spacehub

surshar / claude-code

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

Built by Anthropic · github.com/anthropics

claude-code / CHANGELOG.md
GitHub Actionschore: Update CHANGELOG.md 53a5f3e · on Mar 7, 2026 History
2,132 lines · 132.8 KBPreviewCodeRaw
1# Changelog
2
3## 2.1.71
4
5- Added `/loop` command to run a prompt or slash command on a recurring interval (e.g. `/loop 5m check the deploy`)
6- Added cron scheduling tools for recurring prompts within a session
7- Added `voice:pushToTalk` keybinding to make the voice activation key rebindable in `keybindings.json` (default: space) — modifier+letter combos like `meta+k` have zero typing interference
8- Added `fmt`, `comm`, `cmp`, `numfmt`, `expr`, `test`, `printf`, `getconf`, `seq`, `tsort`, and `pr` to the bash auto-approval allowlist
9- Fixed stdin freeze in long-running sessions where keystrokes stop being processed but the process stays alive
10- Fixed a 5–8 second startup freeze for users with voice mode enabled, caused by CoreAudio initialization blocking the main thread after system wake
11- Fixed startup UI freeze when many claude.ai proxy connectors refresh an expired OAuth token simultaneously
12- Fixed forked conversations (`/fork`) sharing the same plan file, which caused plan edits in one fork to overwrite the other
13- Fixed the Read tool putting oversized images into context when image processing failed, breaking subsequent turns in long image-heavy sessions
14- Fixed false-positive permission prompts for compound bash commands containing heredoc commit messages
15- Fixed plugin installations being lost when running multiple Claude Code instances
16- Fixed claude.ai connectors failing to reconnect after OAuth token refresh
17- Fixed claude.ai MCP connector startup notifications appearing for every org-configured connector instead of only previously connected ones
18- Fixed background agent completion notifications missing the output file path, which made it difficult for parent agents to recover agent results after context compaction
19- Fixed duplicate output in Bash tool error messages when commands exit with non-zero status
20- Fixed Chrome extension auto-detection getting permanently stuck on "not installed" after running on a machine without local Chrome
21- Fixed `/plugin marketplace update` failing with merge conflicts when the marketplace is pinned to a branch/tag ref
22- Fixed `/plugin marketplace add owner/repo@ref` incorrectly parsing `@` — previously only `#` worked as a ref separator, causing undiagnosable errors with `strictKnownMarketplaces`
23- Fixed duplicate entries in `/permissions` Workspace tab when the same directory is added with and without a trailing slash
24- Fixed `--print` hanging forever when team agents are configured — the exit loop no longer waits on long-lived `in_process_teammate` tasks
25- Fixed "❯ Tool loaded." appearing in the REPL after every `ToolSearch` call
26- Fixed prompting for `cd <cwd> && git ...` on Windows when the model uses a mingw-style path
27- Improved startup time by deferring native image processor loading to first use
28- Improved bridge session reconnection to complete within seconds after laptop wake from sleep, instead of waiting up to 10 minutes
29- Improved `/plugin uninstall` to disable project-scoped plugins in `.claude/settings.local.json` instead of modifying `.claude/settings.json`, so changes don't affect teammates
30- Improved plugin-provided MCP server deduplication — servers that duplicate a manually-configured server (same command/URL) are now skipped, preventing duplicate connections and tool sets. Suppressions are shown in the `/plugin` menu.
