Claude Code: What's New in June 2026
A summary of all updates, new features, and bug fixes released for Claude Code during June 2026.
New Features in June 2026
- Added `footerLinksRegexes` setting for regex-matched link badges in the footer row, configurable via user or managed settings(v2.1.176)
- Enabled session titles to be generated in the language of your conversation (can be pinned via the `language` setting)(v2.1.176)
- Added the new `enforceAvailableModels` managed setting to constrain the Default model based on the `availableModels` allowlist.(v2.1.175)
- Added `wheelScrollAccelerationEnabled` setting to allow disabling mouse-wheel scroll acceleration in fullscreen mode(v2.1.174)
- Added usage attribution to the Account & usage dialog (`/usage`) in VSCode, detailing cache misses, long context, subagents, and per-skill/agent/plugin/MCP breakdowns over the last 24 hours or 7 days(v2.1.174)
- Enabled sub-agents to spawn their own sub-agents, supporting up to 5 levels of nesting.(v2.1.172)
- Added a search bar when browsing plugins in the marketplace via the /plugin command.(v2.1.172)
- Amazon Bedrock now reads the AWS region from `~/.aws` config files if `AWS_REGION` is unset, aligning with AWS SDK precedence.(v2.1.172)
- Introduced Claude Fable 5, a Mythos-class model with significantly enhanced capabilities, available by updating to version 2.1.170.(v2.1.170)
- Added the --safe-mode flag (and CLAUDE_CODE_SAFE_MODE) to start Claude Code with all customizations disabled for troubleshooting.(v2.1.169)
- Added the /cd command to move a session to a new working directory without breaking the prompt cache mid-session.(v2.1.169)
- Added a disableBundledSkills setting and CLAUDE_CODE_DISABLE_BUNDLED_SKILLS environment variable to hide bundled skills, workflows, and built-in slash commands from the model.(v2.1.169)
- Added a tip suggesting 'claude agents' when running multiple concurrent sessions.(v2.1.169)
- Added support for configuring up to three fallback models to be tried sequentially when the primary model is unavailable via the new `fallbackModel` setting, which also applies to interactive sessions via the `--fallback-model` flag.(v2.1.166)
- Enabled glob pattern support in the deny rule tool-name position, allowing for broader denial rules (e.g., "*" denies all tools).(v2.1.166)
- Introduced hardened cross-session messaging where messages relayed from other Claude sessions no longer carry user authority, causing receivers to refuse relayed permission requests and auto mode to block them.(v2.1.166)
- Enabled disabling thinking on models that think by default via the Claude API by setting `MAX_THINKING_TOKENS=0`, using `--thinking disabled`, or toggling the per-model thinking setting.(v2.1.166)
- Enabled Claude Code to automatically retry a turn once on the fallback model when the API returns an unexpected non-retryable error.(v2.1.166)
- Added `requiredMinimumVersion` and `requiredMaximumVersion` managed settings to enforce Claude Code version compliance.(v2.1.163)
- Added `/plugin list` command with `--enabled`/`--disabled` filters to view installed plugins.(v2.1.163)
- Added a "c to copy" shortcut to `/btw` that copies the raw markdown answer to the clipboard, preserving formatting.(v2.1.163)
- Enabled Hooks (Stop and SubagentStop) to return `hookSpecificOutput.additionalContext` to provide feedback and continue the turn without error.(v2.1.163)
- Enabled the use of `$` escape syntax in Skills command bodies to include a literal `$` before a digit.(v2.1.163)
- Added the `waitingFor` field to `claude agents --json` output to show what a waiting session is blocked on, such as a permission prompt.(v2.1.162)
- Enabled explicit listing of Grep/Glob tools to provide dedicated search tools on native builds with embedded search.(v2.1.162)
- Added confirmation when setting the `/effort` level as the default for new sessions.(v2.1.162)
- Changed slash command behavior: clicking a command in autocomplete now fills the prompt instead of running immediately; pressing Enter now executes the command.(v2.1.162)
- Made Remote Control status persistent as a footer pill with a link to the session, replacing the startup message.(v2.1.162)
- Enabled inclusion of custom dimension labels (like team or repo) on metric datapoints via OTEL_RESOURCE_ATTRIBUTES.(v2.1.161)
- Introduced a 'Show unused connectors' row in /mcp to collapse unused claude.ai connectors.(v2.1.161)
- Enabled independent result reporting for parallel tool calls; a failure in one Bash command no longer cancels others in the batch.(v2.1.161)
