What Makes Ruflo Different

Most agent frameworks automate execution. Ruflo automates the cycle of proposing ideas, coordinating multiple agents, and learning from every interaction — so your AI systems get smarter over time.

Swarm Intelligence

Agents self-organize into hierarchies, elect leaders via Raft consensus, and coordinate across Byzantine-fault-tolerant networks. Not one agent — a swarm.

Self-Learning Memory

HNSW vector search with sub-millisecond retrieval. SONA pattern matching learns from successful trajectories. Your agents remember and improve across sessions.

Enterprise Security

CVE-hardened with AIDefence. mTLS + ed25519 challenge-response for federation. PII detection with 14-type scanning pipeline before any data leaves your node.

Core Capabilities

Orchestration

Multi-Agent Swarms

Deploy 100+ specialized agents — coders, testers, reviewers, architects, security analysts — that coordinate through a queen-led hierarchy with automatic task routing at 89% accuracy.

Memory

Persistent AgentDB

AgentDB with HNSW vector search stores trajectories, outcomes, and embeddings. Memory persists across sessions. Ask "what did we decide about auth last month?" and get the exact context.

Federation

Cross-Boundary Collaboration

Agents on different machines, organizations, or cloud regions discover each other, prove identity via mTLS, and collaborate on tasks — with automatic PII stripping and behavioral trust scoring.

Interface

Web UI & Goal Planner

Self-hostable web interface with multi-model chat (Claude, GPT, Gemini, Qwen), ~210 MCP tools, and a GOAP A* goal planner that turns plain-English goals into executable agent plans.

How It Works

One command gives Claude Code a nervous system. The hooks system automatically routes tasks, learns from successful patterns, and coordinates agents in the background.

1

Initialize

Run npx ruflo init in your project. This installs the MCP server, 27 hooks, and registers the orchestration layer with Claude Code.

2

Define Agents

Create specialized agents with roles, memory namespaces, and token budgets. Or use 32 pre-built plugins covering core orchestration, swarm management, and autopilot.

3

Collaborate

Use Claude Code normally. Ruflo intercepts tasks, routes to the right agent swarm, coordinates parallel execution, and stores learnings for future runs.

Architecture Overview

Quick Start

Two install paths. Pick based on what you need.

Path A — Plugins Only

Slash commands and agent definitions. Best for trying a single plugin without full install.

# Add the marketplace
/plugin marketplace add ruvnet/ruflo

# Install core + plugins
/plugin install ruflo-core@ruflo
/plugin install ruflo-swarm@ruflo
/plugin install ruflo-federation@ruflo

Note: MCP server is NOT registered. Memory and swarm tools won't be callable.

Path B — Full Install

Everything works as documented. Production use with full memory, swarm, and federation.

# One command installs everything
npx ruflo init

# Your agents are now orchestrated
# Use Claude Code normally — Ruflo handles the rest

Includes hooks, MCP server, memory store, and swarm initialization.

FAQ

Do I need to learn new commands?
No. After init, use Claude Code normally. The hooks system automatically routes tasks, learns from successful patterns, and coordinates agents in the background.
What LLM providers are supported?
Claude (Anthropic), GPT (OpenAI), Gemini (Google), Cohere, and Ollama for local models. Five providers with automatic failover.
Is my data safe?
Yes. Ruflo is CVE-hardened with AIDefence. Federation uses mTLS + ed25519 identity verification. A 14-type PII detection pipeline scans every outbound message before it leaves your node.
Can I self-host the Web UI?
Yes. The Web UI ships as a Docker container with embedded MongoDB. Deploy to Cloud Run, Fly, Kubernetes, or docker-compose. Fully self-hostable.
How many agents can run simultaneously?
Ruflo supports 100+ specialized agents in a single swarm, with 12 auto-triggered background workers and intelligent task routing at 89% accuracy.

Ready to Orchestrate Your Agents?

Join the growing community of developers building the future of autonomous AI collaboration.