31- Updated `/debug` to toggle debug logging on mid-session, since debug logs are no longer written by default
32- Removed startup notification noise for unauthenticated org-registered claude.ai connectors
33
34## 2.1.70
35
36- Fixed API 400 errors when using `ANTHROPIC_BASE_URL` with a third-party gateway — tool search now correctly detects proxy endpoints and disables `tool_reference` blocks
37- Fixed `API Error: 400 This model does not support the effort parameter` when using custom Bedrock inference profiles or other model identifiers not matching standard Claude naming patterns
38- Fixed empty model responses immediately after `ToolSearch` — the server renders tool schemas with system-prompt-style tags at the prompt tail, which could confuse models into stopping early
39- Fixed prompt-cache bust when an MCP server with `instructions` connects after the first turn
40- Fixed Enter inserting a newline instead of submitting when typing over a slow SSH connection
41- Fixed clipboard corrupting non-ASCII text (CJK, emoji) on Windows/WSL by using PowerShell `Set-Clipboard`
42- Fixed extra VS Code windows opening at startup on Windows when running from the VS Code integrated terminal
43- Fixed voice mode failing on Windows native binary with "native audio module could not be loaded"
44- Fixed push-to-talk not activating on session start when `voiceEnabled: true` was set in settings
45- Fixed markdown links containing `#NNN` references incorrectly pointing to the current repository instead of the linked URL
46- Fixed repeated "Model updated to Opus 4.6" notification when a project's `.claude/settings.json` has a legacy Opus model string pinned
47- Fixed plugins showing as inaccurately installed in `/plugin`
48- Fixed plugins showing "not found in marketplace" errors on fresh startup by auto-refreshing after marketplace installation
49- Fixed `/security-review` command failing with `unknown option merge-base` on older git versions
50- Fixed `/color` command having no way to reset back to the default color — `/color default`, `/color gray`, `/color reset`, and `/color none` now restore the default
51- Fixed a performance regression in the `AskUserQuestion` preview dialog that re-ran markdown rendering on every keystroke in the notes input
52- Fixed feature flags read during early startup never refreshing their disk cache, causing stale values to persist across sessions
53- Fixed `permissions.defaultMode` settings values other than `acceptEdits` or `plan` being applied in Claude Code Remote environments — they are now ignored
54- Fixed skill listing being re-injected on every `--resume` (~600 tokens saved per resume)
55- Fixed teleport marker not rendering in VS Code teleported sessions
56- Improved error message when microphone captures silence to distinguish from "no speech detected"
57- Improved compaction to preserve images in the summarizer request, allowing prompt cache reuse for faster and cheaper compaction
58- Improved `/rename` to work while Claude is processing, instead of being silently queued
59- Reduced prompt input re-renders during turns by ~74%
60- Reduced startup memory by ~426KB for users without custom CA certificates
61- Reduced Remote Control `/poll` rate to once per 10 minutes while connected (was 1–2s), cutting server load ~300×. Reconnection is unaffected — transport loss immediately wakes fast polling.
62- [VSCode] Added spark icon in VS Code activity bar that lists all Claude Code sessions, with sessions opening as full editors
63- [VSCode] Added full markdown document view for plans in VS Code, with support for adding comments to provide feedback
64- [VSCode] Added native MCP server management dialog — use `/mcp` in the chat panel to enable/disable servers, reconnect, and manage OAuth authentication without switching to the terminal
65
66## 2.1.69
67
68- Added the `/claude-api` skill for building applications with the Claude API and Anthropic SDK
69- Added Ctrl+U on an empty bash prompt (`!`) to exit bash mode, matching `escape` and `backspace`
70- Added numeric keypad support for selecting options in Claude's interview questions (previously only the number row above QWERTY worked)
71- 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
72- Added Voice STT support for 10 new languages (20 total) — Russian, Polish, Turkish, Dutch, Ukrainian, Greek, Czech, Danish, Swedish, Norwegian
73- Added effort level display (e.g., "with low effort") to the logo and spinner, making it easier to see which effort setting is active
74- Added agent name display in terminal title when using `claude --agent`
75- 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`
76- 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
77- Added `/reload-plugins` command to activate pending plugin changes without restarting
78- Added a one-time startup prompt suggesting Claude Code Desktop on macOS and Windows (max 3 showings, dismissible)
79- Added `${CLAUDE_SKILL_DIR}` variable for skills to reference their own directory in SKILL.md content
80- Added `InstructionsLoaded` hook event that fires when CLAUDE.md or `.claude/rules/*.md` files are loaded into context
81- Added `agent_id` (for subagents) and `agent_type` (for subagents and `--agent`) to hook events
82- Added `worktree` field to status line hook commands with name, path, branch, and original repo directory when running in a `--worktree` session
83- Added `pluginTrustMessage` in managed settings to append organization-specific context to the plugin trust warning shown before installation
84- Added policy limit fetching (e.g., remote control restrictions) for Team plan OAuth users, not just Enterprise
85- Added `pathPattern` to `strictKnownMarketplaces` for regex-matching file/directory marketplace sources alongside `hostPattern` restrictions
86- Added plugin source type `git-subdir` to point to a subdirectory within a git repo
87- Added `oauth.authServerMetadataUrl` config option for MCP servers to specify a custom OAuth metadata discovery URL when standard discovery fails
88- Fixed a security issue where nested skill discovery could load skills from gitignored directories like `node_modules`
89- Fixed trust dialog silently enabling all `.mcp.json` servers on first run. You'll now see the per-server approval dialog as expected
90- Fixed `claude remote-control` crashing immediately on npm installs with "bad option: --sdk-url" (anthropics/claude-code#28334)
91- Fixed `--model claude-opus-4-0` and `--model claude-opus-4-1` resolving to deprecated Opus versions instead of current
92- 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.