- Enhanced fullscreen mode clipboard handling on Linux to use wl-copy/xclip/xsel, copying to both clipboard and PRIMARY selection.(v2.1.161)
- Added a prompt before writing to shell startup files (.zshenv, .zlogin, .bash_login) and ~/.config/git/(v2.1.160)
- Enabled prompting before writing build-tool config files that grant code execution (.npmrc, .yarnrc*, bunfig.toml, .bazelrc, .pre-commit-config.yaml, .devcontainer/, etc.) when in `acceptEdits` mode(v2.1.160)
- Renamed the dynamic-workflow trigger keyword from `workflow` to `ultracode`; asking for a run in your own words still works, and the trigger keyword is highlighted in violet in the prompt input(v2.1.160)
Bug Fixes in June 2026
- Fixed enforcement of available models, preventing alias model picks from redirecting to blocked models via environment variables, and ensuring `/fast` respects the allowlist(v2.1.176)
- Fixed auto mode failing on Fable 5 for organizations without Opus 4.8 enabled by falling back to the best available Opus model(v2.1.176)
- Fixed hook `if` conditions for Read/Edit/Write tool paths so documented patterns like `Edit(src/**)`, `Read(~/.ssh/**)`, and `Read(.env)` now match correctly(v2.1.176)
- Fixed Linux sandbox failing to start when `.claude/settings.json` is an absolute symlink(v2.1.176)
- Fixed `/copy` and mouse-selection copy not reaching the system clipboard inside tmux over SSH, and resolved tmux paste buffer loading issues on versions older than 3.2(v2.1.176)
- Fixed Remote Control silently switching the session model when connecting from web/mobile(v2.1.176)
- Fixed Remote Control disconnect notifications showing bare numeric codes instead of human-readable reasons, and prevented duplicate lines on connection failure transcripts(v2.1.176)
- Fixed Remote Control sessions failing to disconnect when signing into a different account(v2.1.176)
- Fixed `/cd` and worktree moves leaving the session reporting the previous directory's git branch(v2.1.176)
- Fixed an issue where pressing back in one window detached other windows attached to the same `claude agents` session(v2.1.176)
- Fixed backgrounded sessions showing "Working" indefinitely when `/bg` mid-turn had no further work(v2.1.176)
- Fixed background agent search by PR URL so PRs opened during scheduled wakeups or blocked jobs now appear in `claude agents` search(v2.1.176)
- Fixed the agents view input on Windows from showing no text cursor(v2.1.176)
- Fixed `claude --bg -cn <name>` failing to seed the session name(v2.1.176)
- Fixed background sessions to neutralize Windows network paths in persisted state before respawn(v2.1.176)
- Fixed background-session respawn rejecting resume attempts from corrupted state files(v2.1.176)
- Fixed the Windows background-service daemon failing to start if `~/.claude/daemon` has the ReadOnly attribute set(v2.1.176)
- Fixed cloud sessions failing with authentication errors when idle for too long before being claimed(v2.1.176)
- Fixed the `/model` picker incorrectly hiding the model family that Default resolves to; Opus/Sonnet now correctly displays based on the user's plan(v2.1.174)
- Fixed the `/model` picker showing a hardcoded Sonnet version label when `ANTHROPIC_DEFAULT_SONNET_MODEL` is set to a different Sonnet version(v2.1.174)
- Fixed the banner incorrectly stating "Fable 5 is now consuming usage credits" for enterprise accounts using usage-based billing(v2.1.174)
- Fixed Bedrock GovCloud regions (`us-gov-*`) deriving the wrong inference profile prefix, which caused 400 errors on derived model IDs(v2.1.174)
- Fixed background sessions inheriting provider environment variables (gateway URL, custom headers, `/model` aliases) from the shell that started the background daemon(v2.1.174)
- Fixed a 1-2 second pause when exiting Claude Code shortly after a shell command was interrupted or killed on macOS and Linux(v2.1.174)
- Fixed git commit co-author attribution showing an incorrect model name for certain models(v2.1.174)
- Fixed the `/advisor` dialog pre-selecting a saved advisor model that is blocked by the `availableModels` allowlist(v2.1.174)
- Fixed Workflow tool `agent()` subagents missing per-agent attribution headers(v2.1.174)
- Fixed pre-warmed background workers failing with "Could not resolve authentication method" when claimed after being idle(v2.1.174)
- Fixed Fable 5 model names with a [1m] suffix not being normalized by automatically stripping the suffix.(v2.1.173)
- Fixed a spurious "sandbox dependencies missing" startup warning appearing on Windows when the sandbox was enabled.(v2.1.173)
