Source Intelligence
Follow Discord

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

← All releases Home ←olderv2.1.68 v2.1.70newer→

Claude Code v2.1.69

31 entries read diff v2.1.68 → v2.1.69 Markdown

This release is a major update that introduces Remote Control Server mode for persistent multi-session hosting, an Ultraplan feature for remote session planning, a completely overhauled memory system with typed categories, live plugin reloading via /reload-plugins, the --prefill CLI flag, native mouse-based text selection and copying, scrollable UI views, auto theme detection, MCPB bundle tooling, and significant improvements to ToolSearch, command safety, and the hooks system.

Find
Pick an entry · j / k steps through
26 entries

Changesopen

Remote Control Server Mode#

What

A new server subcommand for claude remote-control that runs as a persistent server accepting multiple concurrent sessions from claude.ai/code.

Usage
# Start a server where all sessions share the current directory
claude remote-control server

# Each session gets an isolated git worktree (max N sessions)
claude remote-control server --spawn-worktree-sessions [<N>]

# All sessions share the current directory (default behavior, max N sessions)
claude remote-control server --spawn-same-dir-sessions [<N>]

# Additional server options
claude remote-control server --idle-timeout <ms>    # Timeout for detached sessions
claude remote-control server --max-sessions <n>     # Max concurrent sessions
claude remote-control server --workspace <dir>      # Default working directory
claude remote-control server --auth-token <token>   # Bearer token for auth
claude remote-control server --host <string>        # Bind address
claude remote-control server --port <number>        # Listen port
Details
  • Runs as a persistent daemon that accepts connections without pre-creating an initial session
  • Supports cc+unix:// protocol for Unix domain socket connections
  • Session management with server-sessions.json persistence and server.lock for single-instance enforcement
  • Configurable spawn modes: shared directory (default) or isolated git worktrees
  • Sessions can be created, attached, detached, and destroyed independently
  • Policy-controllable via allow_remote_control organization policy
Evidence

Remote Control server mode (search for "Start a Claude Code session server", "--spawn-worktree-sessions", "server-sessions.json")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Ultraplan#

What

A new remote-session-based planning mode that generates comprehensive implementation plans using a dedicated remote session.

Details
  • Triggered when planning complex implementations
  • Runs a remote session that produces a structured <ultraplan> block
  • Displays "Ultraplanning…" progress indicator during generation
  • The generated plan is written to a plan file for user review
  • After review, the user can approve and proceed to implementation
Evidence

Ultraplan feature (search for "Ultraplanning…", "Ultraplan complete", "ultraplan")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

--prefill CLI Flag#

What

Pre-fill the prompt input with text without automatically submitting it, allowing the user to review and edit before sending.

Usage
claude --prefill "implement a function that"
Details
  • Text appears in the input field ready for editing
  • Unlike --print, does not auto-submit — the user can modify the prompt before pressing Enter
  • Useful for templates, repetitive prompts, or workflow scripting
Evidence

Prefill flag (search for "--prefill", "Pre-fill the prompt input with text without submitting it")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

/reload-plugins Command#

What

Reload and activate plugin changes in the current session without needing to restart Claude Code.

Usage
/reload-plugins
Details
  • Activates pending plugin installations, removals, and configuration changes
  • Replaces the previous behavior of requiring a full restart after plugin changes
  • Messages like "Run /reload-plugins to activate" now appear after plugin changes instead of "Restart Claude Code"
Evidence

Plugin reload (search for "/reload-plugins", "Activate pending plugin changes in the current session")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Mouse-Based Text Selection and Copy#

What

Native mouse text selection with drag-to-select and clipboard copy support in the terminal UI.

Details
  • Click and drag to select text in the terminal output
  • Selected text can be copied to the system clipboard via OSC 52 escape sequence
  • noSelect regions prevent selection of UI chrome (status bars, borders) while allowing content selection
  • Selection state tracks anchor/focus positions with full drag handling
  • from-left-edge option extends no-select regions from the left edge of the terminal
Evidence

Text selection system (search for "isDragging", "copySelection", "noSelect")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Scrollable UI Views#

What

Terminal UI components can now use overflowY: "scroll" to create scrollable content areas with sticky-scroll behavior.

