ToolingPublished All news from that day
Claude Code adds opt-in Code Mode, where Claude writes JavaScript to call tools
Anthropic's experimental Code Mode lets Claude chain tool calls in one step; it is off by default, terminal-only and blockable by an organisation.
Check our sources · 12 facts from 2 sourcesClaude Code can now let Claude write JavaScript that makes several tool calls in one step, in an experimental mode called Code Mode that is off by default. Anthropic's Claude Code changelog lists it as added in version 2.1.323 on 12 November 2026, and its documentation says it may change or be removed in any release.
Anthropic says one piece of code can run commands in parallel, use one result in the next call and return only the part of the output Claude needs to read. It suggests trying it on changes that repeat across many files, searches whose results need filtering, and several independent commands that can run at the same time. The documentation gives no figures for token or time savings.
To turn it on, set the environment variable CLAUDE_CODE_EXPERIMENTAL_CODE_MODE to 1; Claude Code ignores the variable in a project's settings files, so a cloned repository cannot enable it. An organisation can block it by setting the variable to 0 in managed settings.
It works only in an interactive terminal session, with Claude Opus 5.5, Sonnet 5.5, Haiku 5.5 or Fable 5.2, and through the Anthropic API directly; it does not work with -p, the Agent SDK, the VS Code panel, the Desktop app, cloud sessions, Amazon Bedrock or Google Cloud's Agent Platform.
In auto mode, Anthropic says the classifier reviews the code rather than each tool call, and code that is not run in the background stops after ten minutes.
Underlined sentences link to the facts we checked. Dotted words open a short explainer.
Fact 5 of 12 · from Anthropic (Claude Code docs), 18 Nov 2026
One piece of code can run commands in parallel, use one result in the next call, and return the part of the output that Claude needs to read.
Fact 8 of 12 · from Anthropic (Claude Code docs), 18 Nov 2026
Claude Code ignores this variable in a project's settings files, so a repository you clone can't turn on Code Mode for you.
Fact 12 of 12 · from Anthropic (Claude Code docs), 18 Nov 2026
the classifier reviews each piece of code before it runs, not each tool call
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against these 2 sources.
1 Claude Code changelog
Open the source archived copy-
Claude Code's changelog lists version 2.1.323, dated 12 November 2026, as adding Code Mode, an experimental opt-in in which Claude writes JavaScript to call its tools.
citeAdded Code Mode, an experimental opt-in in which Claude writes JavaScript to call its tools
-
Claude Code's changelog lists later Code Mode changes in versions 2.1.326 (16 November 2026) and 2.1.327 (17 November 2026), including support for turning on Code Mode while agent teams are enabled.
citeAdded support for turning on Code Mode while agent teams are enabled; if you already turned it on, sessions with teams on now start in Code Mode
2 Run tool calls from code with Code Mode
Open the source archived copy-
Anthropic's documentation says Code Mode is experimental and off by default and may change or be removed in any release.
citeCode Mode is experimental and off by default. It may change or be removed in any release.
-
Anthropic says that with Code Mode Claude writes a piece of JavaScript that makes the tool calls and Claude Code runs that code in the user's session.
citeWith Code Mode, Claude writes a piece of JavaScript that makes the tool calls, and Claude Code runs that code in your session.
-
Anthropic says one piece of code can run commands in parallel, use one result in the next call, and return the part of the output that Claude needs to read.
citeOne piece of code can run commands in parallel, use one result in the next call, and return the part of the output that Claude needs to read.
-
Anthropic suggests trying Code Mode on tasks that take many related tool calls, such as a change that repeats across many files, a search whose results need filtering before Claude reads them, and several independent commands that can run at the same time.
citeTry Code Mode on tasks that take many related tool calls, such as:
-
The Code Mode documentation page gives no figures for token or time savings.
citeCode Mode is experimental and off by default.
-
Anthropic says to turn on Code Mode by setting the CLAUDE_CODE_EXPERIMENTAL_CODE_MODE environment variable to 1, and that Claude Code ignores this variable in a project's settings files so a cloned repository cannot turn it on.
citeClaude Code ignores this variable in a project's settings files, so a repository you clone can't turn on Code Mode for you.