- Fixed sessions using 1M context without usage credits getting permanently stuck; sessions now automatically compact under the standard context limit.(v2.1.172)
- Fixed a repeating error message about an image being removed when the conversation contained multiple images.(v2.1.172)
- Fixed the agents view incorrectly keeping a session under 'Working' with a busy spinner for up to 30 seconds after the worker responded.(v2.1.172)
- Fixed background agents potentially reading another directory's project settings (approvals, trust) when dispatched to a pre-warmed worker.(v2.1.172)
- Fixed background-session attach failing with EAUTH for sessions started on an older version after the daemon auto-updated.(v2.1.172)
- Fixed a background sub-agent remaining stuck as 'active' in the agent panel after a nested agent it spawned was stopped.(v2.1.172)
- Fixed /model suggestions in the `claude agents` dispatch input rendering with an incorrect slash prefix and showing models disabled for the organization.(v2.1.172)
- Fixed `availableModels` restrictions not being applied correctly to subagent model overrides, the agent dispatch model picker, and the advisor model.(v2.1.172)
- Fixed `availableModels` allowlists incorrectly hiding Opus and Sonnet 1M rows in the /model picker when version-specific IDs were used.(v2.1.172)
- Fixed the /model picker on Bedrock offering models the provider does not serve, which previously caused silent model switching.(v2.1.172)
- Fixed model IDs receiving a doubled 1M-context suffix when `ANTHROPIC_DEFAULT_OPUS_MODEL` already included one.(v2.1.172)
- Fixed the `opusplan` model setting not correctly enabling 1M context in plan mode for entitled users.(v2.1.172)
- Fixed wildcard domain rules (`WebFetch(domain:*.example.com)`) and file permission rules with mid-pattern wildcards not matching correctly.(v2.1.172)
- Fixed the up-arrow prompt history showing the main agent's prompts while a subagent's chat tab was active.(v2.1.172)
- Fixed memory recall failing to find mounted team memory stores (`CLAUDE_MEMORY_STORES`) in remote sessions.(v2.1.172)
- Fixed workflow validation incorrectly rejecting scripts whose prompt strings or comments contained `Date.now()` or `Math.random()`. (v2.1.172)
- Fixed the /plugin marketplace list losing cursor position after navigating out of a long list, and Esc from the plugin browser returning to the wrong tab.(v2.1.172)
- Fixed PowerShell tool calls rendering as raw JSON instead of a command display and permission dialog in VSCode, and stripped ANSI escape codes from displayed shell output.(v2.1.172)
- Fixed issue where session transcripts were not saving or appearing when using the --resume command if the session was launched from the VS Code integrated terminal or a shell inheriting Claude Code environment variables.(v2.1.170)
- Fixed Up/Down arrows jumping past wrapped rows of long input lines; they now move through each visual row first, and history recall enters at the near edge.(v2.1.169)
- Fixed enterprise managed MCP policies not being enforced correctly across various session states (reconnect, IDE-typed configs, initial session after install, before remote settings load) and fixed slow cold starts for orgs without remote settings.(v2.1.169)
- Fixed a ~30-50ms UI stall at the start of each turn for macOS users logged in with claude.ai credentials.(v2.1.169)
- Fixed 'claude -p' being slow or hanging on Windows while waiting for the slash-command/skill scan (regression).(v2.1.169)
- Fixed Remote Control getting stuck on "reconnecting" after resuming a session when an OAuth token refresh occurred simultaneously.(v2.1.169)
- Fixed Git Credential Manager's "Connect to GitHub" popup appearing on Windows at startup when background git commands ran without cached credentials.(v2.1.169)
- Fixed footer hints (e.g. "esc to interrupt") not showing for users with a custom statusline.(v2.1.169)
- Fixed stale permission and dialog prompts reappearing every time you reattached to a remote session whose worker had died while waiting on them.(v2.1.169)
- Fixed 'claude agents --json' omitting blocked and just-dispatched background sessions; added --all to include completed sessions, plus new id and state fields.(v2.1.169)
- Fixed agents view leaving a stale/garbled frame after navigating back from an agent on WSL in Windows Terminal.(v2.1.169)
- Fixed background agents ignoring project-level settings env values (e.g. ANTHROPIC_MODEL) when dispatched onto a pre-warmed worker.(v2.1.169)
- Fixed MCPB plugin cache being spuriously invalidated on Windows, causing unnecessary re-extraction.(v2.1.169)
