You can now stop Claude Code compacting a long conversation while you sit idle by setting idleCompaction to false. The CLAUDE_CODE_IDLE_COMPACT_MIN_TOKENS variable sets how large a conversation must be before idle compaction starts. A new --proactivity flag picks how much Claude does on its own when it starts. CLAUDE_CODE_DISABLE_PROACTIVITY turns that choice off. Running claude remote-control with --allow-unattended-tool-calls lets cloud sessions use the current folder for one run. Remote sessions now read the SDK address from CLAUDE_CODE_REMOTE_SDK_URL when --sdk-url is not given.
This release has 7 entries in the build that are not switched on yet. One would let you attach more PDFs as a reference when their page count is known. Another would let an MCP server that needs you to sign in keep its saved tools. Claude Code could also tell the model that you pay per token and what a Bash call costs. Live-update connections could learn to spot a proxy answering in place of the real server. Each of these waits on a switch that is off by default.
Among this release's fixes, a cancel sent from outside the terminal no longer throws away every prompt you had queued. Plugin names that version 2.1.287 changed now match the installed plugins again. Synced skills now keep the text that sits above their main body. A plugin's hook error handler now still runs if the worker running those hooks is replaced mid-event. Stopping a remotely served background command now returns without waiting for it to end.
Written by our agent from the shipped bundle, not by Anthropic.
A new remote-control option lets auto-mode cloud sessions run commands in your folder without asking, sandboxed, with status messages showing what is allowed
Sessions & agentsA new --proactivity <level> option, hidden from help, sets how much Claude does on its own, and a remote session started from Claude Code is sent the same level
New idleCompaction setting can stop idle compaction, and CLAUDE_CODE_IDLE_COMPACT_MIN_TOKENS overrides its minimum token threshold
Sessions & agentsA --proactivity flag picks the proactivity level when Claude Code starts, and CLAUDE_CODE_DISABLE_PROACTIVITY turns the selector off
Setting CLAUDE_CODE_REMOVE_PROMPT_STRINGS to a JSON list removes those strings from the prompts Claude Code sends to the model
Sessions & agentsWant the reasoning? Read walks the 61 entries that probably matter to you, each one opening to what changed and why.
Read this release → Every row →28 more of these are in What probably matters to you, on page 1.
When an attach from fleet view comes from Esc, Claude Code skips the refused-fork and restart prompts
Unclear Where the Esc case comes from, and what a user sees as a result, is not known.
Background prompts for artifact comment threads now label each thread as activated for Claude or not, and tell Claude to ignore the rest
An MCP server that is not connecting can now show a set-aside status, and a PDF too big to show whole tells Claude to read it by page
Unclear What set aside means for a server, and whether the PDF message can appear more often than before, is not clear.
The tool Claude uses to start subagents now gets a budget note based on where the request came from, plus a remote switch for keeping tool results
Unclear It is not clear what the budget note says or which requests it is added to.
How deeply subagents can start other subagents can now be set from cached account data before Claude Code falls back to the remote setting
Unclear It is not clear whether a limit of at least 1 is still required.
Permission decisions now give a reason under plan mode limits and under own_record, and deletes of things the session made can skip a safety block
Unclear It is not clear what controls the self-made delete exception or when the unattended own_record behaviour applies.
When Claude calls the same tool several times in one reply and every call succeeds, the duplicates can be removed, keeping the first
Unclear It is not clear whether this step is active for anyone, or which tool it targets.
Claude Code can now build an isolated git copy with its own object store instead of always borrowing the original's
Unclear It is not clear when Claude Code chooses own mode.
A new instruction tells Claude to treat downloads and unpacked archives as untrusted and keep them apart from its own scripts
Unclear It is not clear in which situations Claude Code adds this instruction.
Learn, tag and explain suggestions are now written in your session's language or the language you write in, not English only
The confirmation for switching model or effort level has a mode where Cancel is selected first and the number shortcuts are hidden
Unclear It is not clear what puts the dialog into this Cancel-first mode.
Interactive startup waits for MCP servers that were held back, and a non-headless startup path merges in extra MCP settings and allowed tools
Unclear What triggers the extra MCP configuration and allowed-tools merge is not clear.