Details
  • New scroll overflow mode alongside existing hidden and visible
  • stickyScroll attribute auto-scrolls to bottom as new content arrives
  • Tracks scrollHeight, scrollViewportHeight, and scrollTop for programmatic scroll control
  • Content outside the visible viewport is culled for rendering performance
Evidence

Scroll support (search for "stickyScroll", "scrollViewportHeight", overflowY)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Auto Theme Detection#

What

Claude Code can now automatically detect the system light/dark theme setting and follow it.

Details
  • New "Auto (follow system)" theme option
  • On macOS, reads AppleInterfaceStyle from system defaults to detect Dark Mode
  • Falls back to "dark" on non-macOS platforms
  • Theme updates when toggling between light and dark mode
Evidence

Auto theme (search for "Auto (follow system)", "AppleInterfaceStyle")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Memory System Overhaul#

What

Complete restructuring of the persistent memory system with typed categories, frontmatter format, and two-step save process.

Details
  • Four memory types: user (preferences/role), feedback (corrections/guidance), project (ongoing work context), and reference (external system pointers)
  • Memories use markdown files with YAML frontmatter (name, description, type fields)
  • Two-step save: write the memory file, then add a pointer in the memory index
  • Clear guidance on what NOT to save (code patterns, git history, ephemeral details)
  • When users correct information stated from memory, the stored entry must be updated
  • CLAUDE_COWORK_MEMORY_PATH_OVERRIDE environment variable to customize the memory directory
Evidence

Memory system (search for "## Types of memory", "## What NOT to save in memory", "CLAUDE_COWORK_MEMORY_PATH_OVERRIDE")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

MCPB Bundle Tooling#

What

Infrastructure for creating, validating, packing, and unpacking MCP Bundle (MCPB) files — a packaging format for MCP server plugins.

Usage
mcpb pack       # Create a .mcpb bundle from a directory
mcpb validate   # Validate a manifest.json
Details
  • Interactive manifest.json creation wizard with prompts for name, version, tools, prompts, screenshots, and more
  • .mcpbignore support to exclude files from bundles
  • Schema validation for manifest files
  • Bundle unpacking and verification
  • Supports Node.js and Python version constraints
  • Platform compatibility settings (macOS, Linux, Windows)
Evidence

MCPB support (search for "A MCPB bundle", "mcpb pack", ".mcpbignore")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

git-subdir Plugin Source#

What

Plugins can now be sourced from a subdirectory within a larger repository (monorepo support).

Details
  • Uses sparse checkout (--filter=tree:0) to minimize bandwidth for monorepo plugins
  • Only the specified subdirectory is materialized; the rest of the repo is not downloaded
  • Requires git version 2.25 or later for cone mode support
  • Falls back to unshallow fetch if sparse-checkout fails
Evidence

git-subdir support (search for "git-subdir", "Subdirectory within the repo containing the plugin")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Binary Content Handling#

What

Improved detection and handling of binary file content, with disk-based storage for large binary results.

Details
  • Sophisticated binary detection using byte analysis (null bytes, control characters)
  • Extensive list of known binary file extensions (images, audio, video, archives, executables, etc.)
  • Binary content from tool results can be saved to disk and referenced by path
  • Support for Office documents, archives, and other binary MIME types
Evidence

Binary content (search for "Binary content (", "Binary content could not be saved to disk")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

New Hook Events#

What

Three new hook event types for more granular automation control.

Details
  • InstructionsLoaded: Fires when an instruction file (CLAUDE.md or rule) is loaded
  • TaskCompleted: Can prevent continuation after task completion ("TaskCompleted hook prevented continuation")
  • TeammateIdle: Can prevent continuation when a teammate goes idle ("TeammateIdle hook prevented continuation")
  • Hooks support agent_type and agent_id context fields for subagent-aware automation
  • Background hooks with async mode that wake the model on exit code 2
Evidence

New hooks (search for "InstructionsLoaded", "TaskCompleted hook prevented continuation", "TeammateIdle hook prevented continuation")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Settings Visibility and Debug#

What

New settings API that returns effective merged settings plus raw per-source settings in merge order.

Details
  • Shows settings from all sources (User, Project, Local, Managed) with merge ordering
  • Labeled "Ordered low-to-high priority — later entries override earlier ones"
  • Makes it clear where each setting value comes from
Evidence

Settings debug (search for "Effective merged settings plus raw per-source settings in merge order")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

ToolSearch Multi-Select and Search Hints#

ToolSearch now supports loading multiple deferred tools at once with comma-separated select: syntax (e.g., "select:Read,Edit,Grep"). Tools can also display search hints alongside their names for better discoverability. A new partial select mode handles cases where some tools in a multi-select query are found while others are missing.