- Fixed plugin .in_use PID lock files accumulating; stale markers from crashed sessions are now swept once per day.(v2.1.169)
- Fixed untrusted project settings being able to set OTEL client-certificate paths without trust confirmation.(v2.1.169)
- Auto-updater on Windows now stops retrying within a session once 'claude.exe' is held by another process.(v2.1.169)
- Fixed various bugs and improved overall reliability.(v2.1.168)
- Fixed various bugs and improved overall reliability.(v2.1.167)
- Fixed a recurring "image could not be processed" error and associated extra token usage when an unprocessable image was sent in a session.(v2.1.166)
- Fixed remote sessions becoming permanently stuck during startup if a brief backend disruption occurred while the worker was registering.(v2.1.166)
- Fixed Shift+non-ASCII characters (e.g., Shift+ä → Ä) being dropped when typing in terminals using the Kitty keyboard protocol (like WezTerm or Ghostty).(v2.1.166)
- Fixed PowerShell command validation occasionally hanging excessively on Windows when a killed process's children held its output pipes.(v2.1.166)
- Fixed orphaned `claude --bg-pty-host` processes consuming 100% CPU after the daemon died while a connection was active on macOS.(v2.1.166)
- Fixed voice mode incorrectly requiring `/login` to clear a stale authentication check after toggling `/voice".(v2.1.166)
- Fixed managed settings with one invalid entry silently disabling enforcement of their remaining valid policies.(v2.1.166)
- Fixed managed-settings predicates `allowedMcpServers`/`deniedMcpServers` failing to match when they utilized ${VAR} references.(v2.1.166)
- Fixed background agent sessions that entered a git worktree crash-looping with "No conversation found" upon reopening from the `claude agents` list.(v2.1.166)
- Fixed duplicated thinking text appearing in the Ctrl+O transcript view during streaming.(v2.1.166)
- Fixed `/doctor` incorrectly reporting a failed "Not inside a remote session" check when executed inside an active remote session.(v2.1.166)
- Fixed the cursor sticking at the end of the first line when typing a multiline prompt within the `claude agents` dispatch and reply inputs.(v2.1.166)
- Fixed blank lines appearing between background agent rows in the task list on terminals lacking Unicode support.(v2.1.166)
- Fixed various bugs and improved overall reliability.(v2.1.165)
- Fixed `claude -p` hanging indefinitely after its final result when a backgrounded command never exits by stopping background shells after ~5s.(v2.1.163)
- Fixed `claude -p` failing with "ANTHROPIC_API_KEY required" on Bedrock/Vertex/Foundry when `CI=true` and no Anthropic API key is set.(v2.1.163)
- Fixed Bash commands failing under bazel and EDR-protected Go workflows by correcting the override of `$TMPDIR` to only apply to sandboxed commands.(v2.1.163)
- Fixed Bash commands failing on Windows with "EEXIST: file already exists" on the session-env directory when it was read-only or inside OneDrive.(v2.1.163)
- Fixed org-managed permission rules not applying for the entire session when managed settings fetch completed during startup on a fresh config directory.(v2.1.163)
- Fixed background sessions in `claude agents` losing running background tasks when reattaching after a Claude Code update.(v2.1.163)
- Fixed terminal misalignment and a multi-second hang when exiting the agent view by pressing Esc.(v2.1.163)
- Fixed clicking Stop on a background-task chip in the desktop app not clearing the chip when the underlying process was already gone.(v2.1.163)
- Fixed keyboard input becoming permanently unresponsive after a paste operation whose end marker was dropped by the terminal.(v2.1.163)
- Fixed hook `if: "Bash(...)"` conditions firing incorrectly by ensuring the pattern matches commands inside subshells and backticks.(v2.1.163)
- Fixed deny rules on home-directory paths (e.g. `Read(~/Desktop/**)`) not blocking Bash commands that reference the path via `$HOME`.(v2.1.163)
- Fixed a silent startup hang when the configuration directory is read-only or unwritable; Claude Code now starts with in-memory config and shows startup errors.(v2.1.162)
- Fixed WebFetch permission rules not applying to built-in preapproved domains; explicit deny/ask/allow rules now correctly override the auto-allow.(v2.1.162)
- Fixed Windows permission rules failing to match when paths used backslashes (\, \\server\share) or varied case, and fixed Read deny rules failing to hide files from Glob/Grep results.(v2.1.162)
