refcard.dev

Claude Code Changelog

Auto-synced weekly from Anthropic's official release notes ↗

Subscribe via RSS →

v2.1.118

  • — Vim visual mode (v) and visual-line mode (V) with selection, operators, and visual feedback
  • — /usage replaces /cost and /stats — both old commands remain as shortcuts that open the right tab
  • — /theme — create and switch named custom themes; plugins can also ship themes via a themes/ directory
  • — Hooks can now invoke MCP tools directly via type:mcp_tool
  • — DISABLE_UPDATES env var — blocks all update paths including manual claude update (stricter than DISABLE_AUTOUPDATER)
  • — WSL inherits Windows-side managed settings via the wslInheritsWindowsSettings policy key
  • — Auto mode — include "$defaults" in autoMode.allow/soft_deny/environment to add custom rules alongside built-ins

v2.1.117

  • — Forked subagents now enabled on external builds via CLAUDE_CODE_FORK_SUBAGENT=1
  • — /model selections persist across restarts even when a project pins a different model
  • — Default effort for Pro/Max subscribers on Opus 4.6 and Sonnet 4.6 is now high (was medium)
  • — Native builds on macOS/Linux replace Glob/Grep tools with embedded bfs and ugrep via Bash — faster searches
  • — /resume now offers to summarize stale large sessions before re-reading them

v2.1.116

  • — /resume on large sessions is up to 67% faster (40MB+ sessions)
  • — Thinking spinner shows inline progress (still thinking, thinking more, almost done thinking)
  • — /config search now matches option values (e.g. searching "vim" finds Editor mode setting)
  • — /doctor can now be opened while Claude is responding
  • — Bash tool surfaces a hint when gh commands hit GitHub API rate limits

v2.1.113

  • — CLI now spawns a native Claude Code binary instead of bundled JavaScript
  • — sandbox.network.deniedDomains setting — block specific domains even when a wildcard allowedDomains would permit them
  • — /loop improved — Esc cancels pending wakeups; wakeup label now says "Claude resuming /loop wakeup"
  • — /ultrareview improved — faster launch with parallelized checks, diffstat in launch dialog
  • — Bash tool security — deny rules now match commands wrapped in env/sudo/watch/ionice/setsid wrappers
  • — Bash(find:*) allow rules no longer auto-approve find -exec/-delete

v2.1.111

  • — Claude Opus 4.7 xhigh effort level — between high and max, available via /effort and --effort
  • — Auto mode now available for Max subscribers on Opus 4.7
  • — /effort opens an interactive slider with arrow-key navigation when called without arguments
  • — Added "Auto (match terminal)" theme option matching dark/light terminal mode
  • — /less-permission-prompts — scans transcripts and proposes a prioritized allowlist for settings.json
  • — /ultrareview added — comprehensive cloud code review via parallel multi-agent analysis
  • — Ctrl+U now clears the entire input buffer; Ctrl+Y restores it
  • — Plan files now named after your prompt (e.g. fix-auth-race-snug-otter.md) instead of random words

v2.1.110

  • — /tui command and tui setting — switch to flicker-free rendering in the same conversation
  • — Push notification tool — Claude can send mobile push notifications via Remote Control
  • — /focus command — toggle focus view (Ctrl+O now only toggles verbose transcript)
  • — autoScrollEnabled config — disable auto-scroll in fullscreen mode
  • — --resume/--continue now resurrects unexpired scheduled tasks
  • — Session recap enabled for Bedrock, Vertex, Foundry, and DISABLE_TELEMETRY users

v2.1.108

  • — ENABLE_PROMPT_CACHING_1H — opt into 1-hour prompt cache TTL on API key, Bedrock, Vertex, Foundry
  • — FORCE_PROMPT_CACHING_5M — force 5-minute TTL
  • — /recap command — provides context when returning to a session; configurable in /config
  • — CLAUDE_CODE_ENABLE_AWAY_SUMMARY — force recap when telemetry is disabled
  • — /undo is now an alias for /rewind
  • — Model can now discover and invoke built-in slash commands like /init, /review via the Skill tool

v2.1.105

  • — EnterWorktree tool now accepts a path parameter to switch into an existing worktree
  • — PreCompact hook support — hooks can block compaction by exiting with code 2 or returning decision:block
  • — Background monitor support for plugins via a top-level monitors manifest key
  • — /proactive is now an alias for /loop
  • — Stalled API streams now abort after 5 minutes of no data and retry non-streaming
  • — /doctor improved with status icons; press f to have Claude fix reported issues

v2.1.101

  • — /team-onboarding command — generates a teammate ramp-up guide from your Claude Code usage
  • — OS CA certificate store trusted by default — enterprise TLS proxies work without extra setup (CLAUDE_CODE_CERT_STORE=bundled to use only bundled CAs)
  • — Monitor tool added — stream events from background scripts
  • — --exclude-dynamic-system-prompt-sections flag for improved cross-user prompt caching
  • — workspace.git_worktree in status line JSON — set whenever inside a linked git worktree
  • — refreshInterval status line setting — re-run the status line command every N seconds

