Home

Ruflo

Multi-agent AI orchestration for Claude Code — Orchestrate 100+ specialized AI agents across machines, teams, and trust boundaries.

Ruflo adds coordinated swarms, self-learning memory, federated comms, and enterprise security to Claude Code so agents don't just run, they collaborate.

Get Started in 3 Steps

1. Install

npx ruflo@latest init wizard

Or add to Claude Code as an MCP server:

claude mcp add ruflo -- npx ruflo@latest mcp start

2. Create Your First Swarm

npx ruflo@latest swarm init --topology hierarchical --max-agents 8

3. Start Using Ruflo

After init, use Claude Code normally. Hooks automatically route tasks, spawn agents, and learn from successes.

---

What You Get

• 314 MCP tools — Memory, swarm coordination, neural routing, embeddings, security, and more
• 60+ agent types — Specialized roles that self-organize into teams
• 26 CLI commands — Control swarms, manage memory, run benchmarks, security audits
• Self-learning — Agents remember successful patterns and get smarter on every task
• Federation — Agents on different machines securely collaborate
• Enterprise security — AIDefence (PII detection, prompt injection blocking), witness verification (ADR-103)

---

Key Features

Swarm Coordination

Initialize multi-agent teams with automatic anti-drift guardrails. Choose your topology: hierarchical (tight control), mesh (peer collaboration), or adaptive (dynamic).

Memory & Graph Intelligence

Fast vector database (HNSW 150x-12,500x faster) + knowledge graph reasoning (ADR-130). Agents remember across sessions and find relevant patterns in milliseconds.

Neural Learning

SONA self-optimization (<0.05ms adaptation), Mixture of Experts routing, trajectory tracking. Agents learn from every success and failure.

Witness Verification

Signed manifest of all artifacts. ruflo verify confirms the installed version matches the audited baseline. ADR-103.

---

Documentation

• Installation — Setup for macOS, Linux, Windows
• Quick Start — Your first swarm and agent
• CLI Reference — All 26 commands, 140+ subcommands
• MCP Tools — 314 tools for memory, coordination, security
• Agents — 60+ agent types and their roles
• Memory & Graph — How agents remember (HNSW, graph intelligence, ADR-130)
• Intelligence Pipeline — Self-learning (SONA, DISTILL, CONSOLIDATE)
• Hooks & Workers — 17 hooks + 12 background workers
• Hive-Mind — Queen-led consensus and topologies
• Plugins — 21+ native plugins, how to write one
• ADR Index — Architecture decisions (130 ADRs)
• Benchmarks — SOTA comparison vs. LangGraph, AutoGen, CrewAI
• Witness & Verification — ADR-103, how to verify
• Roadmap — What's next
• Troubleshooting — Common issues and fixes
• Contributing — How to submit issues and PRs

---

Live Resources

• npm: ruflo@latest
• GitHub: ruvnet/ruflo
• Benchmarks: SOTA comparison gist
• Summit: RuFlo Summit 2026 (Budapest, June 2–3)

---

Quick Links by Use Case

I want to...	Start here
Try Ruflo without committing	Quick Start
Build a multi-agent team	Swarm Coordination in CLI-Reference
Store and retrieve agent knowledge	Memory & Graph
Understand routing and adaptation	Intelligence Pipeline
Extend with custom behavior	Plugins
Compare vs. competitors	Benchmarks

---

Ruflo v3.10.1 · npm · GitHub · Benchmarks