v1.6.1 — Observable Runtime, tracer bridge & built-in observers

One canonical wire for every AI provider

Pure · Web-first · Multi-provider AI SDK for TypeScript

Anthropic, OpenAI, xAI Grok, Google Gemini and Vertex AI behind one canonical streaming protocol. Zero runtime dependencies, deterministic by design — runs anywhere fetch runs.

Get startedGitHub
npm install @deuz-sdk/core

One API surface across

  • Anthropic
  • OpenAI
  • xAI Grok
  • Google Gemini
  • Vertex AI
  • Yunwu

Canonical delta stream

Every provider response is normalized to one typed StreamPart stream before your code sees it. No raw SSE ever leaks through.

Agentic tool loop

Parallel, self-healing tool execution with runaway guards, budget stops, approval gates and nested sub-agents.

Edge-safe core

Web APIs only — no node:*, no Buffer, no ambient state. Node, Deno, Bun, Vercel Edge and Cloudflare Workers.

Deterministic & testable

Clock, randomness, fetch and keys are injected through one Dependencies seam. Golden-replay tests, no real network.

Memory, RAG & skills

mem0-style memory pipeline, hybrid BM25 + vector retrieval, and progressive SKILL.md disclosure — all edge-safe.

Structured output & UI wire

Schema-typed generateObject / streamObject, React hooks, and a versioned SSE wire you own end to end.

Streaming in three lines

streamChat returns synchronously and never throws — the request starts lazily on first read. Swap the factory to change provider; nothing else moves.

Quickstart
// swap the factory to swap providers
import { streamChat } from '@deuz-sdk/core';
import { createAnthropic } from '@deuz-sdk/core/anthropic';

const anthropic = createAnthropic({ apiKey });
const res = streamChat({
  model: anthropic('claude-opus-4-8'),
  messages: [{ role: 'user', content: 'Hello!' }],
});

for await (const chunk of res.textStream) {
  process.stdout.write(chunk);
}