What changed
Claude Code 2.1.312 shipped on October 27, 2026. On Bedrock, Vertex, Foundry and Mantle, model names Claude Code doesn’t recognize now get the shorter system prompt and are covered by automatic model switches, and prompt and agent hooks now have a concurrency cap. Over 20 fixes span overwrites of large files, session resume, effort display, agent view, cloud sessions and Claude Tag.
New features
-
oidc.user_id_claimfor the Claude apps gatewayThe gateway identified people by the id_token’s
subclaim.oidc.user_id_claimnow lets you use another claim instead, such as Microsoft Entra ID’soid. -
Click to send
In fullscreen, clicking the “to send now” hint under a queued message sends it right away, as its key does.
-
[Cloud sessions] Effort level in the routine model picker
You can now pick an effort level in the model picker when you create or edit a routine that runs in the cloud.
-
[Cloud sessions] Bash editor for setup scripts
The setup script box in cloud environment settings is now a Bash editor with colors, line numbers and a Copy button.
Key improvements
-
Shorter system prompt for unrecognized model names
Bedrock, Vertex, Foundry and Mantle now use the shorter system prompt for model names Claude Code doesn’t recognize, such as custom deployment names and inference profile ARNs, even when they front Sonnet or Haiku. Set
CLAUDE_CODE_SIMPLE_SYSTEM_PROMPT=0to keep the full prompt. -
Automatic model switches cover unrecognized model names
Automatic model switches after a flagged message now also cover model names Claude Code doesn’t recognize, such as custom deployment names. Set
CLAUDE_CODE_DISABLE_REFUSAL_FALLBACK=1to turn them off. -
Concurrency cap for prompt and agent hooks
Prompt and agent hooks now run at most 24 at a time per session, sub-agents included, plus at most 12 that have waited longer than their event’s timeout. A
timeoutunder 1 second now counts as 1 second. -
Subagent
autoCompactWindowtakes effectA subagent’s
autoCompactWindowhad no effect on the window set byCLAUDE_CODE_AUTO_COMPACT_WINDOW. It now lowers that window, though it still never raises it. -
Effort indicator for mid-task changes
The prompt’s effort indicator now says a change made mid-task may wait for your next message, and the note clears when it no longer applies.
-
WebFetch errors in cloud sessions
A host that does not resolve is now explained as possibly outside the allowed network domains, so Claude names the host to allow.
-
“5-hour limit” wording
Usage limit messages now call the 5-hour window the “5-hour limit” instead of the “session limit”.
-
Cloud session names from the server
Cloud sessions started from the CLI now get their names and branch names from the server.
-
OpenTelemetry changes
mcp_server.namenow reports the real name of fixed-URL remote MCP servers declared by plugins from the official Anthropic marketplace, instead ofcustom. Claude Code no longer sends theclaude_code.safety_eventevent. -
[VS Code] Usage limit reset times
Usage limit resets now show a clock time (“Resets at 3:10 PM”) and say “5-hour limit”.
-
[Cloud sessions] Repository step summary
Opening the step shown when Claude adds a repository to a cloud session now shows a short table of what happened instead of raw JSON.
-
[Claude Tag] Retry notice in Slack
Claude Tag’s retry notice now names the model in use, or says rate limits were hit, instead of a generic API error.
Bug fixes
Files & tools
- Claude overwriting a very large CLAUDE.md, rules file or @-included file in a subdirectory after seeing only its start; Write now asks Claude to read it all or edit in place.
- Per-directory and
paths:skills not being found after a Bashcddeeper into a git repository. - The Read tool sometimes failing on PDF pages with “pdftoppm failed: Could not write image” when the clean-up of old session files ran at the same time.
Settings & command loading
- Managed settings and organization policy not being fetched for the rest of the session after signing in from the VS Code extension.
- A command with an invalid name stopping all commands from loading, which could leave
claude -pwith no output.
Session resume & status
- The first reply after resuming a session in the terminal missing the next time the session was resumed (rare).
- Some sessions showing as waiting on you when the last message only mentioned items waiting on you mid-sentence.
Effort & model display
- The spinner, OpenTelemetry
effortvalues and transcripts naming a newly picked effort level during a task before the model applied it; they now show the level in force. - The
/modeland/effortcache warning saying the re-read waits for your next message; a change made while Claude is working can trigger it sooner.
Input & agent view
- Agent view restoring input text on the next launch after it was cleared with Ctrl+C just before exiting.
- Agent view reopening in shell mode after exiting with Ctrl+D from an empty
!input. - Vim mode
d,candywith agjorgkcount larger than the lines left deleting or copying text; like Vim, they now only move the cursor. /verifyshowing a long description meant for Claude in/helpand the/command menu instead of a short one.
VS Code extension
- Sessions ending on a bare exit code when your organization’s settings couldn’t be loaded or were refused at startup; the tab now says why and offers Retry or Sign in again.
Cloud sessions & Remote Control
- Resending an unchanged message after a failed send in a cloud session sometimes running it twice.
- A failed cloud environment creation showing its error only as a toast outside the open dialog; it now appears inside the dialog.
- Edit in a routine’s menu on the Routines page going to the routine’s page without opening its edit dialog.
- An unarchived Remote Control session showing as working until a computer reconnected; it now shows as idle.
Claude Tag
- A Claude Tag channel’s on/off switch and agent settings falling back to workspace defaults when an older private Slack channel goes public and gets a new ID.
- Claude in Slack not replying in channels whose environment has network access set to None.
- Claude in Slack sometimes posting a file in the wrong thread when it was answering in another thread of the same channel.
Code Review
- Code Review reviewing an outdated commit, or none at all, when GitHub delivered a push event late; it now reviews the pull request’s current commit.
Notes
- If you use custom deployment names or inference profile ARNs on Bedrock, Vertex, Foundry or Mantle, Claude Code now uses the shorter system prompt for them. Set
CLAUDE_CODE_SIMPLE_SYSTEM_PROMPT=0to keep the full prompt. - If you don’t want automatic model switches after a flagged message on those names, set
CLAUDE_CODE_DISABLE_REFUSAL_FALLBACK=1. - If you rely on many hooks, prompt and agent hooks now run at most 24 at a time per session, and a
timeoutunder 1 second counts as 1 second. - If you collect OpenTelemetry data,
claude_code.safety_eventis no longer sent, andmcp_server.namenow reports the real name of fixed-URL remote MCP servers from official Anthropic marketplace plugins instead ofcustom.