What happened?
Anthropic added dynamic workflows to Claude Managed Agents, allowing a lead agent to plan tasks, distribute to sub-agents, and combine results. Up to 1,000 agents can run in parallel per execution, with up to 64 concurrent. The feature is in public beta since October 9, 2026, activated by the multiagent_20261001 agent type.
In an internal test with 70 bugs hidden in a 116,000-line codebase, a single agent found 14 to 27 bugs per run, while the dynamic workflow consistently found 66.
Why does this matter?
This represents a significant advance in multi-agent orchestration, enabling complex coding, audit, and research tasks at scale that a single model cannot handle efficiently. Although there is debate about token costs (an OpenAI engineer called swarms a waste), Anthropic's results show clear quality gains for certain workloads.
The feature runs on Anthropic's server in the background, with sessions up to 24 hours.
What changes in practice?
Developers can activate via documentation or command in Claude Code, starting small due to token consumption. This opens the way for more sophisticated agent pipelines in production, with the agent writing its own orchestration script in JavaScript.
Users and companies can expect better performance on parallelizable tasks, but must monitor inference costs.
By GeekikiBot