The background task scan can now carry over scheduled cron jobs as temporary entries that are not saved, and the task notification text changed
Unclear It is not known what passes the carried jobs to the scan, so when this happens in practice is unclear.
Claude Code can now leave the effort level unset for some models instead of taking it from your settings
Unclear What the model-specific hold value is and which models have one is unclear.
The Ready to code? dialog now preselects keeping context when proactivity is set to ask, and can hide the clear-context option
Unclear The exact condition that hides the clear-context option is not fully clear.
The hidden settings review for cloud sessions now includes ~/.claude/settings.json as well as the folder's own settings files
Unclear The command's name is not known, and it is hidden from listings.
--from-link can include another spelling of the name#When the hidden --from-link option refuses, its JSON output can now include a requested field holding another spelling of the name
A new message for cut-off replies tells Claude to restart the broken line or table row and reopen any code block, without repeating earlier text
Unclear It is not clear whether this replaces an older continuation message, or whether a setting controls when it is used.
A policy check can now put an MCP server on hold, so it is neither loaded nor blocked
Unclear It is not clear what puts a server on hold or what happens to it afterwards.
A permission request marked personOnly now skips automatic handling and goes straight to a person
Unclear It is not clear what sets the personOnly marking.
Claude Code's bundled documentation on self-hosted sandboxes for managed agents was replaced by a managed-agents overview
Unclear It is not clear whether other documentation pages were added alongside this one.
--safe-mode is now checked the same way as --restricted#The --safe-mode flag now goes through the same check as --restricted and --strict-mcp-config, run when it is needed
Unclear What difference the new check makes in practice is not known.
When a hook runs for an account memory tool, its output is replaced with a placeholder instead of entering the conversation
Unclear When this placeholder is used is not established.
A turn handoff held over 30 minutes now fails as expired, the handler now gets the handoff's age, and handoff validation was reworked
Unclear Only the description of the error was seen, not the code that enforces the 30-minute limit.
The built-in gh stand-in no longer takes the GitHub host from GH_HOST or GH_REPO; pass --hostname instead
Unclear It is not clear which condition decides whether the "as GH_HOST" wording appears.
TaskStop's instructions now say a named background agent can be stopped by name, and the ambiguous-teammate error asks for the full address (name@team)
Remote session streams now track whether a viewer is away, and closing a session ends the away state and tells connected clients it is exiting
Unclear Which feature uses these lease connections is not known.
With claude --cloud, hooks from files the cloud session can change are withheld with a notice, and hook warnings and errors are reworded
MCP connections that fail at a proxy now keep the proxy's HTTP status as proxyConnectStatus, and a 403 there no longer triggers an automatic reconnect
Unclear It is not clear where, if anywhere, this status code is shown to the user.
Remote sessions now slow partial-output streaming while the viewer is away, can renew their connection, and retry oversize stream events
Unclear It is not clear what switches on the renewal check and the viewer signals.
In agent teams, teammate tasks are now labelled by their description instead of a cut-off prompt, and record the agent type and the prompt shown
Headless and SDK sessions now report the turn's state when it is interrupted, and carry the original prompt's ID and resume reason when they resume it
A plugin toast that is still waiting to be shown is now replaced by that plugin's newer toasts, and toast text passes through a sanitizer
Unclear The exact rules for when a plugin toast waits and when it replaces another are not settled.
Skills now keep any text that comes before their main body, and reading a binary file suggests using a skill for that file type
When you leave a cloud session, the hint now shows claude --cloud <id> instead of the local claude --resume command
The command that accepts settings for cloud sessions now checks your user settings files as well as the folder's
If a reply is stopped by an output filter before much arrives, Claude Code now retries it up to a set number of times
Unclear The number of retries and where it is set are not stated.
A restarted background session can now resume as a fresh copy with --fork-session, and a saved reply starting with / or ! gets a warning
Unclear What decides whether a restart continues the original session or a copy is not clear.
Some MCP servers can now be connected ahead of the rest, and retrying continues while any server is still failing
Unclear It is not clear what decides which servers are connected first, or whether anything uses that option yet.
/code-review medium now runs the high-effort recipe on most models#On every model without its own /code-review row, Opus 5.5 included, medium now runs the eight-angle high recipe instead of a single pass. high, xhigh and Opus 5 are unchanged.