Evidence

ToolSearch improvements (search for "select:Read,Edit,Grep", "searchHint", "partial select")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Deferred Tools Announced via System Reminders#

Deferred tools are now announced through system-reminder messages in the conversation rather than a static list, controlled by the tengu_glacier_2xr feature flag. This makes tool availability more dynamic and context-aware.

Evidence

Deferred tools announcement (search for "Deferred tools are announced via system-reminder messages")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Improved Command Shell Safety Checks#

New obfuscation detection for potentially dangerous shell commands:

  • Quoted newlines followed by #-prefixed lines that could hide arguments
  • Consecutive quote characters at word start (potential obfuscation)
  • Empty quote pairs adjacent to quoted dashes (flag obfuscation)
  • Quoted brace characters inside brace contexts (brace expansion obfuscation)
  • Excess closing braces after quote stripping
Evidence

Safety checks (search for "Command contains a quoted newline followed by a #-prefixed line")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

enableWeakerNetworkIsolation Sandbox Setting#

New sandbox option for macOS that allows access to com.apple.trustd.agent within the sandbox. This is needed for Go-based CLI tools (gh, gcloud, terraform, etc.) to verify TLS certificates when using httpProxyPort with a MITM proxy and custom CA.

Evidence

Network isolation setting (search for "enableWeakerNetworkIsolation", "Reduces security")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Thinking Summaries in Transcript View#

New showThinkingSummaries setting to display thinking summaries when viewing the transcript (ctrl+o). Default: false.

Evidence

Thinking summaries (search for "Show thinking summaries in the transcript view")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

pluginTrustMessage for Enterprise#

Enterprise administrators can now configure a custom message appended to the plugin trust warning shown before installation, via pluginTrustMessage in managed settings or MDM policy. Useful for adding organization-specific guidance.

Evidence

Plugin trust message (search for "pluginTrustMessage", "Custom message to append to the plugin trust warning")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

includeCoworkInstructions Setting#

New setting to control whether built-in commit and PR workflow instructions are included in Claude's system prompt. Default: true.

Evidence

Cowork instructions setting (search for "Include built-in commit and PR workflow instructions")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

pathPattern for Marketplace Restrictions#

Marketplace sources of type strictKnownMarketplaces can now use pathPattern — a regex matched against .path of file and directory sources — to allow filesystem-based marketplaces alongside network source restrictions.

Evidence

Path pattern restrictions (search for "pathPattern", "Regex pattern matched against the .path field")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Improved Subagent System Prompt#

Subagent prompts have been refined:

  • Clearer instruction formatting with separate lines for avoiding emojis and colons before tool calls
  • Different response expectations based on agent role: detailed writeup for some, concise reports for others
  • Agents now use absolute paths in their final responses
  • Verification agent type added for post-task validation
  • Teammates explicitly cannot spawn other teammates (flat team roster)
Evidence

Subagent improvements (search for "Teammates cannot spawn other teammates", "NOTE: You just closed out 3+ tasks")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Improved Summary/Compaction System#

The conversation summary system has been significantly revised:

  • New <<ANALYSIS_INSTRUCTION>> placeholder system for injecting analysis instructions
  • Support for partial compaction of only recent messages when earlier context is retained
  • Lean compaction mode (tengu_lean_cast) with shorter analysis prompts focused on coverage over detail
  • Better handling of large conversation files via optimized compact_boundary detection for files over 64MB
Evidence

Summary improvements (search for "<<ANALYSIS_INSTRUCTION>>", "tengu_lean_cast")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Fast Mode Availability Messaging#

Fast mode status messages now distinguish between general unavailability and network-specific issues:

  • "Fast mode is currently unavailable" for general outages
  • "Fast mode unavailable due to network connectivity issues" for network errors
  • Default status changed from "disabled" to more descriptive states
Evidence

Fast mode messaging (search for "Fast mode is currently unavailable", "Fast mode unavailable due to network connectivity issues")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Inquirer.js Prompt Library#

A full Inquirer.js-style prompt library has been embedded for interactive CLI prompts, with proper error handling (AbortPromptError, CancelPromptError, ExitPromptError), pagination, arrow-key navigation, and multi-select support. This replaces ad-hoc prompt implementations.

