Claude Discovery

← All discoveries

High-resolution image of colorful programming code highlighted on a computer screen.
Photo by Nemuel Sereti on Pexels
tool

Claude Code 2.1.233: Identity Headers, Memory Caps, and a WebFetch Cache Knob

2026-08-18 ยท source:

Claude Code 2.1.233 adds an opt-in setting to forward user identity to gateway proxies for per-user spend attribution and opt-in Linux memory cgroups for Bash commands.

What it is

Claude Code 2.1.233 is a CLI release focused on enterprise gateway controls and resource isolation for Bash-tool commands.

What it does

It adds `forward_user_identity`, an opt-in Anthropic-upstream apps gateway setting that sends the signed-in user's identity as headers so a downstream proxy can attribute spend per user, opt-in memory cgroup limits for Bash commands on Linux via `CLAUDE_CODE_TOOL_MEMORY_LIMIT` so a runaway build can't stall the whole session, GitLab merge request URL support in `--worktree` and the agents view, and a `CLAUDE_CODE_WEBFETCH_CACHE_TTL_MS` variable to tune the WebFetch session URL cache.

Why it matters

Per-user spend attribution behind a shared gateway and hard memory caps on Bash commands are the kind of guardrails that matter once a team, not a solo developer, is running Claude Code against a shared budget.

How to use it

Set `forward_user_identity` on the apps gateway config for Anthropic upstreams, and set `CLAUDE_CODE_TOOL_MEMORY_LIMIT` to cap Bash tool memory on Linux.

Go to source →
claude-codeclirelease-notesgateway