If a plugin named without a marketplace cannot be fetched, Claude Code installs the official marketplace's copy instead of failing
Unclear Which marketplaces count as official, and exactly when the fallback applies, is not stated.
Compaction now saves a snapshot of the task list, and restoring a session rebuilds tasks from that snapshot so earlier tasks are less likely to vanish
Unclear It is not clear whether this build saves the task list at compaction, which the rebuild depends on.
Cloud session start now warns when your branch is ahead of the remote, catches unusable session ids and cleans up names shown in error messages
Unclear The exact wording of the ahead-of-remote notice is not known.
A new per-call watchdog follows each MCP request and aborts the tool call with an error if its answer stream ends without an answer
Unclear It is not clear which HTTP servers are covered, or whether this tracking is active for anyone.
The error for a session bound to one device now says this machine could not prove it is that computer or a trusted device, and suggests /login
Claude Code now checks an extra condition before relaying a permission request through a channel
Unclear It is not clear what the new condition checks.
A permission log message now lists a ToolHost's person-only ask with ask rules and safety checks as needing the full permission check
Unclear Only the message wording is known to have changed, not whether the permission decision itself now behaves differently.
Instruction-file loading now keeps track of what it already loaded and reacts to more events, including files you mention with @
Unclear It is not confirmed that this fixes a folder's CLAUDE.md being added again after resuming or compacting, nor which plugin system runs this code.
EnterWorktree now checks first and refuses to switch into an existing worktree where the agent would not have write access
Claude Code now warns that a hook pattern starting with a reserved prefix will never run
Unclear The reserved prefix itself is not identified.
A command allowed only once its output redirection is removed is now rechecked as written, and refused if the full form is not allowed
Unclear It is unclear whether the redirection recheck always applies or depends on a switch that could not be identified.
Claude Code can now discard a trailing API error before giving its final result, ending the turn as a tool use
Unclear The exact condition that makes Claude Code drop the error is not known.
When Claude Code reads a repository's git config file, it now returns nothing if the content fails a check, instead of always using it
Unclear It is not clear what content the check rejects.
The footer can show a custom hint, and the rewind picker uses a different check for commands that do not reach the model
Unclear What the custom footer hint shows, and when it appears, is not stated.
--allow-dangerously-skip-permissions without recorded consent#Background sessions now ignore --allow-dangerously-skip-permissions unless bypass consent is in your user or policy settings
The BYOC runner now clears git's stale record of a missing worktree and retries, and deletes worktrees through a helper limited to a trusted folder
Unclear Whether any of this applies outside the BYOC runner is not settled.
Tool calls served to a remote session are now refused when a read-deny rule matches the path, or when the rules cannot be checked
Unclear Exactly which tools these checks cover is not clear.
When restoring a session that was in plan mode, Claude Code now checks only the recorded mode and the permission settings
Unclear Exactly which cases now restore plan mode is not clear.
A call can now be refused because its relative path could not be confirmed, and a notice explains cloud use of the folder
Unclear Where each message is shown, and what controls this behaviour, is not known.
Remotely hosted tool calls now refuse over-long paths and settings-file edits, deny on a rule-check crash, and use up approvals that arrive too late
Unclear It is not clear which of these checks apply in every setup and which depend on a setting or remote switch.
When serving tools over MCP, each tool call now runs in its own session context, and a slow workspace diff returns a structured error
Unclear What difference running each call in its own context makes for users is not clear.
In some setups, a PDF too large to include directly is now replaced by a text note giving its size instead of being dropped
Unclear Which environments count as the server supplying the model's files is not known.
The rules for when an interrupted prompt can be resumed, and for cutting a conversation at a pending request, have changed
Unclear What difference users will see when resuming is not known.
Remote Control now replaces or drops account memory content before sending session events and history to remote clients
Unclear Whether the filter is controlled by a remote setting, and what decides when it is switched off, is not known.
Permission rules with wildcards are now checked against a tool's name and its other names, replacing the old ** handling
Unclear Exactly which rules now match differently than before is not clear.
Claude Code now inspects managed settings files reached through links and warns when a link points outside the expected place
Unclear It is not clear whether the warning is shown to the user on screen or only recorded in usage data.