Evidence

Inquirer integration (search for "AbortPromptError", "[Inquirer] Hook functions can only be called from within a prompt")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Managed Plugin Trust Messaging#

Plugins managed by organization policy now display "Managed by your organization — contact your admin" instead of generic status messages.

Evidence

Managed plugins (search for "Managed by your organization")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

1 entry

Bug Fixesopen

#

  • Fixed sandbox cleanup: cleanupAfterCommand is now exposed on the sandbox API for proper resource cleanup after command execution (search for "cleanupAfterCommand")
  • Fixed lock file handling: Force-removal of stale lock files now logs the action and handles failures gracefully (search for "Force-removed lock file at", "Failed to force-remove lock file")
  • Fixed symlink resolution in path operations: New Fsq function properly resolves symlinks by walking up the directory tree and handling broken symlinks gracefully (search for "realpathSync" in added functions)
  • Removed microcompact feature: The Context microcompacted flow has been removed, simplifying the compaction pipeline (search for "microcompact" — present in old, absent in new context references)
  • Removed coral_reef_opus2 client data check and crystal_beam budget token feature — these were internal feature gates that have been cleaned up
1 entry

Notesopen

#

The Remote Control server subcommand is a significant architectural addition. Previous versions only supported basic remote-control connections — the new server mode enables persistent multi-session hosting with session lifecycle management, making it suitable for team environments and CI/CD integrations. Both --spawn-same-dir-sessions (default) and --spawn-worktree-sessions require the server subcommand; attempting to use them with the base remote-control command will produce an error.

The memory system overhaul changes the recommended format for memory files. While existing plain-text memories will likely continue to work, adopting the new frontmatter format (name, description, type fields) is recommended for better relevance matching in future conversations.

Verbatim
Official · Anthropic

Anthropic’s official release notes

Published verbatim by Anthropic for v2.1.69. 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.

