Ruflo - A Multi-Agent AI Orchestration Platform for Claude Code

Claude Flow has been rebranded as Ruflo and reorganized into a multi-agent platform that orchestrates over 100 specialized agents on top of Claude Code.

• A single line, npx ruvflo init, organizes a swarm, attaching task-specific learning, cross-session memory, and secure communication with agents on other machines.
• Installation splits into a lightweight Claude Code Plugin approach that leaves no files like .claude/ in the workspace, and a full CLI installation. The full installation includes 98 agents, 60+ commands, 30 skills, an MCP server, hooks, and a daemon.
• 32 native plugins cover core orchestration, memory/RAG, code quality, security, architecture, DevOps, and IoT/trading domains.
• Memory is built on HNSW vector search and SONA self-learning patterns, and routes across Claude, GPT, Gemini, Cohere, and Ollama together.
• The Web UI beta offers multi-model chat and parallel execution of about 210 MCP tools, and users can also add custom MCP servers.
• The Goal Planner UI turns natural-language goals into A*-based GOAP plans, and a live agent dashboard monitors execution status.
• Agent Federation has built-in zero trust communication, mTLS + ed25519 authentication, 14 types of PII detection, behavior-based trust scoring, and HIPAA/SOC2/GDPR modes.
• The underlying core uses the Cognitum.One agentic architecture and a Rust-based engine. The project is MIT licensed.