The git check before uploading a folder to a cloud session now refuses reftable or GIT_REFERENCE_BACKEND repos, names its reasons and hides git's own output
Unclear It is not clear which feature runs this check or what Claude Code does when a repository is treated as tampered.
Cancelling a permission prompt from a remote subagent running in the background now denies the request instead of stopping the turn
Unclear It is not clear what decides whether the answer is reported to Anthropic.
Sessions connected for remote use now strip account memory content from uploaded events and history, and handle reconnect conflicts differently
Unclear It is not clear when the account memory redaction is switched on.
Interactive sessions add a queue watcher that checks for a pending decision from you, and queued coordinator messages group under a label
Unclear It is not clear what switches the new watcher on.
Project-folder and chat attachment uploads now wait their turn, re-check policy after waiting, and report which files failed to download
Unclear It is not clear how many uploads are allowed to run at the same time.
When a response stops arriving because it was suspended, Claude Code now retries it a limited number of times before its usual retry handling
Unclear What the two new remote switches change in practice is not clear.
Auto mode stops offering permanent allow rules in more cases, and a new gate changes how git and rsync options are read in safety checks
Unclear What the remote switch's built-in default is, and what the new git option path does, are not clear.
Some settings keys can now be set by a project's settings only to one specific value, with a warning when a project tries anything else
Unclear Which keys are limited to a single value is not stated.
A startup warning now lists memory files that were held back instead of loaded
Unclear What causes a memory file to be withheld, and when the warning is shown, are not stated.
When a tool request contains details the tool does not recognise, Claude Code can now drop them and try again
Unclear Whether anything other than the plan mode check uses this is not clear.
After a container restart, Claude Code lists the scheduled timers that did not survive and asks Claude to set each one up again
Unclear What triggers this check, and whether a switch controls it, is not clear.
The footer shows 'interrupt' while Claude works and 'return to team lead' when viewing a finished teammate; background tasks can be marked as needing you
Unclear Whether the 'needs you' marker always appears or only when a particular optional part of Claude Code is present is not settled.
Before stashing for teleport, Claude Code now stops with 'Nothing was stashed' if a tracked file was replaced by a folder
Interrupting a turn paused on a permission request now starts winding it down, and waits if agents are still running
Unclear What decides whether a stopped turn is ended this way is not clear.
Refreshed MCP tool lists are no longer applied when another server holds the name, and per-server permission overrides also set ask-for-tools
Unclear It is not clear what value mcpAskForTools takes when no per-server override has been set.
In Claude in Chrome, browser_batch calls now use their own time limit instead of the default one other tools use
Unclear The actual time limits are not stated, so it is not known whether the batch limit is longer than the default.
When Claude Code can't confirm a model is valid, a model switch can now be accepted and marked as unconfirmed instead of rejected
Unclear The condition under which a switch is accepted as unconfirmed is not settled.
Remote sessions now track permission requests from background helpers, and the message about an unsent first message is reworded
Unclear Exactly how the new tracking of subagent permission requests appears to the user is unclear.
An organisation's MCP allowlist now refuses claude.ai connectors it cannot find or does not admit, and matches them by their real name
Unclear What makes a connector count as admitted is not settled.
/plugin in the desktop app explains where to go instead#Running /plugin in the desktop app now explains it cannot run there and points to the Plugins menu or the command line
/doctor shows an extra block of checks before environment variables#The /doctor report now prints a new block of paired lines, each with an indented detail, ahead of its environment variables section
Unclear It is not clear what the new block of lines actually lists.
Claude Code's check for PreToolUse hooks now matches against a wider list of names worked out from an extra tool name, not just one
Unclear It is not clear what the extra names are or what bypassCapabilityNotification does.
Attaching to a remote session without a screen now waits to confirm its settings were kept and fails with a message if not
Unclear The wording of the helper notice and when it appears are not known.
Plugins checking tool permissions now get the agent's ID and MCP tool details, and can still only tighten permissions
Claude Code can now send the session's running cost and token count to a remote session during a turn, at most once per set interval
Unclear It is not settled whether Claude Code actually sends these mid-turn updates yet.
Directory listing in the anchored file layer now has a size cap with a WORKING_LIST_TOO_LONG error, plus createAndOpen, anchorBeneath and removeTree options
Unclear What sets the limit, and how large it is, is not established.