- Fixed an interrupt (Esc) sent at the very start of a turn being silently dropped in stream-json/SDK sessions, resulting in no "Interrupted" feedback.(v2.1.162)
- Fixed API 400 "no low surrogate in string" errors occurring for classifier side-queries and MCP server descriptions containing emoji near a truncation boundary.(v2.1.162)
- Fixed MCP per-server `timeout` values below 1000 ms being incorrectly floored to 1 second; sub-1000 ms values are now ignored, falling back to the default timeout.(v2.1.162)
- Fixed the LSP tool's `workspaceSymbol` operation returning no results; it now correctly accepts and passes a `query` parameter to the language server.(v2.1.162)
- Fixed `claude agents` cutting live status text (tool args, replies, prompts, exec output) short on wide terminals; status detail now uses the full terminal width.(v2.1.162)
- Fixed `claude agents` truncating long session names at 40 columns; the name column now adjusts based on terminal width.(v2.1.162)
- Fixed `claude agents` attach occasionally failing to connect on the first try after a background-service restart.(v2.1.162)
- Fixed Ctrl+V image pasting doing nothing in the `claude agents` dispatch input and reply box; pasting non-image content now shows a hint.(v2.1.162)
- Fixed backgrounding a session with ← silently losing the conversation if the background service failed to start; the session now remains in the list as a failed row that can be woken with Enter.(v2.1.162)
- Fixed replies from the agents view that fail to send being lost; they are now queued for delivery upon the next session start.(v2.1.162)
- Fixed cross-session messaging (`SendMessage`) breaking silently when temporary directory variables pointed to deep paths.(v2.1.162)
- Fixed opening a running background session from `claude agents` stalling for 5 seconds before attaching.(v2.1.162)
- Fixed workflow animations, prompt keyword shimmer, and the /effort dialog ignoring the "Reduce motion" setting.(v2.1.161)
- Fixed managed-settings policies (forceLoginOrgUUID/forceLoginMethod) incorrectly blocking third-party provider sessions when the org pin was active.(v2.1.161)
- Fixed background subagent output corrupting stdout when using --output-format text or json with claude -p.(v2.1.161)
- Fixed /usage-credits incorrectly initiating a re-login for Team/Enterprise admins instead of navigating to the organization's usage settings.(v2.1.161)
- Fixed /autofix-pr incorrectly reporting it cannot run on the default branch when operating inside a git worktree or another repository.(v2.1.161)
- Fixed the --resume picker failing to show sessions from the current directory if it was not a git worktree (e.g., jj workspaces).(v2.1.161)
- Fixed Windows hooks that explicitly invoke bash (e.g., /usr/bin/bash script.sh) failing with command not found errors.(v2.1.161)
- Fixed OpenTelemetry log events being silently dropped if emitted before telemetry initialization was complete.(v2.1.161)
- Fixed claude mcp list/get/add commands from printing secrets to the terminal by redacting credential headers, URL secrets, and preventing expansion of ${VAR} references.(v2.1.161)
- Fixed Workflow agents spawned with isolation: "worktree" being blocked from editing files inside their own worktree during background sessions.(v2.1.161)
- Fixed background sessions dispatched from claude agents booting using a stale model from the daemon's environment instead of the model specified in settings.json.(v2.1.161)
- Fixed a potential crash that occurred when rendering Write tool results after resuming a session.(v2.1.161)
- Fixed completed subagents getting stuck showing as running when an error occurred during result finalization.(v2.1.161)
- Fixed EADDRINUSE errors for tools binding Unix sockets under $TMPDIR when CLAUDE_CODE_TMPDIR was set to a deep path.(v2.1.161)
- Fixed copy-on-select not writing to the Windows clipboard on WSL by using PowerShell interop instead of OSC 52(v2.1.160)
- Fixed restoring a completed session from `claude agents` dropping chat history and re-running the original prompt(v2.1.160)
- Fixed background sessions re-attached after overnight retire losing their conversation and re-running the original prompt(v2.1.160)
- Fixed `claude --bg` occasionally failing with "socket missing" when the background daemon was cold-starting on a loaded machine(v2.1.160)
- Fixed an issue on Windows where the directory a background session was started in could not be deleted after `claude rm` until the background daemon exited(v2.1.160)
- Fixed background agents that resumed work being shown under Completed in the agents list(v2.1.160)
