Anthropic: Managed Agents gain multiagent sessions, Outcomes, webhooks and vault refresh in public beta
Claude Managed Agents is Anthropic's managed platform for autonomous agents, which on May 6, 2026 received four new features in public beta: multiagent sessions, the Outcomes mechanism for defining goals, webhooks for session and vault lifecycle events, and background refresh for mcp_oauth credentials. New filters for sessions by status and events by type and creation time were also added.
This article was generated using artificial intelligence from primary sources.
On May 6, 2026, Anthropic published four expansions to the Claude Managed Agents platform in the Claude Platform release notes. All new functionality is available in public beta under the standard managed-agents-2026-04-01 beta header.
Claude Managed Agents is Anthropic’s managed platform for autonomous agents — a server-side harness that runs Claude as an agent in a secure sandbox, with built-in tools and SSE streaming. It launched on April 8, 2026.
What are the four new capabilities?
The release notes list four specific items for May 6:
Multiagent sessions and Outcomes — Multiagent sessions allow multiple agents to work within the same session, while Outcomes is a mechanism for defining goals the agent must achieve. Both features are now in public beta.
Webhooks for Managed Agents — External systems can subscribe to session and vault lifecycle events via webhook endpoints, eliminating the need to poll agent state.
Background refresh for vault credentials — mcp_oauth credentials in the vault are now refreshed in the background, preventing session interruptions due to expired OAuth tokens toward MCP servers.
New filters for sessions and events — Sessions can be filtered by status, events by type, and events now support filtering by creation time.
Why is multiagent sessions a big deal?
Until this update, a Claude Managed Agent operated as a single agent per session. Multiagent sessions open the door to orchestrating multiple agents through the same contextual unit, which is a prerequisite for workflows where one agent coordinates specialized sub-agents (e.g., researcher + coder + reviewer pattern).
The Outcomes mechanism arrives as a natural complement: if a session has multiple agents, the session goal must be explicit and measurable rather than implicit in the prompt. Outcomes provides that layer.
What do webhooks and filters change in practice?
Webhooks solve an operational pain point — previously, organizations had to poll GET requests for session status or vault access. Event types cover session and vault lifecycle.
The new filters (status for sessions, type and creation time for events) address the practical needs of teams managing dozens of agents — without filters, listing endpoints were drowning useful data in noise.
All four items remain behind the beta header, meaning Anthropic retains room for breaking changes before the GA phase.
Frequently Asked Questions
- What are multiagent sessions?
- Multiagent sessions are a feature within the Claude Managed Agents platform that allows multiple agents to work within the same session. They are available in public beta under the standard 'managed-agents-2026-04-01' beta header.
- What is the Outcomes feature?
- Outcomes is a mechanism for explicitly defining goals that a Managed Agent should achieve within a session. It is also in public beta under the same beta header as multiagent sessions.
- What events do the new webhooks cover?
- Webhooks for Claude Managed Agents cover session and vault lifecycle events, enabling external systems to react to agent state changes without polling.
Related news
GitHub: validation of agentic behavior via dominator analysis from compiler theory achieves 100% accuracy vs 82% agent self-assessment
vLLM: Mooncake distributed KV cache store integration delivers 3.8× higher throughput and 46× lower P50 TTFT for multi-turn agentic workloads
Anthropic: 10 ready-made financial-services agent templates + Claude Opus 4.7 at 64.37% on Vals AI Finance benchmark