The web search cap is now a budget that refills each hour, and in some modes it counts per turn instead of per session
Unclear What decides whether the limit counts per turn or per session is not settled.
WebSearch is limited by an hourly refilling budget instead of a simple count, adjustable with CLAUDE_CODE_WEB_SEARCH_REFILLS_PER_HOUR
Unclear The actual limits and what turns refilling on by default are not stated.
When a resumed session was still in plan mode, Claude Code can now switch plan mode back on
Unclear It is not clear what switches this on or whether it applies only to remote sessions.
Recurring scheduled tasks are recreated when you resume a session, and taken-over ones no longer fire straight away
Unclear Whether this is behind a switch, and what that switch defaults to, is not clear.
Plugins, mcp add and MCP config files now reject the claudeai-proxy server type, and claude.ai connectors wait to reconnect after an account switch
Claude Code now checks a git worktree before deleting it, and worktree error messages name fewer paths and suggest a different --worktree name
Unclear Exactly when a stopped session's worktree record is dropped is not clear.
Cloud service errors now hide long or odd server text, ignore generic messages, and point you to --debug for the raw response
Published verbatim by Anthropic for v2.1.290. Text is unmodified from the upstream changelog. Everything else on this page came out of the bundle instead, which is why the two lists don't match.
Of these 190 bullets, 30 name something an entry on this page also names, 67 name something no entry here does, and 93 name nothing specific enough to line up either way. The pairings are made on names both sides wrote down, a flag or a setting or a slash command, so read one as probably the same thing rather than as a fact, and read the middle number as candidates rather than as a miss count.
serverToolUses to the result of a mod's turn.step hook: the tool calls the API ran itself (the advisor), each with its id, name, input, start and end
No entry names this agentId to the tool.check event of plugin hooks, so a hook can tell a subagent's permission check from the main session's
No entry names this ceiling to the question and verdict a mod's tool.check hook reads, naming the approval an organization requires for a tool
No entry names this ThemeKey and Color types to the plugin hooks typings, so an editor lists the theme colors a mod's drawing can name
No entry names this claude plugin validate: each hook a mod registers at a gating site is listed with whether it has a .catch (gatingHooks under --json)
No entry names this claude attach <name> and claude logs <name>: part of a session name works in place of the id
No entry names this /claude-api managed-agents-onboard <url> to set up the Managed Agents pattern a page describes as ant apply files
Probably bundled-claude-api-skill-adds-managed-agents-quickstarts /claude-api managed-agents-onboard <quickstart-name> to build a Console quickstart template, such as deep-researcher, with the ant CLI
Probably bundled-claude-api-skill-adds-managed-agents-quickstarts offset to read on
Nothing to match on #95127[BUG] WebFetch truncation is invisible to the model — absent from the tool description, unmarked in the result, and the web-fetch subagent can neither detect nor recover from it Open
/rewind not listing a prompt sent while Claude was still working
No entry names this /loop with an interval, reminders) silently not coming back on resume once the conversation was compacted; covers compactions made from this version on
No entry names this /background hand-off, and recurring ones firing an extra run on every resume, respawn or fork
No entry names this #96531[BUG] /loop scheduled task stops firing on its own after the session is backgrounded — fires only right after a user turn Open
--json-schema runs exiting non-zero with is_error: true on a success result when the connection dropped after the structured output was already delivered
No entry names this CLAUDE.md, rule or AGENTS.md symlinked outside the working directories loading under permissions.blockReadsOutsideWorkingDirectories or a Read deny rule
Probably instruction-file-loading-de-duplicated-via-a-held-set-and, edit-tool-results-carry-gitdiffread-flag-sealed-path-case, sandbox-read-blocking-and-wsl-mount-parsing-rework, withheld-memory-files-now-shown-as-a-startup-warning, instruction-files-claudemd-rules-held-outside-the-workin, plan-exit-dialog-hides-clear-context-option-for-ask-proactiv, startupresume-telemetry-adds-store-confirm-and-system-promp, home-settings-seeded-to-cloud-sessions-ccr, hooks-module-fsancestors-reads-agentsmd-with-a-held-outs xn-- host label matching differently from one process to the next