93- 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)
94- Fixed ghost dotfiles (`.bashrc`, `HEAD`, etc.) appearing as untracked files in the working directory after sandboxed Bash commands on Linux
95- Fixed Shift+Enter printing `[27;2;13~` instead of inserting a newline in Ghostty over SSH
96- Fixed stash (Ctrl+S) being cleared when submitting a message while Claude is working
97- Fixed ctrl+o (transcript toggle) freezing for many seconds in long sessions with lots of file edits
98- Fixed plan mode feedback input not supporting multi-line text entry (backslash+Enter and Shift+Enter now insert newlines)
99- Fixed cursor not moving down into blank lines at the top of the input box
100- Fixed `/stats` crash when transcript files contain entries with missing or malformed timestamps
101- 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)
102- Fixed `--setting-sources user` not blocking dynamically discovered project skills
103- Fixed duplicate CLAUDE.md, slash commands, agents, and rules when running from a worktree nested inside its main repo (e.g. `claude -w`)
104- Fixed plugin Stop/SessionEnd/etc hooks not firing after any `/plugin` operation
105- Fixed plugin hooks being silently dropped when two plugins use the same `${CLAUDE_PLUGIN_ROOT}/...` command template
106- Fixed memory leak in long-running SDK/CCR sessions where conversation messages were retained unnecessarily
107- Fixed API 400 errors in forked agents (autocompact, summarization) when resuming sessions that were interrupted mid-tool-batch
108- Fixed "unexpected tool_use_id found in tool_result blocks" error when resuming conversations that start with an orphaned tool result
109- Fixed teammates accidentally spawning nested teammates via the Agent tool's `name` parameter
110- Fixed `CLAUDE_CODE_MAX_OUTPUT_TOKENS` being ignored during conversation compaction
111- Fixed `/compact` summary rendering as a user bubble in SDK consumers (Claude Code Remote web UI, VSCode extension)
112- Fixed voice space bar getting stuck after a failed voice activation (module loading race, cold GrowthBook)
113- Fixed worktree file copy on Windows
114- Fixed global `.claude` folder detection on Windows
115- Fixed symlink bypass where writing new files through a symlinked parent directory could escape the working directory in `acceptEdits` mode
116- 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
117- 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
118- Fixed multi-GB memory spike when committing with large untracked binary files in the working tree
119- 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.
120- Fixed Android app crash when running local slash commands (`/voice`, `/cost`) in Remote Control sessions
121- Fixed a memory leak where old message array versions accumulated in React Compiler `memoCache` over long sessions
122- Fixed a memory leak where REPL render scopes accumulated over long sessions (~35MB over 1000 turns)
123- 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
124- Fixed a memory leak in interactive mode where hook events could accumulate unboundedly during long sessions
125- Fixed hang when `--mcp-config` points to a corrupted file
126- Fixed slow startup when many skills/plugins are installed
127- Fixed `cd <outside-dir> && <cmd>` permission prompt to surface the chained command instead of only showing "Yes, allow reading from <dir>/"
128- Fixed conditional `.claude/rules/*.md` files (with `paths:` frontmatter) and nested CLAUDE.md files not loading in print mode (`claude -p`)
129- Fixed `/clear` not fully clearing all session caches, reducing memory retention in long sessions
130- Fixed terminal flicker caused by animated elements at the scrollback boundary
131- Fixed UI frame drops on macOS when using MCP servers with OAuth (regression from 2.1.x)
132- Fixed occasional frame stalls during typing caused by synchronous debug log flushes
133- Fixed `TeammateIdle` and `TaskCompleted` hooks to support `{"continue": false, "stopReason": "..."}` to stop the teammate, matching `Stop` hook behavior
134- Fixed `WorktreeCreate` and `WorktreeRemove` plugin hooks being silently ignored
135- Fixed skill descriptions with colons (e.g., "Triggers include: X, Y, Z") failing to load from SKILL.md frontmatter
136- Fixed project skills without a `description:` frontmatter field not appearing in Claude's available skills list
137- Fixed `/context` showing identical token counts for all MCP tools from a server
138- Fixed literal `nul` file creation on Windows when the model uses CMD-style `2>nul` redirection in Git Bash
139- Fixed extra blank lines appearing below each tool call in the expanded subagent transcript view (Ctrl+O)
140- Fixed Tab/arrow keys not cycling Settings tabs when `/config` search box is focused but empty
141- Fixed service key OAuth sessions (CCR containers) spamming `[ERROR]` logs with 403s from profile-scoped endpoints
142- Fixed inconsistent color for "Remote Control active" status indicator
143- Fixed Voice waveform cursor covering the first suffix letter when dictating mid-input
144- Fixed Voice input showing all 5 spaces during warmup instead of capping at ~2 (aligning with the "keep holding…" hint)
145- Improved spinner performance by isolating the 50ms animation loop from the surrounding shell, reducing render and CPU overhead during turns
146- Improved UI rendering performance in native binaries with React Compiler
147- Improved `--worktree` startup by eliminating a git subprocess on the startup path
148- Improved macOS startup by eliminating redundant settings-file reloads when managed settings resolve
149- Improved macOS startup for Claude.ai enterprise/team users by skipping an unnecessary keychain lookup
150- Improved MCP `-p` startup by pipelining claude.ai config fetch with local connections and using a concurrency pool instead of sequential batching
151- Improved voice startup by removing imperceptible warmup pulse animations that were causing re-render stutter
152- 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.