-
Anthropic says an organization can block Code Mode by adding CLAUDE_CODE_EXPERIMENTAL_CODE_MODE with the value 0 to its managed settings.
citeAdd CLAUDE_CODE_EXPERIMENTAL_CODE_MODE with the value 0 to your managed settings so that nobody in your organization can use Code Mode
-
Anthropic lists the supported models as Claude Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.2, and says Code Mode needs Claude Code v2.1.323 or later.
citeClaude Opus 5.5, Sonnet 5.5, Haiku 5.5, or Fable 5.2
-
Anthropic says Code Mode doesn't work with -p, the Agent SDK, the VS Code panel, the Desktop app or cloud sessions, and can't be used through a gateway, Amazon Bedrock, Claude Platform on AWS, Google Cloud's Agent Platform, Microsoft Foundry, or Claude Models on Google Cloud (Anthropic managed).
citeCode Mode doesn't work with `-p`, the Agent SDK, the VS Code panel, the Desktop app, or cloud sessions
-
Anthropic says that in auto mode the classifier reviews each piece of code before it runs, not each tool call, and that code that is not run in the background stops after ten minutes.
citethe classifier reviews each piece of code before it runs, not each tool call
Every sentence above is checked against these 2 sources. Each fact has its own address you can cite.
Source 1 · Anthropic (Claude Code docs) · 22 Sep 2026
Claude Code changelog
Claude Code's changelog lists version 2.1.323, dated 12 November 2026, as adding Code Mode, an experimental opt-in in which Claude writes JavaScript to call its tools.
Added Code Mode, an experimental opt-in in which Claude writes JavaScript to call its tools
Claude Code's changelog lists later Code Mode changes in versions 2.1.326 (16 November 2026) and 2.1.327 (17 November 2026), including support for turning on Code Mode while agent teams are enabled.
Added support for turning on Code Mode while agent teams are enabled; if you already turned it on, sessions with teams on now start in Code Mode
Source 2 · Anthropic (Claude Code docs) · 18 Nov 2026
Run tool calls from code with Code Mode
Anthropic's documentation says Code Mode is experimental and off by default and may change or be removed in any release.
Code Mode is experimental and off by default. It may change or be removed in any release.
Anthropic says that with Code Mode Claude writes a piece of JavaScript that makes the tool calls and Claude Code runs that code in the user's session.
With Code Mode, Claude writes a piece of JavaScript that makes the tool calls, and Claude Code runs that code in your session.
Anthropic says one piece of code can run commands in parallel, use one result in the next call, and return the part of the output that Claude needs to read.
One piece of code can run commands in parallel, use one result in the next call, and return the part of the output that Claude needs to read.
Show all 10 factsShow fewer facts
Anthropic suggests trying Code Mode on tasks that take many related tool calls, such as a change that repeats across many files, a search whose results need filtering before Claude reads them, and several independent commands that can run at the same time.
Try Code Mode on tasks that take many related tool calls, such as:
The Code Mode documentation page gives no figures for token or time savings.
Code Mode is experimental and off by default.
Anthropic says to turn on Code Mode by setting the CLAUDE_CODE_EXPERIMENTAL_CODE_MODE environment variable to 1, and that Claude Code ignores this variable in a project's settings files so a cloned repository cannot turn it on.
Claude Code ignores this variable in a project's settings files, so a repository you clone can't turn on Code Mode for you.
Anthropic says an organization can block Code Mode by adding CLAUDE_CODE_EXPERIMENTAL_CODE_MODE with the value 0 to its managed settings.
Add CLAUDE_CODE_EXPERIMENTAL_CODE_MODE with the value 0 to your managed settings so that nobody in your organization can use Code Mode
Anthropic lists the supported models as Claude Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.2, and says Code Mode needs Claude Code v2.1.323 or later.
Claude Opus 5.5, Sonnet 5.5, Haiku 5.5, or Fable 5.2
Anthropic says Code Mode doesn't work with -p, the Agent SDK, the VS Code panel, the Desktop app or cloud sessions, and can't be used through a gateway, Amazon Bedrock, Claude Platform on AWS, Google Cloud's Agent Platform, Microsoft Foundry, or Claude Models on Google Cloud (Anthropic managed).
Code Mode doesn't work with `-p`, the Agent SDK, the VS Code panel, the Desktop app, or cloud sessions
Anthropic says that in auto mode the classifier reviews each piece of code before it runs, not each tool call, and that code that is not run in the background stops after ten minutes.
the classifier reviews each piece of code before it runs, not each tool call
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep