Anthropic opens Claude Managed Agents dynamic workflows to public beta, with up to 1,000 sub-agents
Anthropic's new dynamic workflow feature lets a main Claude agent write a program that splits work across as many as 1,000 sub-agents. In an Anthropic test, it found 66 of 70 planted bugs in a 116,000-line codebase.
Photo: TechCrunch / Wikimedia Commons, CC BY 2.0. Anthropic logo via Wikimedia Commons (public domain).
Anthropic launched a public beta of dynamic workflows for Claude Managed Agents on October 10, AI Understanding reported. The feature lets a main agent generate a workflow program that defines how sub-agents divide labor, work in parallel and hand off results.
According to the report, a workflow supports up to 1,000 agents, with a current limit of 64 running at the same time. Each agent keeps its own conversation history but shares the session's files and sandbox.
In a test Anthropic reported on a 116,000-line codebase with 70 planted bugs, a single agent found 14, 15 and 27 bugs across three runs, while the dynamic workflow found 66 in each run. AI Understanding noted that Anthropic did not disclose the division of labor used, time taken, token usage or false-positive rates.
Anthropic also expanded the public beta of Claude Code Projects, which adds persistent, project-based task management, to all Pro and Max users, the report said.