- Fixed `claude agents` freezing for several seconds when returning to the session list due to the auto-updater re-checking on every exit(v2.1.160)
- Fixed Esc, arrow keys, and typing becoming unresponsive on Windows when attached to a background session or in the agent view while the host is under heavy CPU load(v2.1.160)
- Fixed background agents emitting terminal sync-output markers to unsupported terminals (Apple Terminal, tmux), preventing render artifacts when entering a running agent(v2.1.160)
- Fixed mouse wheel scrolling prompt history instead of the transcript right after opening a session from the agents list(v2.1.160)
- Fixed CJK IME composition appearing at the bottom-left of the screen instead of at the input caret in the `claude agents` view(v2.1.160)
- Fixed valid `file:///C:/...` links being rewritten to a broken path on Windows terminals with hyperlink support(v2.1.160)
- Fixed voice mode failing to connect when the project directory or branch name contains non-ASCII or special characters(v2.1.160)
- Fixed the auto mode unavailability message on third-party providers (Bedrock/Vertex/Foundry) to point to the `CLAUDE_CODE_ENABLE_AUTO_MODE` opt-in(v2.1.160)
- Fixed `/effort ultracode` incorrectly blaming the dynamic workflows setting when the model cannot run xhigh; ultracode is no longer offered on models that do not support it(v2.1.160)
- Fixed model-not-found errors suggesting `--model` when running via the SDK or other hosts where the CLI flag doesn't apply(v2.1.160)
- Fixed Claude's past replies disappearing from scrollback when resuming a brief mode session with brief mode turned off(v2.1.160)
- Fixed vim mode `p` pasting on the line below instead of at the cursor when the register was yanked with `v$`(v2.1.160)
Improvements in June 2026
- Improved Bedrock credential caching: credentials from `awsCredentialExport` are now cached until their `Expiration` time instead of a fixed 1 hour(v2.1.176)
- Background sessions now show clearer guidance when a window left open across an auto-update cannot submit a reply, and `claude daemon status` now explains version-skew behavior(v2.1.176)
- When the Default model resolves to a disallowed model with `enforceAvailableModels` enabled, it now correctly falls back to the first allowed model.(v2.1.175)
- Fixed skill hot-reload re-sending the entire skill listing when only a single skill changed; now only changed skills are re-announced(v2.1.174)
- Added the `model` attribute to the `claude_code.lines_of_code.count` OTEL metric.(v2.1.172)
- Disabled mouse tracking on Windows consoles that do not fully support it.(v2.1.172)
- Improved performance in long conversations by removing redundant message normalization and avoiding full history transforms when streaming tool-use state is unchanged.(v2.1.172)
- Reduced idle CPU usage by preventing the /goal status chip from re-rendering the terminal at 5 Hz while idle, and reducing UI re-renders when subagents run in parallel.(v2.1.172)
- Improved Claude in Chrome tool loading by batching browser tool calls into a single request.(v2.1.172)
- Improved the non-interactive Usage Policy refusal message to suggest starting a new session or changing the model.(v2.1.172)
- The /code-review command now keeps the `ultra` option visible when not signed into claude.ai, with an explanation that cloud review requires an account.(v2.1.172)
- Shortened the Remote Control footer indicator to "/rc active" and hid it on narrow terminals.(v2.1.172)
- Stopped promoting /loop in remote sessions, as pending loops do not keep the container alive.(v2.1.172)
- /workflows now opens immediately even while a turn is in progress.(v2.1.169)
- Improved TaskCreate reliability: malformed inputs are repaired automatically and validation errors for unloaded tools include the schema.(v2.1.169)
- Improved the error message shown when your organization has disabled API key authentication, with guidance based on where the active API key comes from.(v2.1.169)
- Reduced CPU usage while responses stream and during spinner animations.(v2.1.169)
- Restored a default 5-minute idle timeout on Vertex/Foundry so a stalled stream aborts instead of hanging indefinitely; set API_FORCE_IDLE_TIMEOUT=0 to opt out.(v2.1.169)
- Remote-managed settings with an invalid entry now apply their remaining valid policies and surface the validation error, instead of silently dropping the whole payload.(v2.1.169)
- Background sessions now preserve --ide, --chrome, --bare, --remote-control, and other flags across retire→wake, and respawn state validation was hardened.(v2.1.169)