153- Improved memory usage in long sessions by stabilizing `onSubmit` across message updates
154- Improved LSP tool rendering and memory context building to no longer read entire files
155- Improved session upload and memory sync to avoid reading large files into memory before size/binary checks
156- Improved file operation performance by avoiding reading file contents for existence checks (6 sites)
157- Improved documentation to clarify that `--append-system-prompt-file` and `--system-prompt-file` work in interactive mode (the docs previously said print mode only)
158- Reduced baseline memory by ~16MB by deferring Yoga WASM preloading
159- Reduced memory footprint for SDK and CCR sessions using stream-json output
160- Reduced memory usage when resuming large sessions (including compacted history)
161- Reduced token usage on multi-agent tasks with more concise subagent final reports
162- Changed Sonnet 4.5 users on Pro/Max/Team Premium to be automatically migrated to Sonnet 4.6
163- Changed the `/resume` picker to show your most recent prompt instead of the first one. This also resolves some titles appearing as `(session)`.
164- Changed claude.ai MCP connector failures to show a notification instead of silently disappearing from the tool list
165- Changed example command suggestions to be generated deterministically instead of calling Haiku
166- Changed resuming after compaction to no longer produce a preamble recap before continuing
167- [SDK] Changed task creation to no longer require the `activeForm` field — the spinner falls back to the task subject
168- [VSCode] Added compaction display as a collapsible "Compacted chat" card with the summary inside
169- [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
170- [VSCode] Fixed RTL text (Arabic, Hebrew, Persian) rendering reversed in the chat panel (regression in v2.1.63)
171
172## 2.1.68
173
174- Opus 4.6 now defaults to medium effort for Max and Team subscribers. Medium effort works well for most tasks — it's the sweet spot between speed and thoroughness. You can change this anytime with `/model`
175- Re-introduced the "ultrathink" keyword to enable high effort for the next turn
176- Removed Opus 4 and 4.1 from Claude Code on the first-party API — users with these models pinned are automatically moved to Opus 4.6
177
178## 2.1.66
179
180- Reduced spurious error logging
181
182## 2.1.63
183
184- Added `/simplify` and `/batch` bundled slash commands
185- Fixed local slash command output like /cost appearing as user-sent messages instead of system messages in the UI
186- Project configs & auto memory now shared across git worktrees of the same repository
187- Added `ENABLE_CLAUDEAI_MCP_SERVERS=false` env var to opt out from making claude.ai MCP servers available
188- Improved `/model` command to show the currently active model in the slash command menu
189- Added HTTP hooks, which can POST JSON to a URL and receive JSON instead of running a shell command
190- Fixed listener leak in bridge polling loop
191- Fixed listener leak in MCP OAuth flow cleanup
192- Added manual URL paste fallback during MCP OAuth authentication. If the automatic localhost redirect doesn't work, you can paste the callback URL to complete authentication.
