Claude Code in October: mods, clearer limits and ultrathink

Claude Code is Anthropic's AI coding assistant: it writes code, runs commands and edits projects on its own. On November 3, Anthropic's developer account ClaudeDevs posted a recap of October. Here are the highlights in plain words.

What is new

• Mods. The Claude Code interface can now be extended with your own modifications, and Claude can write them for you on request.
• Clear limits. The /usage command explains what is eating your limits, and a suggested fix applies with a single press of f.
• Savings on pauses. When a long conversation sits idle, Claude compacts it into a summary ahead of time and works from that summary when you come back.
• Fair warning. On Pro and Max plans, Claude warns you if resuming an old long conversation would use a noticeable share of the five-hour limit.
• The word ultrathink. Type it in a message and Claude uses its highest effort level (xhigh) for that one reply, without changing the setting for the whole session. Subagents, helper AI workers, can now get their own effort level.
• Fewer cut-offs. Commands that hit their timeout keep running in the background.
• Cloud and integrations. Cloud sessions start preparing while you type, and you can comment on any line of a code diff. Connected services (MCP servers) can ask a clarifying question mid-task. Max and Team plans now include a monthly Claude API credit.

The theme of the month is clear: fewer billing surprises and more autonomous work. For companies that matters most, because spending on AI assistants becomes more predictable.

Sources: ClaudeDevs on X, Claude Code changelog.