Skip to content

Claude Code

Claude Code 2.1.310 brings Monitor to Bedrock and Vertex

3 min read

Written by an AI model from the linked sources and checked against them by a second model. How we work

All about: Claude Code versions and changes →

Claude Code 2.1.310, out on October 23, adds the Monitor tool and plugin monitors on Bedrock, Vertex and Foundry, and in sessions where telemetry or nonessential traffic is disabled, the release notes on GitHub say. On those setups Claude can now follow a background script’s output as it prints, and the block on long foreground sleeps applies there too.

Monitor, part of Claude Code since version 2.1.98 on April 9, runs a shell command and turns each line of its stdout into a notification that wakes the main session; lines arriving within 200ms of each other come as one, and a command that stays silent costs Claude nothing. Without it, a command started with run_in_background reports once, when it finishes, and following it along the way means polling with /loop at a full API call per iteration.

Setting DISABLE_TELEMETRY or CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC, which cuts all nonessential traffic at once, also switches off Claude Code’s feature-flag evaluation, the Claude docs say. That can leave Remote Control unavailable in the same sessions where Monitor now works; DISABLE_ERROR_REPORTING has no such effect.

Gateways and long conversations

Requests used to fail behind some gateways and compatible servers that reject the effort-2025-11-24 beta value or system messages sent mid-conversation; Claude Code now retries without them. Version 2.1.305 already did the same for tool search, after a 2.1.304 regression had every request failing on Vertex, Bedrock and other providers whose endpoint or gateway rejected it.

After a “Prompt is too long” error, compaction kept too much of a conversation full of dense text such as base64 or SVG, so the next request could fail the same way; 2.1.310 fixes that. Non-interactive sessions, run with -p or through the Agent SDK, also use less memory when compacting a very large transcript file.

Until this release, setting cleanupPeriodDays to a very large value could make background cleanup delete some recent files, pasted images among them. The setting adjusts how long Claude Code keeps session transcripts, which sit in plaintext under ~/.claude/projects/ for 30 days by default so that sessions can be resumed.

Hooks and the Linux sandbox

On Windows, PreToolUse and PermissionRequest hooks could let a tool call through when Git Bash was missing and the hook’s command was only quoted strings, such as a quoted path; 2.1.310 closes that gap. On Linux, /sandbox and the sandbox startup warning now name Ubuntu’s AppArmor restriction on user namespaces, and how to lift it, when that restriction keeps bubblewrap from starting the sandbox.

File read permission prompts now show where a symbolic link leads. Auto mode’s prompt before the first read outside the working directories now says when blocking those reads makes shell commands ask for permission. And an MCP tool call that times out in a non-interactive session while the server’s elicitation, its request for the user’s input, sits unanswered now returns an error telling Claude that no answer came and not to repeat the call.

VS Code and cloud sessions

In VS Code, pressing Enter on the plan review card’s “Send feedback and keep planning” button approved the plan; it now sends your comments back. The panel also stops showing “Something went wrong” after “Switch branch & continue” on a cloud session opened from the Web tab of Session history.

Enterprise custom roles gain permissions for Claude Code cloud sessions and for Remote Control, so Owners can control access to each by role. Turning either on in admin settings now asks for confirmation and lists the custom roles that hold that permission.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.