- Background sessions are now told that shared-checkout edits are blocked until they enter a worktree, avoiding a wasted rejected edit before EnterWorktree.(v2.1.169)
- The "CLAUDE.md is too long" warning threshold now scales with the model's context window.(v2.1.169)
- Improved color contrast for skill tags in the slash-command menu.(v2.1.169)
- Promo credit claims for Apple/Google-billed subscribers without a payment method now explain where to add one.(v2.1.169)
- The `claude update` command now announces the target version before starting the download, improving user feedback during updates.(v2.1.166)
- Typing a URL into the agent list now correctly filters to the session whose first prompt contained that URL.(v2.1.166)
- Fixed flickering issues observed in JetBrains IDE terminals (IntelliJ, PyCharm, WebStorm, etc.) on version 2026.1+ by enabling synchronized output.(v2.1.166)
- stdio MCP servers now receive the same `CLAUDE_CODE_SESSION_ID` as hooks/Bash on `--resume`.(v2.1.163)
- Background agent sessions now update to a new Claude Code version in the background, preventing waits on cold restarts upon opening.(v2.1.163)
- Provided clearer descriptions for built-in commands and skills in the / menu.(v2.1.163)
- The subscription-switch suggestion now appears in the startup announcement slot instead of a toast notification.(v2.1.163)
- Agent sessions dispatched from the state-grouped view now correctly start in the directory the agent view was opened from.(v2.1.163)
- Fixed a stray "(no content)" line left in the transcript after closing panel dialogs like /mcp and /plugins.(v2.1.163)
- Renamed Windsurf to Devin Desktop across the `/ide` menu, `/terminal-setup`, and `/scroll-speed` commands to align with the editor's rebrand.(v2.1.162)
- Made startup quieter by grouping notices by severity and consolidating session info/announcements onto a single line per launch.(v2.1.162)
- Rewrote startup warnings to be shorter, clearer, and include concrete fixes.(v2.1.162)
- Launch-prompt warnings (deep link/pre-filled prompt) now remain pinned below the input until acted upon, preventing them from scrolling away.(v2.1.162)
- Failed turns now display a compact warning line instead of a multi-line red error block.(v2.1.162)
- Improved background service startup and `claude update` verification to wait for endpoint-security scanning of new binaries instead of timing out prematurely.(v2.1.162)
- Background dispatch spawn failures now report the error class name when an errno is unavailable.(v2.1.162)
- Removed the "Claude in Chrome enabled" and "marketplace installed" startup messages; model auto-updates and team onboarding tips are now shown as quiet notices under the logo.(v2.1.162)
- Updated claude agents rows to show done/total status before details when work is fanned out, with peek showing the longest-running item.(v2.1.161)
- Improved fullscreen mode hint to show the correct key for native selection based on the terminal.(v2.1.161)
- Improved terminal rendering performance by stabilizing the layout engine's JIT compilation profile.(v2.1.161)
- Improved rendering performance specifically for large file writes.(v2.1.161)
- Added a tip in VSCode suggesting disabling terminal GPU acceleration or running /terminal-setup to resolve garbled glyphs.(v2.1.161)
- Edit no longer requires a separate Read after viewing a file with `grep`: single-file `grep`/`egrep`/`fgrep` commands now satisfy the read-before-edit check(v2.1.160)
- Improved performance of opening recently-inactive background agent sessions in `claude agents`(v2.1.160)
- Improved auto mode classifier latency by reducing reasoning on routine actions, lowering the chance of "could not evaluate this action" blocks(v2.1.160)
- Improved background-session teardown (`claude rm`/`stop`, idle reap) to send SIGTERM to running shell subprocesses before SIGKILL, so cleanup handlers run(v2.1.160)
- Removed `CLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDE`; the environment variable is now a no-op(v2.1.160)
- Removed the JetBrains plugin install suggestion from startup(v2.1.160)
All Releases in June 2026
v2.1.177v2.1.1762 features18 fixesThis release introduces configurable link badges in the footer and ensures session titles are generated in the conversation's language. Numerous critical bugs were fixed across model enforcement, remote control stability, background session management, and Linux/Windows compatibility. Credential caching for Bedrock has also been significantly improved to respect expiration times.
