Initializes a multi-agent swarm with configurable topology for coordinated parallel work.
npx skills add https://github.com/ruvnet/claude-flow --skill swarm-init
Please help me install this Agent Skill. Skill: swarm-init Source: https://github.com/ruvnet/claude-flow/tree/main/plugins/ruflo-swarm/skills/swarm-init Command: npx skills add https://github.com/ruvnet/claude-flow --skill swarm-init
Coordinating multiple AI agents on complex multi-file tasks often leads to drift, conflicting edits, and lost context. This Skill sets up a structured swarm with a defined topology so agents can work in parallel with clear coordination.
Ask the AI to initialize a hierarchical swarm with 8 agents using the swarm-init skill before starting your multi-file refactoring task.
High-intent search queries and answers about installing and using this skill.
Call the MCP tool mcp__plugin_ruflo-core_ruflo__swarm_init with a topology, maxAgents, and strategy, or run npx @claude-flow/cli@latest swarm init with the same flags. Then spawn named agents with the Task tool using run_in_background for parallel execution.
For teams of 10 or more agents, use the hierarchical-mesh topology, which combines layered coordination with peer-to-peer messaging. Smaller teams of around 8 agents typically work well with the standard hierarchical topology.
Skip swarm setup for single-file edits or quick questions, since coordination overhead outweighs the benefit. The skill is intended for complex multi-file tasks needing three or more coordinated agents, such as feature implementation or security audits.
Each agent uses EnterWorktree to work in an isolated git worktree, preventing conflicting edits to the same working tree. Agents coordinate through SendMessage, which requires spawning them with a name parameter for addressability.
Yes, spawn all agents in one message using the Task tool with run_in_background set to true. This launches them concurrently, and SendMessage handles inter-agent coordination while they execute.
agentmattersDecompose large features into verifiable work units for multi-agent execution.
mariasebarespersonaMerge parallel worktree branches into main with conflict detection and verification.
AcosmiCoordinate multiple autonomous agents through structured inter-agent collaboration and orchestration modes.
Dmccarty30Coordinate distributed swarm workloads across multiple agents with MCP tools.