  • Added the /claude-api skill for building applications with the Claude API and Anthropic SDK
  • Added Ctrl+U on an empty bash prompt (!) to exit bash mode, matching escape and backspace
  • Added numeric keypad support for selecting options in Claude's interview questions (previously only the number row above QWERTY worked)
  • Added optional name argument to /remote-control and claude remote-control (/remote-control My Project or --name "My Project") to set a custom session title visible in claude.ai/code
  • Added Voice STT support for 10 new languages (20 total) — Russian, Polish, Turkish, Dutch, Ukrainian, Greek, Czech, Danish, Swedish, Norwegian
  • Added effort level display (e.g., "with low effort") to the logo and spinner, making it easier to see which effort setting is active
  • Added agent name display in terminal title when using claude --agent
  • Added sandbox.enableWeakerNetworkIsolation setting (macOS only) to allow Go programs like gh, gcloud, and terraform to verify TLS certificates when using a custom MITM proxy with httpProxyPort
  • Added includeGitInstructions setting (and CLAUDE_CODE_DISABLE_GIT_INSTRUCTIONS env var) to remove built-in commit and PR workflow instructions from Claude's system prompt
  • Added /reload-plugins command to activate pending plugin changes without restarting
  • Added a one-time startup prompt suggesting Claude Code Desktop on macOS and Windows (max 3 showings, dismissible)
  • Added ${CLAUDE_SKILL_DIR} variable for skills to reference their own directory in SKILL.md content
  • Added InstructionsLoaded hook event that fires when CLAUDE.md or .claude/rules/*.md files are loaded into context
  • Added agent_id (for subagents) and agent_type (for subagents and --agent) to hook events
  • Added worktree field to status line hook commands with name, path, branch, and original repo directory when running in a --worktree session
  • Added pluginTrustMessage in managed settings to append organization-specific context to the plugin trust warning shown before installation
  • Added policy limit fetching (e.g., remote control restrictions) for Team plan OAuth users, not just Enterprise
  • Added pathPattern to strictKnownMarketplaces for regex-matching file/directory marketplace sources alongside hostPattern restrictions
  • Added plugin source type git-subdir to point to a subdirectory within a git repo
  • Added oauth.authServerMetadataUrl config option for MCP servers to specify a custom OAuth metadata discovery URL when standard discovery fails
  • Fixed a security issue where nested skill discovery could load skills from gitignored directories like node_modules
  • Fixed trust dialog silently enabling all .mcp.json servers on first run. You'll now see the per-server approval dialog as expected
  • Fixed claude remote-control crashing immediately on npm installs with "bad option: --sdk-url" (anthropics/claude-code#28334)
  • Fixed --model claude-opus-4-0 and --model claude-opus-4-1 resolving to deprecated Opus versions instead of current
  • Fixed macOS keychain corruption when using multiple OAuth MCP servers. Large OAuth metadata blobs could overflow the security -i stdin buffer, silently leaving stale credentials behind and causing repeated /login prompts.
  • Fixed .credentials.json losing subscriptionType (showing "Claude API" instead of "Claude Pro"/"Claude Max") when the profile endpoint transiently fails during token refresh (anthropics/claude-code#30185)
  • Fixed ghost dotfiles (.bashrc, HEAD, etc.) appearing as untracked files in the working directory after sandboxed Bash commands on Linux
  • Fixed Shift+Enter printing [27;2;13~ instead of inserting a newline in Ghostty over SSH
  • Fixed stash (Ctrl+S) being cleared when submitting a message while Claude is working
  • Fixed ctrl+o (transcript toggle) freezing for many seconds in long sessions with lots of file edits
  • Fixed plan mode feedback input not supporting multi-line text entry (backslash+Enter and Shift+Enter now insert newlines)
  • Fixed cursor not moving down into blank lines at the top of the input box
  • Fixed /stats crash when transcript files contain entries with missing or malformed timestamps
  • Fixed a brief hang after a streaming error on long sessions (the transcript was being fully rewritten to drop one line; it is now truncated in place)
  • Fixed --setting-sources user not blocking dynamically discovered project skills
  • Fixed duplicate CLAUDE.md, slash commands, agents, and rules when running from a worktree nested inside its main repo (e.g. claude -w)
  • Fixed plugin Stop/SessionEnd/etc hooks not firing after any /plugin operation
  • Fixed plugin hooks being silently dropped when two plugins use the same ${CLAUDE_PLUGIN_ROOT}/... command template
  • Fixed memory leak in long-running SDK/CCR sessions where conversation messages were retained unnecessarily
  • Fixed API 400 errors in forked agents (autocompact, summarization) when resuming sessions that were interrupted mid-tool-batch
  • Fixed "unexpected tool_use_id found in tool_result blocks" error when resuming conversations that start with an orphaned tool result
  • Fixed teammates accidentally spawning nested teammates via the Agent tool's name parameter
  • Fixed CLAUDE_CODE_MAX_OUTPUT_TOKENS being ignored during conversation compaction
  • Fixed /compact summary rendering as a user bubble in SDK consumers (Claude Code Remote web UI, VSCode extension)
  • Fixed voice space bar getting stuck after a failed voice activation (module loading race, cold GrowthBook)
  • Fixed worktree file copy on Windows
  • Fixed global .claude folder detection on Windows
  • Fixed symlink bypass where writing new files through a symlinked parent directory could escape the working directory in acceptEdits mode
  • Fixed sandbox prompting users to approve non-allowed domains when allowManagedDomainsOnly is enabled in managed settings — non-allowed domains are now blocked automatically with no bypass
  • Fixed interactive tools (e.g., AskUserQuestion) being silently auto-allowed when listed in a skill's allowed-tools, bypassing the permission prompt and running with empty answers
