Raven

Overview

Raven is a self-evolving multi-agent orchestration ecosystem and a host agent that brings built-in and third-party agents together to carry out complex tasks. It generates DAGs and orchestrates specialized agents, supports harness self-evolution, and is powered by EverOS for memory and context across sessions.

Key Features

• Generates DAGs and orchestrates multiple specialized agents for complex tasks.
• Includes four built-in agents: Raven-Research, Raven-Code, Raven-Design, and Raven-Oncall.
• Connects third-party agents via ACP, CLI, or OpenAI-compatible APIs, with presets for 13 agents.
• Uses EverOS Memory to preserve user context, agent experience, and world knowledge across sessions.
• Uses SkillForge to retrieve relevant skills from local libraries, EverOS memory, and SkillHub's catalog.
• Combines event monitoring and scheduled execution for proactive reminders and follow-up work.

Use Cases

• Deep research for complex questions, literature reviews, and technical analysis, with structured reports and traceable sources.
• Agentic software development for feature implementation, debugging, refactoring, data processing, and data analysis.
• Visual design for PowerPoint slide decks, brand assets, charts, diagrams, and web interfaces.
• Unattended workflow automation for experimentation, optimization, and continuous monitoring.

Getting Started

• On Linux, macOS, or WSL2, run: curl -fsSL https://raven\.evermind\.ai/install\.sh | bash
• On native Windows PowerShell, run: irm https://raven\.evermind\.ai/install\.ps1 | iex
• For a container deployment, clone the repository, enter the Raven directory, and run: docker compose -f docker/docker-compose.yml up
• After container deployment, open http://localhost:18793 in your browser.
• Run raven web to launch the WebUI, or raven web --stop to stop the background service.

Deployment & Requirements

• The shell installer supports Linux, macOS, and WSL2.
• Native Windows PowerShell is supported.
• Container deployment requires Git and Docker with Docker Compose.

Before You Adopt

• License: Apache-2.0. Review its terms before using, modifying, or distributing the project.
• The Curator component is explicitly experimental and is distributed with the source repository rather than the installed package.
• Third-party agent orchestration depends on ACP, CLI, or OpenAI-compatible APIs.