193- Fixed memory leak when navigating hooks configuration menu
194- Fixed listener leak in interactive permission handler during auto-approvals
195- Fixed file count cache ignoring glob ignore patterns
196- Fixed memory leak in bash command prefix cache
197- Fixed MCP tool/resource cache leak on server reconnect
198- Fixed IDE host IP detection cache incorrectly sharing results across ports
199- Fixed WebSocket listener leak on transport reconnect
200- Fixed memory leak in git root detection cache that could cause unbounded growth in long-running sessions
201- Fixed memory leak in JSON parsing cache that grew unbounded over long sessions
202- VSCode: Fixed remote sessions not appearing in conversation history
203- Fixed a race condition in the REPL bridge where new messages could arrive at the server interleaved with historical messages during the initial connection flush, causing message ordering issues.
204- Fixed memory leak where long-running teammates retained all messages in AppState even after conversation compaction
205- Fixed a memory leak where MCP server fetch caches were not cleared on disconnect, causing growing memory usage with servers that reconnect frequently
206- Improved memory usage in long sessions with subagents by stripping heavy progress message payloads during context compaction
207- Added "Always copy full response" option to the `/copy` picker. When selected, future `/copy` commands will skip the code block picker and copy the full response directly.
208- VSCode: Added session rename and remove actions to the sessions list
209- Fixed `/clear` not resetting cached skills, which could cause stale skill content to persist in the new conversation
210
211## 2.1.62
212
213- Fixed prompt suggestion cache regression that reduced cache hit rates
214
215## 2.1.61
216
217- Fixed concurrent writes corrupting config file on Windows
218
219## 2.1.59
220
221- Claude automatically saves useful context to auto-memory. Manage with /memory
222- Added `/copy` command to show an interactive picker when code blocks are present, allowing selection of individual code blocks or the full response.
223- Improved "always allow" prefix suggestions for compound bash commands (e.g. `cd /tmp && git fetch && git push`) to compute smarter per-subcommand prefixes instead of treating the whole command as one
224- Improved ordering of short task lists
225- Improved memory usage in multi-agent sessions by releasing completed subagent task state
226- Fixed MCP OAuth token refresh race condition when running multiple Claude Code instances simultaneously
227- Fixed shell commands not showing a clear error message when the working directory has been deleted
228- Fixed config file corruption that could wipe authentication when multiple Claude Code instances ran simultaneously
229
230## 2.1.58
231
232- Expand Remote Control to more users
233
234## 2.1.56
235
236- VS Code: Fixed another cause of "command 'claude-vscode.editor.openLast' not found" crashes
237
238## 2.1.55
239
240- Fixed BashTool failing on Windows with EINVAL error
241
242## 2.1.53
243
244- Fixed a UI flicker where user input would briefly disappear after submission before the message rendered
245- Fixed bulk agent kill (ctrl+f) to send a single aggregate notification instead of one per agent, and to properly clear the command queue
246- Fixed graceful shutdown sometimes leaving stale sessions when using Remote Control by parallelizing teardown network calls
247- Fixed `--worktree` sometimes being ignored on first launch
248- Fixed a panic ("switch on corrupted value") on Windows
249- Fixed a crash that could occur when spawning many processes on Windows
250- Fixed a crash in the WebAssembly interpreter on Linux x64 & Windows x64
251- Fixed a crash that sometimes occurred after 2 minutes on Windows ARM64
252
253## 2.1.52
254
255- VS Code: Fixed extension crash on Windows ("command 'claude-vscode.editor.openLast' not found")
256
257## 2.1.51
258
259- Added `claude remote-control` subcommand for external builds, enabling local environment serving for all users.
260- Updated plugin marketplace default git timeout from 30s to 120s and added `CLAUDE_CODE_PLUGIN_GIT_TIMEOUT_MS` to configure.
261- Added support for custom npm registries and specific version pinning when installing plugins from npm sources
262- BashTool now skips login shell (`-l` flag) by default when a shell snapshot is available, improving command execution performance. Previously this required setting `CLAUDE_BASH_NO_LOGIN=true`.
263- Fixed a security issue where `statusLine` and `fileSuggestion` hook commands could execute without workspace trust acceptance in interactive mode.
264- Tool results larger than 50K characters are now persisted to disk (previously 100K). This reduces context window usage and improves conversation longevity.
265- Fixed a bug where duplicate `control_response` messages (e.g. from WebSocket reconnects) could cause API 400 errors by pushing duplicate assistant messages into the conversation.
266- Added `CLAUDE_CODE_ACCOUNT_UUID`, `CLAUDE_CODE_USER_EMAIL`, and `CLAUDE_CODE_ORGANIZATION_UUID` environment variables for SDK callers to provide account info synchronously, eliminating a race condition where early telemetry events lacked account metadata.
