Helios Engine v0.3.4
Helios Engine just got a massive upgrade.
This release transforms it into a full-featured Rust framework for intelligent, tool-using, multi-agent systems.
What’s New
Forest of Agents: Multi-agent collaboration with context sharing and task delegation.
RAG System: Retrieval-Augmented Generation powered by InMemory or Qdrant vector stores.
Feature Flags: Build only what you need with optional local/offline model support.
Streaming Support: True token-by-token streaming for remote and local models.
Core Capabilities
Agent System: Define multiple agents with distinct personalities, goals, and tools.
Tool Registry: Extensible system for registering and composing custom tools.
Session Memory: Persistent conversation state, metadata, and context tracking.
Chat Management: Integrated history management and conversation orchestration.
Thinking Tags: Automatic model reasoning visualization.
Dual Mode: Run in Auto, Online (API), or Offline (local model) modes.
Built-In Tools
File Tools: Search, read, write, edit, and list local files.
Web & API Tools: Web scraping, HTTP requests, JSON parsing.
System Tools: Shell command execution, system info retrieval, timestamp ops.
Text Tools: Search, replace, format, and analyze text.
Extensive Suite: 16+ built-in utilities ready for production use.
LLM Integration
Works with OpenAI, any OpenAI-compatible API, or local models via llama.cpp.
Offers HuggingFace integration for local inference.
Includes a built-in HTTP server that exposes OpenAI-compatible endpoints.
Full control over parameters like temperature, max_tokens, and stop.
Technical Highlights
Built on Tokio for async performance.
Type-safe, extensible, and minimal overhead.
Functions as both CLI and library crate.
Clean and quiet offline output—no debug clutter.
Crate: https://crates.io/crates/helios-engine
Docs: docs.rs/helios-engine
Top comments (0)