swarm

Introduces the /swarm multi-agent parallel fan-out mode in the Claude Code ecosystem, which was recently disabled by default with hard limits added, due to runaway recursion and cost explosion.

This paper introduces Collective-State JEPA (CS-JEPA), a recurrent joint-embedding predictive architecture that lets every robot in a swarm predict the same future collective state from local observations and limited messages. The method shows label-efficient improvements in prediction error and inter-robot agreement, plus planning-relevant value estimation.

A developer built a 24/7 multi-agent swarm for marketing a SaaS product, where each agent is a full Hermes agent rather than a simple LLM with tools.

A critical commentary comparing AI agent orchestrators to middle managers, reviewer agents to flunkies, and swarms to factories producing bullshit jobs, highlighting perceived inefficiencies in multi-agent systems.

SwarmResearch introduces an orchestrator-subagent harness where a Shepherd Agent steers a population of Search Agents to explore diverse solutions for open-ended optimization problems, achieving better or comparable results to state-of-the-art methods on 13/15 tasks.

Introduces codemap-based assistance features for coding agents, including Swarm and Pavel optimization techniques.

A detailed guide on using Claude Code's Dynamic Workflows pattern to orchestrate multiple parallel subagents from a single lead agent, with 9 steps covering task decomposition, isolation, and review.

Introduces an open-source course called 'AI Engineering from Scratch', containing 503 lessons, covering from linear algebra to autonomous swarms, with special emphasis on Agent Engineering and Multi-Agent phases.

A tool called 'orchestrate' enables users to run a swarm of AI agents with multiple planners, verifiers, and workers, accessible via plugin commands.