Treyspace SDK

The Treyspace SDK is an open-source framework designed for Retrieval-Augmented Generation (RAG) specifically focused on canvas-based knowledge graphs. This SDK efficiently ingests Excalidraw canvas data and integrates it into the Helix graph-vector database, facilitating advanced semantic, relational, and spatial clustering. Enhanced by large language model (LLM) capabilities, it provides intelligent analysis and insights for users.

Key Features

• Responses API: A robust wrapper around the OpenAI responses API, optimized for handling canvas-specific contexts.
• Canvas AI Engine: Server-Sent Events (SSE) endpoints under /api/ai/* that orchestrate a comprehensive RAG pipeline with graph-based retrieval.
• Helix Bridge: Access the Helix graph database through endpoints /api/clusters and /api/mcp-bridge for effective semantic clustering.
• SDK & Server: Utilize the SDK either as a library in applications or run it as a standalone server to suit various development needs.
• Development-Focused: Streamlined for adaptability, with no authentication, rate limiting, or billing concerns, making it convenient for developers.

Users can access a hosted version with a complete user interface and an integrated Excalidraw canvas at Treyspace.

Example Usage

To demonstrate the capabilities of the SDK, here are some basic commands:

node examples/responses.mjs "Summarise this Excalidraw canvas"

OR to load a complete request:

node examples/responses.mjs --payload ./payload.json

For advanced functionality, SDK usage as a library can be easily implemented:

import { createHelixRagSDK, executeFullPipeline } from 'treyspace-sdk';

// Create SDK instance
const sdk = createHelixRagSDK();

// Sync canvas data
await sdk.syncCanvas({
  boardId: 'my-board',
  excalidrawData: { elements: [...] }
});

// Get clusters
const clusters = await sdk.refreshClusters({ boardId: 'my-board' });

// Or run full pipeline
const result = await executeFullPipeline({
  boardId: 'my-board',
  userMessage: 'Explain this diagram',
  elements: canvasElements
});

console.log(result.text);

For additional details, refer to the API Reference for a comprehensive overview of endpoints, or explore the Pipeline Guide for an in-depth step-by-step workflow on processing canvas data.

Explore and contribute to the project through GitHub, and connect with the community on Twitter or join the Discord community. Join a growing ecosystem focused on enhancing knowledge graph interactions through innovative canvas technologies.