FR
Live

Claude, the essentials — edition of November 17, 2026

Claude Code's Code Mode matures as Claude Meetings strains under demand

By ·

Three Claude Code releases in a few days built out the experimental Code Mode and added administrative guardrails. Anthropic's status page also recorded an overnight model outage and a Claude Meetings beta that couldn't keep up with demand.

Key points

Claude Code builds out Code Mode and tightens its guardrails

The main product change this week is Code Mode, introduced in Claude Code 2.1.323. Anthropic describes it as an experimental opt-in in which Claude writes JavaScript to call its tools. The release points to documentation for turning it on, and for turning it off across an organization. Version 2.1.326 then removed one friction point. Code Mode can now be enabled while agent teams are on. For users who had already enabled it, sessions with teams switched on now start in Code Mode.

Around that feature, the three releases add several checks for people who configure Claude Code. Version 2.1.326 shows a startup warning when an --allowedTools, --disallowedTools or --tools value contains an entry that doesn't look like a complete rule. It also explains why MCP servers stop loading when an administrator adds managed-mcp.json during a session. Version 2.1.325 adds a line under the error shown when an organization's policy blocks a request. The line tells the user to run /rewind and change the message, or run /clear. For plugin authors, 2.1.323 adds an engines field for claude-code in plugin.json. Plugins outside their stated Claude Code version range are not installed, updated or loaded, and install and update pick the newest version that fits. It also adds type checking of a plugin's TypeScript hooks module to claude plugin validate. Each finding is a warning, and --strict turns it into a failure.

Sources: Claude Code 2.1.323 release notes — Claude Code · Claude Code 2.1.325 release notes — Claude Code · Claude Code 2.1.326 release notes — Claude Code

Watermarking and self-hosted agents on the API side

The API release notes dated 16 November say that text generated by Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.5 and Claude Haiku 4.5 carries Anthropic's text watermark on supported platforms. A page titled 'How Claude marks AI-generated content' lists those platforms. The visible excerpt also mentions requests to these models that set sampling parameters such as temperature and top_p. The rest of that passage is cut off, so its implications can't be stated here.

The notes dated 12 November describe self-hosted Claude Managed Agents. Customers run the whole Managed Agents service on their own infrastructure, not only self-hosted sandboxes. They also store agents, sessions and session history there. According to the notes, the server sends model requests to a provider the customer chooses.

Sources: Anthropic API release notes, November 16, 2026 — Anthropic (release notes) · Anthropic API release notes, November 12, 2026 — Anthropic (release notes)

Reliability: a model outage, a strained beta and a sign-in problem

Claude Meetings (beta) is the incident still open. Anthropic's status page says the service saw higher-than-expected demand today and that Claude was unable to join many calls. Mitigations are in place and the company is monitoring join reliability. Claude Meetings is re-enabled for all Enterprise customers, though Claude may take longer than usual to join. Adding Claude to new calls and Auto-join remain paused for Team customers.

Earlier, the status page recorded elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2. The incident affected claude.ai, the Claude API, Claude Code and Claude Cowork. It ran from 23:32 PT on 15 November (07:32 UTC on 16 November) to 00:45 PT (08:45 UTC) on 16 November, and is marked resolved.

A third notice concerns Sign in with Apple on the Claude iOS app, which mainly affects new accounts. The entry is listed as resolved, but its text says investigation was continuing. New users can sign up with Google or email in the meantime.

Sources: Claude unable to join meetings — status.claude.com · Elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2 — status.claude.com · Some users unable to sign up with Apple on the Claude iOS app — status.claude.com

Security and safety: a patched decoder and two Anthropic publications

On the software side, Claude Code 2.1.325 fixed a security issue in how zstd-compressed responses are decoded. The release notes give no further detail on the nature or severity of the problem, so none is asserted here.

Anthropic also published two pieces whose titles point to its safety and security process: 'Updating our coordinated vulnerability disclosure policy timelines' and 'Building strong safety evaluations'. Only the titles reached us, so the changes to the disclosure timelines and the content of the evaluation guidance are not reported here.

Sources: Claude Code 2.1.325 release notes — Claude Code · Updating our coordinated vulnerability disclosure policy timelines — Anthropic · Building strong safety evaluations — Anthropic

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.