may be outdated, resolve using this priority:

Installed package source — check the user's project first, as it reflects the exact version in use: node_modules/opperai/
Source code: https://github.com/opper-ai/opper-node
Documentation: https://docs.opper.ai

opper-node-sdk