> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# 9 Claude Code updates from October worth turning on today
- URL: https://globalfeed.ai/en/9-claude-code-updates-from-october-worth-turning-on-today/
- Published: 2026-11-03T21:42:17.000Z
- Updated: 2026-11-03T21:42:17.000Z
- Description: Mods, idle compaction, ultrathink, /usage fixes and monthly API credits on Max, Ultra and Team. What each one does and how to turn it on.
- Author: GlobalFeed Editor
- Tags: Claude Code, Anthropic, guide, software development, x-Anthropic, dil-en, elle

The Claude Code team shipped almost daily in October: dozens of releases between 2.1.287 on 1 October and 2.1.316 on 31 October. In a month that busy, the changes that matter most are easy to miss. We read the [changelog](https://code.claude.com/docs/en/changelog?ref=globalfeed.ai) and the related docs end to end and picked nine updates that make a difference in daily use, with what each one does and how to turn it on. One of them goes straight to your wallet: Max, Ultra and Team plans now come with monthly API credits.

## 1\. Mods: turn Claude Code into your own tool

The biggest addition of the month arrived on 1 October: Claude Mods. [According to the docs](https://code.claude.com/docs/en/plugins/mods/overview?ref=globalfeed.ai), a mod is a plugin that changes how Claude Code looks and behaves. It is made of JavaScript or TypeScript handlers that Claude Code calls when something happens, such as a tool call, a submitted prompt or part of the interface being drawn. A handler can watch the event, change it or take it over. That lets you add your own pane beside the transcript, add a command that runs instantly even while Claude is working, or hold a risky tool call and ask yourself a question first. A pane that charts how full your context window is after each request is a typical example.

You do not need to know how to write one: describe what you want in a session and Claude writes the mod. To try a ready-made one, Anthropic ships a built-in mod called "You should know", where a side agent watches your back and flags things you or Claude might miss. Turn it on with `/plugin enable cc-plugin-you-should-know@builtin`. The docs are clear on one point: a mod is code that runs with your permissions, so install mods only from authors you trust.

## 2\. Coming back to an old session costs less

If you left a long session for an hour, the prompt cache had expired by the time you came back, and your next message reprocessed the whole conversation. Idle compaction, added in 2.1.299 on 14 October, fixes that: a conversation over 200K tokens that sits idle for about 54 minutes is compacted just before its one-hour prompt cache expires, so only the summary is processed when you return. It reached the VS Code extension and the Code tab of the Claude desktop app on 30 October. To turn it off, use the "Idle compaction" row in `/config` or set `"idleCompaction": false` in settings. On Pro and Max, since 19 October, Claude Code also warns you first when resuming a very long conversation would use a big share of your 5-hour limit; you can resume anyway or start fresh, and Claude can look up the old one if you refer to it.

## 3\. "ultrathink" for one hard problem

Since 2.1.311 on 26 October, typing `ultrathink` in a message raises effort to `xhigh` for that turn only, so you can make Claude think harder on one problem without changing your session setting. Typed while Claude is working, it raises the rest of that turn. Per the changelog it works on Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.1 through Anthropic's API, and it reached VS Code on 31 October. `CLAUDE_CODE_DISABLE_ULTRATHINK` turns the keyword off.

## 4\. Pick the effort for subagents

Subagents used to inherit the session's effort level. Since 31 October you can tell Claude which level to run them at: if your session is on high but you want the subagents for a task on medium, just ask. A new `CLAUDE_CODE_SUBAGENT_EFFORT` variable sets the default. Running simple search and scan work at lower effort is an easy way to make your limits last longer.

## 5\. /usage tells you what is eating your limits

On 14 October `/usage` gained warnings for max effort, workflows, always-on context and large tool results. In 2.1.305 on 20 October each warning became a one-line fix that names its source, and pressing `f` applies the fixes right there. The same fixes appear in VS Code and in `/doctor`.

## 6\. Long builds are no longer cut off

A shell command that hit its timeout used to be killed, leaving a long build or test suite half done. Since 2.1.311 it keeps running in the background, and the result shows the output so far and says the command is still running, so Claude can move on while the build finishes.

## 7\. A new switch for throwaway containers

Even in bypassPermissions mode, Claude Code still asks before `rm` and `rmdir` commands on critical paths such as your home or root directory. That is intentional. If you work in a throwaway container and want those to run without asking too, you can start Claude Code with `CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS=1` since 2.1.311\. [The permission modes docs](https://code.claude.com/docs/en/permission-modes?ref=globalfeed.ai#turn-off-the-critical-path-removal-prompt-in-bypasspermissions-mode) say it is read only from the launch environment, such as your shell, and ignored in a settings file's `env` block. The name is a fair warning: it is meant for environments with nothing to lose.

## 8\. MCP servers can check in with you mid-task

MCP servers that support it can now ask you for input in the middle of a task. On 1 October Claude Code added URL prompts for steps done in the browser, such as signing in. On 31 October cloud sessions gained a form card, so you can fill in or decline a form an MCP server asks for instead of the request being cancelled without you seeing it, and 2.1.317 on 3 November added the same for links a server asks you to visit.

## 9\. Monthly API credits on Max, Ultra and Team

The best news for builders is on the plan side. [According to the Claude Platform docs](https://platform.claude.com/docs/en/about-claude/api-credits-for-subscribers?ref=globalfeed.ai), Max includes $100 of API credits a month and Ultra $200\. Team plans get $20 per Standard seat and $100 per Premium seat, pooled into one balance capped at $500; a team with three Standard and two Premium seats gets $260 a month. The credits cover the Claude API, Claude Managed Agents, the Claude Agent SDK and the Console playground. They do not cover Claude Code or extra usage in the Claude apps, which keep running on your plan's own limits.

To claim them, link a Claude Console organization in Settings > Billing on claude.ai (Organization settings > Billing on Team). No payment method is needed, and new subscribers can claim after 7 days. Credits expire at the end of each billing cycle with no rollover, and Free, Pro and Enterprise plans are not eligible. You cannot change the linked organization yourself later, so link the one you plan to build in.

## Where to start

Most of this arrives with an update: run `claude --version` and make sure you are on 2.1.316 or later. Then open `/usage` to see what is using your limits. And if you are on Max, Ultra or Team, claim the API credits on your billing page; that is a ready monthly budget for trying your own app or agent.

Sources: [Claude Code changelog](https://code.claude.com/docs/en/changelog?ref=globalfeed.ai), [Mods docs](https://code.claude.com/docs/en/plugins/mods/overview?ref=globalfeed.ai), [Permission modes docs](https://code.claude.com/docs/en/permission-modes?ref=globalfeed.ai#turn-off-the-critical-path-removal-prompt-in-bypasspermissions-mode), [API credits docs](https://platform.claude.com/docs/en/about-claude/api-credits-for-subscribers?ref=globalfeed.ai)