Skip to content
Divinci AI SDK

Divinci AI SDK

Build powerful AI integrations with our official TypeScript SDKs

Get Started View on GitHub

See the whole pipeline

Section titled “See the whole pipeline”

New here? Follow one PDF from upload to a grounded answer — parsed, chunked, quizzed, embedded, routed, attacked, guarded, and served. Every stage is a real stage of the Divinci pipeline, and every tool named is one you can actually pick.

One document. The whole pipeline. → A scroll-through walkthrough of all seven stages, with the vendors you can choose at each one.

Bring your coding agent

Section titled “Bring your coding agent”

Building with Divinci in Claude Code, Cursor, Gemini CLI, or Grok? Install the Divinci Agent Skill — a drop-in package that teaches your agent the CLI, SDKs, and API (auth patterns, RAG ingestion, releases, the footguns) — and connect the hosted MCP server for live access to your account:

Terminal window
claude mcp add divinci --transport http https://mcp.divinci.app/mcp
Install the Agent Skill One zip, works in Claude Code and any SKILL.md-compatible agent.
Connect via MCP Give your assistant live access to workspaces, RAG, and releases.

Choose Your SDK

Section titled “Choose Your SDK”

Client SDK

Build custom chat UIs in React, Vue, or vanilla JavaScript with real-time streaming.

Learn more →

Server SDK

Manage workspaces, releases, and RAG documents from your Node.js backend.

Learn more →

MCP SDK

Integrate with Claude Desktop, Cursor, and other MCP-compatible AI assistants.

Learn more →

Embed Script

Drop a fully-rendered chat widget onto any page with one <script> tag — no bundler required.

Learn more →

CLI

Manage workspaces, chat with AI, search knowledge bases, and control releases from your terminal.

View CLI docs →

TypeScript First

Full TypeScript support with auto-generated types from our OpenAPI spec.

View API reference →

Quick Example

Section titled “Quick Example”
import { DivinciClient } from "@divinci-ai/client";


const client = new DivinciClient({
  releaseId: "rel_abc123",
  apiKey: "divinci_key_...",
});


// Open a thread, then stream a response token-by-token
const thread = await client.chat.create();


await client.chat.stream(thread.id, "Tell me about AI", {
  assistantName: "your-assistant-id", // the release's assistant/model id
  onChunk: (text) => process.stdout.write(text),
  onComplete: (message) => console.log("\n\nDone:", message.id),
});

Example projects

Section titled “Example projects”