svelte-rust-agents-sdk

A multi-agent chat system with a Svelte 5 frontend and Rust backend featuring real-time streaming, LLM observability, and modular worker architecture.

• Real-time streaming — Responses stream token-by-token as they're generated
• Multi-agent pipeline — Frontline, Orchestrator, Workers, Evaluator
• LLM observability — Token usage and response time displayed per message
• Modular workers — Search (Serper), Email (SendGrid), General conversation
• Evaluator loop — Optional quality validation with retry logic

Architecture

┌─────────────────┐     WebSocket      ┌─────────────────────────────────────┐
│                 │◄──────────────────►│              Agent                  │
│  Svelte 5 UI    │    (streaming)     │                                     │
│  (SvelteKit)    │                    │  ┌─────────┐    ┌─────────────┐     │
│                 │                    │  │Frontline│───►│ Orchestrator│     │
└─────────────────┘                    │  └─────────┘    └──────┬──────┘     │
                                       │                        │            │
                                       │         ┌──────────────┼──────────┐ │
                                       │         ▼              ▼          ▼ │
                                       │    ┌────────┐    ┌────────┐  ┌─────┐│
                                       │    │ Search │    │ Email  │  │ Gen ││
                                       │    │(Serper)│    │(SGGrid)│  │     ││
                                       │    └────────┘    └────────┘  └─────┘│
                                       │                        │            │
                                       │                   ┌────▼────┐       │
                                       │                   │Evaluator│       │
                                       │                   └─────────┘       │
                                       └─────────────────────────────────────┘

Technologies Used

• Client: Svelte 5, SvelteKit, TypeScript
• Agent: Rust 1.92, Axum, Tokio
• LLM: OpenAI API

Prerequisites

• Docker & Docker Compose

Environment

Create a .env file in the project root:

# Required
OPENAI_API_KEY=sk-...

# Optional
OPENAI_MODEL=gpt-4o           # Main model for routing decisions
WORKER_MODEL=gpt-4o-mini      # Model for workers
SERPER_API_KEY=...            # For web search (serper.dev)
SENDGRID_API_KEY=...          # For email sending
[email protected]
RUST_LOG=info

Run

docker compose up

• Client: http://localhost:3000
• Agent: http://localhost:8000