ruvnet/ruflo

🌊 The leading agent meta-harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

What it solves

RuFlo is an "agent meta-harness" designed to transform LLMs (specifically Claude Code and Codex) from simple text generators into functional, autonomous agents. It provides the necessary execution layer—including tools, memory, loops, and security guardrails—that allows agents to collaborate in swarms, learn from past tasks, and operate across different machines securely.

How it works

RuFlo acts as a coordination layer between the user and the LLM. It uses a router to dispatch tasks to specialized agent swarms, which are supported by a self-learning memory system (AgentDB) and a plugin architecture. It integrates with the Model Context Protocol (MCP) to provide agents with a vast array of tools. The system can be deployed as a CLI tool, a set of Claude Code plugins, or a self-hosted Web UI.

Who it’s for

It is built for developers and enterprises who want to move beyond basic chat interfaces to production-grade agentic workflows, specifically those using Claude Code or Codex, and those needing multi-agent coordination and persistent, self-optimizing memory.

Highlights

  • 100+ Specialized Agents: Pre-built agents for coding, security auditing, documentation, and architecture.
  • Swarm Coordination: Supports hierarchical and mesh topologies for agents to work as a team with consensus.
  • Self-Learning Memory: Uses HNSW-indexed AgentDB for fast vector search and trajectory learning to improve over time.
  • Zero-Trust Federation: Allows agents on different machines or organizations to collaborate securely without leaking data.
  • GOAP Planning: Includes a Goal-Oriented Action Planning (GOAP) system that decomposes plain-English goals into executable A* paths.
  • MetaHarness: A tool to audit agent setups, grade readiness, and scan for security risks before deployment.