v2.1.94

  • — Default effort changed from medium to high for API-key, Bedrock/Vertex/Foundry, Team, and Enterprise users
  • — Amazon Bedrock powered by Mantle — set CLAUDE_CODE_USE_MANTLE=1
  • — /cost now shows per-model and cache-hit breakdown for subscription users
  • — /release-notes is now an interactive version picker
  • — Bedrock interactive setup wizard accessible from the login screen
  • — forceRemoteSettingsRefresh policy — block startup until remote managed settings are freshly fetched (fail-closed)

v2.1.91

  • — MCP tool result persistence override via _meta["anthropic/maxResultSizeChars"] — up to 500K chars
  • — disableSkillShellExecution setting — disable inline shell execution in skills and plugin commands
  • — Plugins can now ship executables under bin/ and invoke them as bare commands from Bash tool
  • — /vim command removed (toggle vim mode via /config → Editor mode instead)

v2.1.90

  • — /powerup — interactive lessons teaching Claude Code features with animated demos

v2.1.89

  • — "defer" permission decision for PreToolUse hooks — headless sessions can pause at a tool call and resume
  • — CLAUDE_CODE_NO_FLICKER=1 — opt into flicker-free alt-screen rendering with virtualized scrollback
  • — PermissionDenied hook — fires after auto mode classifier denials; return {retry:true} to let the model retry
  • — Named subagents added to @ mention typeahead suggestions
  • — Auto mode — denied commands appear in /permissions → Recent tab where you can retry with r

v2.1.85

  • — Conditional if field for hooks using permission rule syntax — filter when hooks run to reduce overhead
  • — /proactive (now /loop) fires add timestamp markers in transcripts
  • — MCP OAuth now follows RFC 9728 Protected Resource Metadata discovery
  • — PreToolUse hooks can satisfy AskUserQuestion by returning updatedInput + permissionDecision allow

v2.1.84

  • — PowerShell tool for Windows as an opt-in preview
  • — ANTHROPIC_DEFAULT_OPUS/SONNET/HAIKU_MODEL_SUPPORTS env vars — override effort/thinking capability detection for 3p providers
  • — TaskCreated hook fires when a task is created via TaskCreate
  • — WorktreeCreate hook supports type:http — return the created worktree path via hookSpecificOutput.worktreePath
  • — allowedChannelPlugins managed setting for team/enterprise admins
  • — Rules and skills paths: frontmatter now accepts a YAML list of globs

v2.1.83

  • — managed-settings.d/ drop-in directory — separate teams deploy independent policy fragments, merged alphabetically
  • — CwdChanged and FileChanged hook events for reactive environment management (e.g. direnv)
  • — sandbox.failIfUnavailable setting — exit with error when sandbox cannot start instead of running unsandboxed
  • — Transcript search — press / in transcript mode (Ctrl+O) to search, n/N to step through matches
  • — Agents can now declare initialPrompt in frontmatter to auto-submit a first turn
  • — Changed "stop all background agents" keybinding from Ctrl+F to Ctrl+X Ctrl+K

v2.1.81

  • — --bare flag for scripted -p calls — skips hooks, LSP, plugin sync, and skill walks; requires API key auth
  • — --channels permission relay — channel servers can forward tool approval prompts to your phone

v2.1.80

  • — rate_limits field in statusline scripts — shows Claude.ai rate limit usage (5-hour and 7-day windows)
  • — effort frontmatter support for skills and slash commands to override effort level when invoked
  • — source:'settings' plugin marketplace source — declare plugin entries inline in settings.json

v2.1.79

  • — --console flag to claude auth login for Anthropic Console (API billing) authentication

v2.1.78

  • — StopFailure hook event — fires when a turn ends due to an API error
  • — "${CLAUDE_PLUGIN_DATA}" variable for plugin persistent state that survives plugin updates
  • — effort, maxTurns, and disallowedTools frontmatter support for plugin-shipped agents
  • — Response text now streams line-by-line as it is generated

v2.1.77

  • — Increased default max output tokens for Claude Opus 4.6 to 64K; upper bound for Opus/Sonnet 4.6 raised to 128K
  • — allowRead sandbox filesystem setting — re-allow read access within denyRead regions
  • — /copy now accepts an optional index — /copy N copies the Nth-latest assistant response

v1.7.0

  • — Added /ultrareview for deep multi-agent PR review (local diff or GitHub PR#)
  • — Added /fast to toggle fast mode on Opus 4.6 (faster output, no model downgrade)
  • — EnterWorktree and ExitWorktree tools for isolated agent tasks in git worktrees
  • — Background agent support via run_in_background parameter on the Agent tool

v1.6.0

  • — Extended thinking support with think, think hard, think harder, ultrathink keywords
  • — --effort flag for headless mode thinking budget control (low/medium/high)
  • — MCP HTTP transport support in addition to stdio and SSE

v1.5.0

  • — Plan mode via Shift+Tab — Claude presents a plan before executing any action
  • — /memory command to open and edit user-level CLAUDE.md
  • — Skills system for custom agent configurations in .claude/skills/

v1.0.0

  • — Initial stable release of Claude Code
  • — MCP server support with stdio and SSE transports
  • — CLAUDE.md memory file system with import syntax
  • — Project and user-level settings.json