Claude-code-flow hive-mind

install

source · Clone the upstream repo

git clone https://github.com/ruvnet/ruflo

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/ruvnet/ruflo "$T" && mkdir -p ~/.claude/skills && cp -r "$T/.agents/skills/hive-mind" ~/.claude/skills/ruvnet-claude-code-flow-hive-mind && rm -rf "$T"

manifest: .agents/skills/hive-mind/SKILL.md

Hive-Mind Skill

Purpose

Byzantine fault-tolerant consensus and distributed swarm coordination.

When to Trigger

Multi-agent distributed tasks
Fault-tolerant operations needed
Collective decision making
Complex coordination patterns

Topologies

Topology	Description	Use Case
`hierarchical`	Queen controls workers	Default, anti-drift
`mesh`	Fully connected peers	Research, exploration
`hierarchical-mesh`	Hybrid	Recommended for complex
`adaptive`	Dynamic based on load	Auto-scaling

Consensus Strategies

Strategy	Tolerance	Use Case
`byzantine`	f < n/3 faulty	Untrusted environment
`raft`	f < n/2 faulty	Leader-based, consistent
`gossip`	Eventual	Large scale, availability
`crdt`	Conflict-free	Concurrent updates
`quorum`	Configurable	Tunable consistency

Commands

Initialize Hive-Mind

npx claude-flow hive-mind init --topology hierarchical-mesh --consensus raft

Spawn Queen

npx claude-flow hive-mind spawn --role queen --name coordinator

Check Consensus Status

npx claude-flow hive-mind consensus --status

View Sessions

npx claude-flow hive-mind sessions --active

Best Practices

Use hierarchical for coding tasks (anti-drift)
Use raft consensus for consistency
Keep agent count under 8 for coordination
Run frequent checkpoints