Claude Code’s New Mode Turns Tool Chains Into One Powerful Move!

Claude Code just gained a new way to handle complex work, and it changes what happens between a request and a finished result. Anthropic added Code Mode, an experimental opt-in feature that lets Claude write JavaScript to chain several tool calls in one step.
At the same time, Anthropic said Claude can join video calls as a visible participant through Claude Meetings, which is now in beta for Team and Enterprise plans. One update targets developer workflows inside the terminal; the other places Claude directly inside team conversations.
Claude Code Can Orchestrate Several Tools at Once
Code Mode shipped in Claude Code v2.1.323 last week. It works with Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 and Fable 5.2, but it runs only in an interactive terminal session on the native build, v2.1.323 or later.
Without Code Mode, Claude requests each Bash, Read or Edit call directly, then reads every result before choosing the next step. With Code Mode, Claude writes a piece of JavaScript, and Claude Code runs it inside your session. That lets one piece of code run commands in parallel, feed one result into the next call, and return only the output Claude needs to read.
The shift matters because multi-step work often forces an agent to spend round-trips moving between tools. A change repeated across many files, a search that needs filtering before Claude reads the results, or several independent commands that can run at the same time all fit the recommended use cases.
You still see the tool calls in the conversation, including items such as a diff, but you do not see the code by default. Press Ctrl+O to open the transcript viewer and expand a collapsed Code row.
Code Mode supports background code for up to 2 hours while you keep talking to Claude. If Claude waits, the code stops after 10 minutes, including time spent at permission prompts.
Opt-In Access Comes With Tight Boundaries
To turn on Code Mode, set CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1 in your shell or in the env block of ~/.claude/settings.json. Before the first session, Claude Code asks you to confirm. Select “Yes, turn on Code Mode,” and Claude Code restarts with the message, “Code Mode (experimental) is on”.
A Code Mode session resumes only while the variable remains set to 1. Setting it to 0 anywhere turns the feature off, even if 1 appears elsewhere. Admins can block Code Mode across an organization with a 0 in managed settings, but they cannot force it on.
- Windows needs Git for Windows.
- Code Mode requires a Claude subscription or Console account.
- It is available only through an interactive terminal.
- It does not support -p, Agent SDK, the VS Code panel, Desktop app or cloud sessions.
- It is not available through gateways, Amazon Bedrock, Claude Platform on AWS, Google Cloud or Microsoft Foundry.
- Organization support has no HIPAA configuration or zero data retention.
- Blocked settings include API key helper, OpenTelemetry export and DISABLE_TELEMETRY.
- Unavailable features include Remote Control, Claude in Chrome and computer use.
Background runs also need a session started in auto or bypassPermissions mode. Version requirements include v2.1.323 or later, and v2.1.326 or later with agent teams.
Each tool call inside the code still passes your deny and ask rules, along with PreToolUse hooks under the tool’s usual name, such as Bash. In Auto mode, a classifier reviews each piece of code rather than each tool call, and it cannot see values computed at run time, such as a path produced by search output. In other modes, code starts without review and you are prompted per tool call as usual.
Anthropic has published no benchmark or token-saving figure for Code Mode. One independent test reported system-tool context of 11.1k tokens with Code Mode versus 24.6k without, but that figure does not establish a benchmark for every workflow.
From Programmatic Tools to Claude in the Meeting
The idea of chaining tools has already moved through several forms. Cloudflare coined “Code Mode” on September 26, 2025, and its Agents SDK turns MCP schemas into a TypeScript API that runs in V8 isolates. Anthropic then published code execution with MCP, while its API feature, Programmatic Tool Calling, uses Python.
Those examples include large token reductions: one example fell from 150,000 to 2,000 tokens, a 98.7% saving, while average usage on complex research tasks dropped from 43,588 to 27,297 tokens. Claude Code applies the pattern to the CLI’s own tools, though its documentation gives no token figure.
Separate testing also points to how different AI systems can be when hunting software vulnerabilities. Echo tested Anthropic’s Claude Mythos 5.1, OpenAI’s GPT-5.6 Sol accessed through Daybreak, and Z.ai’s GLM-5.3 against three widely used open-source projects using the same vulnerability-hunting framework and validation process. Of 121 distinct potential issues, only one was found by all three models, while 83% were found by just one.
Anthropic’s other update moves Claude beyond the terminal. Claude Meetings can join video calls as a visible participant, follow the conversation and start work during the meeting. “Your team can ask Claude questions during the meeting, out loud or in chat,” Anthropic said.
“By default, Claude answers in writing, and it can use your connected tools with your permission.” Claude can also work with the team live to build docs, whiteboards or other artifacts while teammates on the invite refine them. It writes the notes and action items, so no one on the call has to.
Claude Meetings is starting to roll out in beta on Team and Enterprise plans, with support for other video conferencing platforms coming soon. Companies in the video chat space include Zoom Video (ZM), Microsoft (MSFT) and Alphabet (GOOG) (GOOGL).
Code Mode gives Claude more room to coordinate work, while Claude Meetings gives it a place to participate in that work with a team. Both features point toward an agent that does more than answer one prompt: it connects tools, follows ongoing activity and turns conversation into action.
Based on
- Anthropic Adds Code Mode to Claude Code: Claude Now Chains Tool Calls in JavaScript — marktechpost.com
- Anthropic says Claude can now join video calls as visible participant — markets.businessinsider.com
- Claude code has plenty of growth ahead, says Jim Cramer — cnbc.com
- Three Leading AI Models Hunted the Same Code for Vulnerabilities and Agreed on Just 1 in 121 | Markets Insider — markets.businessinsider.com




