// AI Platform Drift
Claude Managed Agents web tools now enforce allowed_hosts in limited networking
Claude Platform release notes · published 2026-10-07 23:07 UTC
WHAT CHANGEDEffective October 7, 2026, Claude Managed Agents with limited networking now apply allowed_hosts restrictions to web_search and web_fetch tools. A web_fetch call to a disallowed host returns url_not_allowed to the agent; web_search omits results from disallowed hosts. Session creation or update now fails with a 400 error when web tools have allowed_domains entries not covered by allowed_hosts. Host matching is exact unless the entry starts with *., so docs.example.com does not match example.com.
WHY IT MATTERSExisting agents with web tools enabled will fail to create or update sessions if their allowed_domains contain hosts not in allowed_hosts. Running agents will receive url_not_allowed errors or empty search results instead of fetching pages they could reach before. This breaks any agent workflow that depends on web_search or web_fetch reaching hosts not explicitly listed in allowed_hosts.
WHAT TO DOAudit every agent session configuration with limited networking and web tools enabled. For each allowed_domains entry, verify a matching allowed_hosts entry exists—add exact hosts or wildcard prefixes like *.example.com to allowed_hosts, or remove the entry from allowed_domains. Test session creation and agent web tool calls against your production host lists before October 7.
Model-reported confidence 95% (not measured accuracy)
Primary source ↗
written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+## Claude Platform release notes — October 7, 2026 +https://platform.claude.com/docs/en/release-notes/overview#october-7-2026-5 +Wed, 07 Oct 2026 00:00:00 GMT + +In Claude Managed Agents, a cloud environment with limited networking now also applies its allowed_hosts to the web_search and web_fetch tools. A web_fetch call for a URL on a host that allowed_hosts does not match returns a url_not_allowed error result to the agent. web_search omits results from such hosts. When allowed_hosts lists no hosts, neither tool returns a page or a search result. allow_package_managers and allow_mcp_servers add no hosts for these tools. To let the tools reach a host, add it to allowed_hosts, which also opens it to the sandbox. unrestricted networking and self-hosted environments do not limit these tools. See Environment networking. +With limited networking, creating a session fails with a 400 error when an enabled web tool's allowed_domains has an entry not within allowed_hosts. So does a session update that adds such an entry. An allowed_hosts entry matches one exact host unless it starts with *., so docs.example.com is not within ["example.com"]. To fix the error, add the host to allowed_hosts or remove the entry from allowed_domains. See Restrict web search and web fetch domains.
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
| triage verdict | MATERIAL at 0.92, needed 0.45 |
|---|---|
| triage reasoning | Breaking changes to Managed Agents web tool behavior and session validation rules affect production deployments. |
| signals | breaking API change: web_search and web_fetch tools now respect allowed_hosts in limited networking · new error behavior: 400 error on session creation/update when allowed_domains entry not in allowed_hosts · new error result: url_not_allowed returned to agent for web_fetch calls outside allowed_hosts · behavior change: web_search omits results from disallowed hosts · new constraint: allowed_hosts matching rules (exact host vs wildcard prefix) |
| suppression | — |
| triage model | anthropic/claude-haiku-4.5 ($0.00000) |
| writer model | anthropic/claude-sonnet-4.5 ($0.00000) |