  • Fixed multi-GB memory spike when committing with large untracked binary files in the working tree
  • Fixed Escape not interrupting a running turn when the input box has draft text. Use Up arrow to pull queued messages back for editing, or Ctrl+U to clear the input line.
  • Fixed Android app crash when running local slash commands (/voice, /cost) in Remote Control sessions
  • Fixed a memory leak where old message array versions accumulated in React Compiler memoCache over long sessions
  • Fixed a memory leak where REPL render scopes accumulated over long sessions (~35MB over 1000 turns)
  • Fixed memory retention in in-process teammates where the parent's full conversation history was pinned for the teammate's lifetime, preventing GC after /clear or auto-compact
  • Fixed a memory leak in interactive mode where hook events could accumulate unboundedly during long sessions
  • Fixed hang when --mcp-config points to a corrupted file
  • Fixed slow startup when many skills/plugins are installed
  • Fixed cd <outside-dir> && <cmd> permission prompt to surface the chained command instead of only showing "Yes, allow reading from <dir>/"
  • Fixed conditional .claude/rules/*.md files (with paths: frontmatter) and nested CLAUDE.md files not loading in print mode (claude -p)
  • Fixed /clear not fully clearing all session caches, reducing memory retention in long sessions
  • Fixed terminal flicker caused by animated elements at the scrollback boundary
  • Fixed UI frame drops on macOS when using MCP servers with OAuth (regression from 2.1.x)
  • Fixed occasional frame stalls during typing caused by synchronous debug log flushes
  • Fixed TeammateIdle and TaskCompleted hooks to support {"continue": false, "stopReason": "..."} to stop the teammate, matching Stop hook behavior
  • Fixed WorktreeCreate and WorktreeRemove plugin hooks being silently ignored
  • Fixed skill descriptions with colons (e.g., "Triggers include: X, Y, Z") failing to load from SKILL.md frontmatter
  • Fixed project skills without a description: frontmatter field not appearing in Claude's available skills list
  • Fixed /context showing identical token counts for all MCP tools from a server
  • Fixed literal nul file creation on Windows when the model uses CMD-style 2>nul redirection in Git Bash
  • Fixed extra blank lines appearing below each tool call in the expanded subagent transcript view (Ctrl+O)
  • Fixed Tab/arrow keys not cycling Settings tabs when /config search box is focused but empty
  • Fixed service key OAuth sessions (CCR containers) spamming [ERROR] logs with 403s from profile-scoped endpoints
  • Fixed inconsistent color for "Remote Control active" status indicator
  • Fixed Voice waveform cursor covering the first suffix letter when dictating mid-input
  • Fixed Voice input showing all 5 spaces during warmup instead of capping at ~2 (aligning with the "keep holding…" hint)
  • Improved spinner performance by isolating the 50ms animation loop from the surrounding shell, reducing render and CPU overhead during turns
  • Improved UI rendering performance in native binaries with React Compiler
  • Improved --worktree startup by eliminating a git subprocess on the startup path
  • Improved macOS startup by eliminating redundant settings-file reloads when managed settings resolve
  • Improved macOS startup for Claude.ai enterprise/team users by skipping an unnecessary keychain lookup
  • Improved MCP -p startup by pipelining claude.ai config fetch with local connections and using a concurrency pool instead of sequential batching
  • Improved voice startup by removing imperceptible warmup pulse animations that were causing re-render stutter
  • Improved MCP binary content handling: tools returning PDFs, Office documents, or audio now save decoded bytes to disk with the correct file extension instead of dumping raw base64 into the conversation context. WebFetch also saves binary responses alongside its summary.
  • Improved memory usage in long sessions by stabilizing onSubmit across message updates
  • Improved LSP tool rendering and memory context building to no longer read entire files
  • Improved session upload and memory sync to avoid reading large files into memory before size/binary checks
  • Improved file operation performance by avoiding reading file contents for existence checks (6 sites)
  • Improved documentation to clarify that --append-system-prompt-file and --system-prompt-file work in interactive mode (the docs previously said print mode only)
  • Reduced baseline memory by ~16MB by deferring Yoga WASM preloading
  • Reduced memory footprint for SDK and CCR sessions using stream-json output
  • Reduced memory usage when resuming large sessions (including compacted history)
  • Reduced token usage on multi-agent tasks with more concise subagent final reports
  • Changed Sonnet 4.5 users on Pro/Max/Team Premium to be automatically migrated to Sonnet 4.6
  • Changed the /resume picker to show your most recent prompt instead of the first one. This also resolves some titles appearing as (session).
  • Changed claude.ai MCP connector failures to show a notification instead of silently disappearing from the tool list
  • Changed example command suggestions to be generated deterministically instead of calling Haiku
  • Changed resuming after compaction to no longer produce a preamble recap before continuing
  • [SDK] Changed task creation to no longer require the activeForm field — the spinner falls back to the task subject
  • [VSCode] Added compaction display as a collapsible "Compacted chat" card with the summary inside
  • [VSCode] The permission mode picker now respects permissions.disableBypassPermissionsMode from your effective Claude Code settings (including managed/policy settings) — when set to disable, bypass permissions mode is hidden from the picker
  • [VSCode] Fixed RTL text (Arabic, Hebrew, Persian) rendering reversed in the chat panel (regression in v2.1.63)
Read it on GitHub →
System prompt

The system prompt was not captured for this release, so this page cannot say whether it moved.

Which releases were captured →