Summary
On August 28, 2026, Anthropic's Claude Developer Platform added the ability to restrict which sites a Managed Agents agent's web_search and web_fetch tools can reach via allowed_domains or blocked_domains, and introduced an anthropic-workspace-id response header that returns the workspace an API key or access token resolved to.
What changed
Managed Agents can now constrain their built-in web_search and web_fetch tools to an allow-list or block-list of domains, and every Claude API response now carries an anthropic-workspace-id header identifying the resolving workspace.
Why it matters
Domain scoping closes a real exfiltration and prompt-injection surface for autonomous agents that browse the web, and the workspace-ID header improves multi-workspace attribution, auditing, and billing hygiene — both aimed squarely at enterprise governance of hosted agents.
Evidence excerpt
You can now restrict which sites a Claude Managed Agents agent's web_search and web_fetch tools can reach. Set allowed_domains or blocked_domains on the tool's entry in the agent_toolset config; the anthropic-workspace-id response header returns the workspace an API key or access token resolved to.