v2.1.1751 featureThis release introduces the `enforceAvailableModels` managed setting, which tightens security by ensuring the Default model respects the configured `availableModels` allowlist. If the default model is disallowed, it now gracefully falls back to the first available model.
v2.1.1742 features10 fixesThis release introduces a new setting to control mouse-wheel scroll acceleration in fullscreen mode and enhances usage transparency by adding detailed attribution breakdowns to the VSCode Account & usage dialog. Several critical bugs were fixed, including issues with model selection display, background session environment inheritance, and authentication failures for idle workers.
v2.1.1732 fixesThis release focuses on stability and compatibility for Fable 5 users by correctly handling model name suffixes. Additionally, a misleading startup warning related to sandbox dependencies on Windows has been resolved.
v2.1.1723 features18 fixesThis release introduces significant hierarchical agent capabilities, allowing sub-agents to spawn nested sub-agents up to five levels deep. Numerous bug fixes address stability issues related to 1M context sessions, model selection across various interfaces, and background agent operations. Performance has also been improved, particularly in long conversations and reducing idle CPU consumption.
v2.1.1701 feature1 fixThis release introduces Claude Fable 5, a powerful new Mythos-class model, accessible by updating to version 2.1.170. A critical bug affecting session saving and resumption when launched from integrated terminals or specific shell environments has also been resolved.
v2.1.1694 features15 fixesThis release introduces new configuration options for troubleshooting and managing bundled skills, including a new --safe-mode flag and a setting to disable built-in commands. Several critical bugs were fixed, particularly around MCP policy enforcement, UI responsiveness on macOS and Windows, and agent session management. Performance has also been improved with reduced CPU usage during streaming and better handling of long input lines.
v2.1.1681 fixThis release focuses primarily on stability and reliability improvements. Users should experience fewer unexpected issues and a more robust coding environment.
v2.1.1671 fixThis release focuses primarily on stability and reliability improvements. Users should experience fewer unexpected issues and a more robust coding environment.
v2.1.1665 features13 fixesThis release introduces robust fallback model support, allowing sessions to automatically retry on secondary models when the primary API endpoint is overloaded. Security has been enhanced through hardened cross-session messaging, and several critical bugs affecting terminal input, remote sessions, and managed settings have been resolved. Additionally, glob pattern support has been added to tool deny rules for more flexible configuration.
v2.1.1651 fixThis release focuses primarily on stability and reliability improvements. Users should experience fewer unexpected issues and a more robust coding environment.
v2.1.1635 features11 fixesThis release introduces new managed settings for version enforcement and adds the `/plugin list` command for better plugin management. Several critical bugs were fixed, including hangs in background processes and permission rule application issues. Additionally, background agent sessions now update seamlessly without blocking the user upon restart.
v2.1.1625 features15 fixesThis release focuses heavily on stability and user experience within the agents view and startup process. Key changes include fixing numerous bugs related to permission handling, agent status display, and session management reliability. Users will notice clearer startup messages, improved handling of failed turns, and better behavior when interacting with slash commands.
v2.1.1614 features14 fixesThis release introduces better metric visibility by including custom dimensions on datapoints and improves parallel tool execution reliability by ensuring failures don't halt other concurrent calls. Numerous bugs were fixed, including issues with motion reduction settings, policy regressions affecting third-party providers, and various session/output corruption problems across different operating systems and environments.
v2.1.1603 features18 fixesThis release introduces new safety prompts before writing to sensitive startup and configuration files, enhancing security during code generation. Numerous bugs related to background session management, Windows clipboard interaction, and agent view responsiveness have been fixed. Performance has also been improved for opening inactive agent sessions and for the auto mode classifier.