Anthropic Adds Dynamic Workflows to Claude Managed Agents for Mass Parallel Orchestration

Anthropic announced an update to Claude Managed Agents introducing dynamic workflows capable of orchestrating up to 1,000 AI agents in parallel for a single execution. Under this framework, a primary master agent decomposes complex tasks, dispatches them to sub-agents, and aggregates the results upon completion. Scaling agent orchestration to 1,000 parallel threads marks a major advancement in multi-agent architecture and automated codebase analysis. It enables enterprises to solve complex, large-scale problems—such as comprehensive software security audits—far more effectively than a single agent could. In a benchmark test involving a 116,000-line codebase containing 70 seeded bugs, dynamic workflows consistently detected 66 vulnerabilities, compared to only 14 to 27 detected by a single agent. The workflow runtime executes dynamic orchestration scripts in the background while keeping user sessions responsive.