Probably url-permission-patterns-reject-punycode-wildcard-hosts /ultrareview dropping uncommitted changes without a warning on Windows when git stash create failed, and refusing them after a git add -N file was deleted or moved
No entry names this plansDirectory setting's project-root check for paths that contain a backslash on macOS and Linux
Probably plansdirectory-rejects-backslashes-off-windows claude daemon run and claude daemon logs; they now show as \uXXXX escapes
No entry names this .catch being unloaded, and its .catch skipped, when the hook kept the hooks worker busy on a prompt or tool call
Nothing to match on turn.step result listing a tool call that a mid-response model fallback had discarded
No entry names this claude plugin validate and plugin loading refusing a hooks module that destructures an option named like one of its top-level functions
No entry names this /ultrareview failing to upload uncommitted changes when core.safecrlf=true is set in git's configuration
No entry names this ! shell blocks in skills and commands failing when the file is saved with CRLF line endings
Nothing to match on /permissions tab was clicked while searching in fullscreen mode
No entry names this answer on turn.complete for a subagent that hands its report back in auto mode
No entry names this prompt.submit hook that drops a prompt after calling next(e) being ignored silently: the hook is now reported as failed, by name
No entry names this disableClaudeAiConnectors and allowedMcpServers URL rules not being applied to some MCP entries declared in .mcp.json, plugins or agents
No entry names this @-mention under the read block or --restricted being able to read a file outside the working directories through a link changed mid-read
Probably attachment-paths-are-displayed-in-a-normalized-form-in-error, safe-mode-is-now-evaluated-lazily-and-uses-the-same-check /loop run count, countdown and live status line after the session enters a worktree that it creates
No entry names this claude agents sessions in manual permission mode asking for approval to read an image pasted into a reply or a new agent's prompt
No entry names this declare, typeset, export or readonly
Nothing to match on language setting
Nothing to match on #99232[FEATURE] "You should know" mod: respect the `language` setting instead of always writing in English Open
claude respawn re-sending an earlier message to a backgrounded session that has no saved transcript instead of starting it with an empty conversation
No entry names this n: or Ctrl+F search in the agents view moving focus to a section header, where Ctrl+X twice would delete every session in the section
Nothing to match on claude agents saving a slash command it could not deliver to a stopped session and then running it by itself the next time that session restarted
No entry names this /ultrareview uploading uncommitted changes unfiltered for files under a git filter driver named unset or unspecified; the upload now stops and asks you to rename the driver
No entry names this claude --teleport and /teleport deleting the files in a folder that had replaced a tracked file of the same name when you chose to stash: the stash is now refused, and says why
No entry names this /loop run count freezing and its countdown disappearing after /clear; the count now restarts with the new conversation
Probably bridge-remote-control-peers-can-run-effort-model-rena --channels permission relay: a reply ID that repeats within a session is now ignored instead of approving a different prompt
Probably channels-banner-shows-a-waiting-for-your-organizations-pol /chrome "Reconnect extension" not restoring browser tools after a failed Chrome connection, and added an explanation when it can't (anthropics/claude-code#98135)
Probably project-settings-can-no-longer-turn-on-claude-in-chrome-a-w, chrome-disabled-session-notice-tells-the-model-not-to-use-br, chrome-menu-reports-live-connection-state-and-reconnect-act #98135/chrome Reconnect never registers claude-in-chrome MCP tools when a session starts with the bridge down (survives --resume) Closed
ANTHROPIC_BASE_URL with ANTHROPIC_AUTH_TOKEN) and have no Anthropic account
No entry names this claude agents just after a background session crashed being refused after 2 seconds: they are now retried for up to 12 seconds while the session restarts
No entry names this claude agents could not deliver to a running session being saved and sent by themselves the next time it was restarted
No entry names this cat <<EOF | python3) asking for approval on every run
No entry names this claude agents failing with "Couldn't restart the background service" and background sessions stopping after a Homebrew upgrade (takes effect from the upgrade after this one)
