# You can now let a Claude Managed Agents agent use dynamic workflows

By Steven Van · 2026-10-09

The beta lets agents run multi-agent programs in the background, with progress events and session budgets to track and cap the work.

Large tasks such as reviewing hundreds of documents or cross-checking sources can now be split across agents that work in parallel. [Anthropic](<https://creatorstoolbox.com/tools/anthropic>) has added dynamic workflows to Claude Managed Agents in beta, using the managed-agents-2026-04-01 beta header.

## How a workflow runs

The agent writes a program that runs multiple agents in phases and combines their results. The server executes it in the background while the original agent keeps working or ends its turn.

Describe the task in a user.message, and the agent decides whether and when to start a run. No additional API call is needed. Instructions in that message or the agent's system prompt can guide when to use workflows, how to split the work, how to handle failures and what time limit to apply.

## Enable and configure agents

Set multiagent.type to "multiagent\_20261001" and workflows to {"type": "enabled"}. This type enables both workflows and subagent delegation by default. For workflows alone, also set subagents to {"type": "disabled"}.

A workflow can define its own agents, use existing agents listed in workflows.predefined\_agents, or combine both. Agents defined within the workflow use the session agent's model. To use another model for some of the work, list predefined agents configured with that model.

## Track runs and spending

Follow runs through workflow\_run.\* events on the session's event stream. Each agent in a run also works in a session thread that can be listed, read and streamed.

Every agent in a run uses tokens. A budget set when creating a session caps spending across the session, including workflow runs, and pauses runs when the budget is reached. Configuration changes apply to new sessions, so updating an agent does not change sessions already underway.

The [dynamic workflows documentation](<https://platform.claude.com/docs/en/managed-agents/multiagent-orchestration#dynamic-workflows>) covers setup and instructions for guiding runs.

![Anthropic](<https://www.google.com/s2/favicons?domain=anthropic.com&sz=128>)

Anthropic

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

[View Anthropic →](<https://creatorstoolbox.com/tools/anthropic>)

[Original source](<https://platform.claude.com/docs/en/release-notes/overview#october-9-2026>)

[Read You can now let a Claude Managed Agents agent use dynamic workflows on Creators Toolbox](<https://creatorstoolbox.com/blog/anthropic-you-can-now-let-a-claude-managed-agents-agent-use-dynamic-workflows>)

---
Canonical source: https://creatorstoolbox.com/blog/anthropic-you-can-now-let-a-claude-managed-agents-agent-use-dynamic-workflows