Claude, the essentials — edition of November 7, 2026
Anthropic tightens access controls and showcases Claude in scientific research
A beta IP allowlist, a beta Admin API sign-out and a revised Claude Code auto mode all concern who and what is trusted to act. Alongside them, Anthropic published a run of pieces on Claude in scientific work, agent memory and inference.
- Claude Console now has a beta IP allowlist. Requests from addresses not on the list are refused, which Anthropic says limits where a leaked API key can be used.
- A beta Admin API lets Claude Enterprise administrators sign a member out of Claude everywhere. The member stays in the organization and can sign in again.
- Claude Code 2.1.319 changes auto mode when telemetry or feature flags are off, including behind LLM gateways. Calls the server cannot check are now asked about or denied, with no fallback to the built-in classifier.
- Claude Code now runs MCP server tool calls that execute as tasks in the background. It reports the result to Claude when they finish.
- Anthropic published several pieces on Claude in science, covering scientific model-building, vault particles and how researchers use Claude. It also published pieces on agent memory and provable inference.
Narrowing where credentials and sessions are valid
Anthropic's API release notes, listed under 5 November, describe two administrative controls, both in beta. In the Claude Console, an organization can now set an IP allowlist. Requests to covered services from addresses not on the list are refused. Anthropic frames the purpose as limiting where a leaked API key can be used. Organizations can add IPv4 and IPv6 addresses or CIDR ranges, with a ceiling of 1,000 per organization.
The second control concerns people rather than keys. With the Admin API, an administrator of a Claude Enterprise (claude.ai) organization can sign a member out of Claude everywhere. Anthropic specifies that the member stays in the organization and can sign in again, so this revokes sessions without removing anyone. A related change appears in Claude Code 2.1.318: the Claude apps gateway gains assume_role on mantle upstreams. The release note says this serves a role in another AWS account and per-developer AWS cost attribution.
Sources: Claude Console IP allowlist (beta) and Admin API sign-out (beta), API release notes — Anthropic (release notes) · Claude Code 2.1.318 release notes — Claude Code
Claude Code: auto mode gets stricter when the server cannot check
Version 2.1.319 changes auto mode in setups where telemetry or feature flags are off, including those behind LLM gateways. The release note says calls the server cannot check are now asked about or denied, and that there is no fallback to the built-in classifier. Anyone who wants to keep using the classifier can set CLAUDE_CODE_AUTO_MODE_SERVER=0. The note indicates that Bedrock, Vertex and Foundry are unchanged.
A second note covers LLM gateways. Setting CLAUDE_CODE_AUTO_MODE_SERVER=1 turns on the server-side check even when CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS is set. If managed settings set the latter, they must set the former to 1 as well. The same release adds terminal support for MCP tasks: when an MCP server runs a tool call as a task, Claude Code moves it to the background immediately and tells Claude the result when it finishes. Version 2.1.318 adds smaller items for the mods Claude Code can write in a session. It records the user's answer when Claude Code asks to turn them on. It also adds an effort option to a mod's $.model.fork, so a quick side job can run at low effort whatever effort the session uses, and the result reports the effort sent.
Sources: Claude Code 2.1.319 release notes — Claude Code · Claude Code 2.1.318 release notes — Claude Code
Claude in the laboratory: a cluster of science pieces
Anthropic published several items on scientific work. They carry the titles "AI for scientific discovery: How researchers use Claude", "Teaching AI to build scientific models" and "Investigating vault particles with Claude". The feed gives only headlines for these pieces, so it supports no claim about their findings or methods. What it shows is Anthropic presenting scientific research as a use case worth documenting, both in general and in one case study on vault particles.
The same day's feed also lists a 2026 paper under Anthropic's name, "Prokaryotic homologs of the major vault protein form engineerable vault nanocapsules", by Athukoralage et al. It sits next to the vault-particle piece, but the feed does not state how the two are connected.
Sources: AI for scientific discovery: How researchers use Claude — Anthropic · Teaching AI to build scientific models — Anthropic · Investigating vault particles with Claude — Anthropic · Athukoralage et al. 2026, Prokaryotic homologs of the major vault protein form engineerable vault nanocapsules — Anthropic
Agents, inference and people: other Anthropic publications
Anthropic also published "The components of agent memory", "Project EVE: Claude plays MMORPGs" and "Progress on provable inference for large language models". Again the feed carries only titles, so their content cannot be summarized here. Taken together they show Anthropic writing publicly about how agents retain information, how Claude behaves in long-running game worlds, and how inference might be made verifiable.
On the hiring side, the Anthropic Alignment Science blog announced that applications are open for the Anthropic Residency, aimed at established technical leaders. The item gives no further detail on the programme's format, duration or terms.
Sources: The components of agent memory — Anthropic · Project EVE: Claude plays MMORPGs — Anthropic · Progress on provable inference for large language models — Anthropic · Anthropic Residency: applications open for established technical leaders — Anthropic Alignment Science blog
This edition is an original synthesis written by Claude from aggregated news — Anthropic's own sources first (release notes, status, newsroom, research, engineering), then the press, Hacker News, Reddit and GitHub, under the editorial supervision of Héra SASU. Every fact links to its article, publisher named. See the live feed →
Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.