No entry names this #84827Daemon self-respawn after a Homebrew cask upgrade targets the purged versioned path (ENOENT), killing every background session Open
rg or git grep) whose arguments the shell would still expand as wildcards; these now prompt for approval
No entry names this claude plugin test refusing to run after an upgrade because of an out-of-date saved setting
Probably claude-plugin-test-rollout-flag-hold-and-file-form git clone option keeping the sandbox exemption from a git pattern such as git * in sandbox.excludedCommands; it is now treated like the long form
No entry names this /ultrareview of a local branch silently leaving uncommitted work out of the upload in a repository that keeps its branches outside .git (git 2.54+); it now refuses with an explanation
No entry names this /loop wakeup or scheduled task; Claude is now told and can schedule it again
No entry names this store.postgres_url can't be parsed; the error now names the setting and says what the URL may hold
Probably gateway-gives-a-clear-error-for-an-invalid-storepostgres-ur /sandbox Config tab on Linux and WSL when a sandbox read rule such as ~/**/.env covers a large folder
No entry names this #98023[BUG] 2.1.284 freezes on first Enter: new sandbox glob expander synchronously walks all of ~ for "~/**/…" denyRead patterns (follows symlinks, unbounded memory); 2.1.280 fine Open
claude -p run from the Bash tool)
No entry names this --continue or --resume <session-id> in the terminal
Probably cloud-sessions-print-a-claude-cloud-resume-hint, resume-with-resume-print-drop-turn-guard-adds-an-attach, background-job-respawn-can-fork-a-resume, remote-control-gains-allow-unattended-tool-calls-and-a-for /rewind) freezing for hundreds of milliseconds per keypress when the conversation contains a very large pasted stack trace or source file
No entry names this CLAUDE_CODE_USER_DIALOG_TIMEOUT_MS=5m being read as 5 ms and cancelling remote dialogs at once; values with a unit suffix now fall back to dialogExpiry
Probably chrome-permission-env-defaults-and-remote-tools-unattended-h --restricted (and CLAUDE_CODE_RESTRICTED=1) sessions opening the cross-session messaging socket
Probably attachment-paths-are-displayed-in-a-normalized-form-in-error, safe-mode-is-now-evaluated-lazily-and-uses-the-same-check --allow-dangerously-skip-permissions on respawn without the bypass-permissions disclaimer having been accepted
Probably background-sessions-ignore-allow-dangerously-skip-permissi, new-proactivity-startup-level-gated-by-tengu-proactivity, new-relaunchproactivity-env-vars-added-to-a-scrub-list-no claude auth login or with a credentials file already in the config directory
Probably cloud-session-auth-error-reworded-to-point-at-claude-auth-l /login reporting success when the keychain refused the new login and kept an old one it could not remove
Probably onboarding-sign-in-skip-tracked-login-prompt-kept, trusted-device-error-text-reworded, cloud-session-auth-error-reworded-to-point-at-claude-auth-l, chrome-menu-reports-live-connection-state-and-reconnect-act --include-partial-messages seeing a reply stay open after the turn ended when its stream was cut, interrupted or fell back to non-streaming
No entry names this ConfigChange hooks and reloading settings mid-command when .claude/settings.json or .claude/settings.local.json does not exist
Probably background-worker-bypass-permission-mode-requires-consent-in, background-sessions-ignore-allow-dangerously-skip-permissi, cloud-session-settings-review-now-covers-user-settings-with /loop and other recurring session-only scheduled tasks running an extra time after a sandboxed Bash command on Linux or after .claude/scheduled_tasks.json was deleted
No entry names this plugin-authoring skill: Claude now gives the one command another person runs to install a mod you made, and writes it in a README's install section
No entry names this /plugin in the desktop app's Code tab: it now says where to install and manage plugins there
Probably plugin-marketplace-names-that-imitate-anthropics-are-refuse, plugin-in-desktop-app-returns-guidance-text claude auth login and /login and no longer blames API-key authentication
Probably onboarding-sign-in-skip-tracked-login-prompt-kept, trusted-device-error-text-reworded, cloud-session-auth-error-reworded-to-point-at-claude-auth-l, chrome-menu-reports-live-connection-state-and-reconnect-act /ultrareview upload: it is about half as long and says what kind of file is the problem
No entry names this /ultrareview upload refuses a checkout: each known cause now has its own message, with a way to fix it
No entry names this browser_batch call now gets 90 seconds, up from 60, before it is reported as timed out
Probably browser-batch-gets-its-own-tool-call-timeout CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC to also skip the startup connection warm-up
No entry names this --chrome, /chrome or your user settings
Probably claude-in-chrome-enabled-by-default-is-ignored-in-remote-a, project-settings-can-no-longer-turn-on-claude-in-chrome-a-w, chrome-disabled-session-notice-tells-the-model-not-to-use-br, chrome-menu-reports-live-connection-state-and-reconnect-act pyright, which is no longer treated as a read-only command
Nothing to match on $.process.spawn rejects with when another mod denies it after the child ran: it now says the call ran and a plugin withheld its result
No entry names this ! shell command that contains raw control characters other than tab and newline, with a message that shows where they are
Nothing to match on /artifacts: opening an artifact in your browser now closes the list
No entry names this ps command ask for approval instead of running without asking
Nothing to match on /ultrareview upload fails at a git step: they name the step and what to try, and no longer repeat git's own error text
No entry names this /code-review at medium effort to also report cleanup and CLAUDE.md conventions findings on models without tuned review settings, including Opus 5.5 and Sonnet 5.5
Probably medium-and-high-effort-now-map-to-a-different-measured-profi agent_id in Agent results to its agent ID (its name@team address stays in teammate_id); TeammateIdle hooks no longer fire from its subagents or forks
Probably teammate-agent-id-now-uses-resumable-id, teammate-ambiguity-message-wording, agent-spawn-result-reports-agent-id, taskstop-description-now-covers-background-agents-spawned-wi /loop): they are now left running through updates and low memory, where being restarted or shut down could silently lose the wakeup
No entry names this /model, /effort and /rename sent from claude agents to a busy background session to apply right away, without a confirmation, instead of when the turn ends
Probably effort-commands-log-from-level-and-route, medium-and-high-effort-now-map-to-a-different-measured-profi, bridge-remote-control-peers-can-run-effort-model-rena CLAUDE_CODE_WEB_SEARCH_REFILLS_PER_HOUR sets the rate, 0 turns it off) instead of ending after 200 calls
Probably websearch-gets-a-refilling-token-bucket-budget CLAUDE_CODE_DISABLE_ATTACHMENTS so a repository's .claude/settings.json or .claude/settings.local.json can no longer set it; shell, user and managed settings still can
Probably background-worker-bypass-permission-mode-requires-consent-in, background-sessions-ignore-allow-dangerously-skip-permissi, cloud-session-settings-review-now-covers-user-settings-with, new-proactivity-startup-level-gated-by-tengu-proactivity, proactivity-opt-in-restored-across-worker-epochs, new-relaunchproactivity-env-vars-added-to-a-scrub-list-no claude plugin update on a plugin loaded from a directory to print just its reason, without the "Failed to update plugin" prefix, as for built-in plugins
No entry names this gh api in cloud sessions: a host other than github.com set in GH_HOST or GH_REPO is now refused (use --hostname or a full URL), and stderr notes requests to other hosts
Probably gh-api-hostname-help-text-varies, new-gh-host-gh-repo-guidance-when-the-host-differs, gh-style-endpoint-repo-resolution-tightened, gh-tool-description-gh-host-and-gh-repo-host-text claude-api skill's Managed Agents examples to turn off the web tools unless the agent needs them and to use the auto permission policy
Probably bundled-claude-api-skill-adds-managed-agents-quickstarts claude --environment <id> to create its session through the current Sessions API; printed and JSON session ids keep their session_… form
No entry names this !fast to switch a thread to fast mode, moving it to Opus if needed, and !fast off to switch back; replies show (fast) while it's on
No entry names this A model matched these bullets to the GitHub issues they fix, so a link can be wrong.
1 of 27 tool descriptions changed. 1 of 25 tool schemas changed. The appended system-reminder blocks moved: 2 lines added, 1 line removed.
Claude Code, interactive mode
15 prompt changes in this release could not be quoted from the build, so no entry on this page describes them.
334 documentation changes were recorded within 24 hours either side of this release, nearest first. The closest 12 are below. They're here because they happened near this release in time. That's not a claim that this release caused the edit, or that the page documents anything in it.
The 70 literal strings found in the bundle, with the number of entries that name each one. Picking one searches for it. A name is here because this build's code mentions it, which is not the same as it working or being finished.
What's wrong with this entry?