267- Fixed slash command autocomplete crashing when a plugin's SKILL.md description is a YAML array or other non-string type
268- The `/model` picker now shows human-readable labels (e.g., "Sonnet 4.5") instead of raw model IDs for pinned model versions, with an upgrade hint when a newer version is available.
269- Managed settings can now be set via macOS plist or Windows Registry. Learn more at https://code.claude.com/docs/en/settings#settings-files
270
271## 2.1.50
272
273- Added support for `startupTimeout` configuration for LSP servers
274- Added `WorktreeCreate` and `WorktreeRemove` hook events, enabling custom VCS setup and teardown when agent worktree isolation creates or removes worktrees.
275- Fixed a bug where resumed sessions could be invisible when the working directory involved symlinks, because the session storage path was resolved at different times during startup. Also fixed session data loss on SSH disconnect by flushing session data before hooks and analytics in the graceful shutdown sequence.
276- Linux: Fixed native modules not loading on systems with glibc older than 2.30 (e.g., RHEL 8)
277- Fixed memory leak in agent teams where completed teammate tasks were never garbage collected from session state
278- Fixed `CLAUDE_CODE_SIMPLE` to fully strip down skills, session memory, custom agents, and CLAUDE.md token counting
279- Fixed `/mcp reconnect` freezing the CLI when given a server name that doesn't exist
280- Fixed memory leak where completed task state objects were never removed from AppState
281- Added support for `isolation: worktree` in agent definitions, allowing agents to declaratively run in isolated git worktrees.
282- `CLAUDE_CODE_SIMPLE` mode now also disables MCP tools, attachments, hooks, and CLAUDE.md file loading for a fully minimal experience.
283- Fixed bug where MCP tools were not discovered when tool search is enabled and a prompt is passed in as a launch argument
284- Improved memory usage during long sessions by clearing internal caches after compaction
285- Added `claude agents` CLI command to list all configured agents
286- Improved memory usage during long sessions by clearing large tool results after they have been processed
287- Fixed a memory leak where LSP diagnostic data was never cleaned up after delivery, causing unbounded memory growth in long sessions
288- Fixed a memory leak where completed task output was not freed from memory, reducing memory usage in long sessions with many tasks
289- Improved startup performance for headless mode (`-p` flag) by deferring Yoga WASM and UI component imports
290- Fixed prompt suggestion cache regression that reduced cache hit rates
291- Fixed unbounded memory growth in long sessions by capping file history snapshots
292- Added `CLAUDE_CODE_DISABLE_1M_CONTEXT` environment variable to disable 1M context window support
293- Opus 4.6 (fast mode) now includes the full 1M context window
294- VSCode: Added `/extra-usage` command support in VS Code sessions
295- Fixed memory leak where TaskOutput retained recent lines after cleanup
296- Fixed memory leak in CircularBuffer where cleared items were retained in the backing array
297- Fixed memory leak in shell command execution where ChildProcess and AbortController references were retained after cleanup
298
299## 2.1.49
300
301- Improved MCP OAuth authentication with step-up auth support and discovery caching, reducing redundant network requests during server connections
302- Added `--worktree` (`-w`) flag to start Claude in an isolated git worktree
303- Subagents support `isolation: "worktree"` for working in a temporary git worktree
304- Added Ctrl+F keybinding to kill background agents (two-press confirmation)
305- Agent definitions support `background: true` to always run as a background task
306- Plugins can ship `settings.json` for default configuration
307- Fixed file-not-found errors to suggest corrected paths when the model drops the repo folder
308- Fixed Ctrl+C and ESC being silently ignored when background agents are running and the main thread is idle. Pressing twice within 3 seconds now kills all background agents.
309- Fixed prompt suggestion cache regression that reduced cache hit rates.
310- Fixed `plugin enable` and `plugin disable` to auto-detect the correct scope when `--scope` is not specified, instead of always defaulting to user scope
311- Simple mode (`CLAUDE_CODE_SIMPLE`) now includes the file edit tool in addition to the Bash tool, allowing direct file editing in simple mode.
312- Permission suggestions are now populated when safety checks trigger an ask response, enabling SDK consumers to display permission options
313- Sonnet 4.5 with 1M context is being removed from the Max plan in favor of our frontier Sonnet 4.6 model, which now has 1M context. Please switch in /model.
314- Fixed verbose mode not updating thinking block display when toggled via `/config` — memo comparators now correctly detect verbose changes
315- Fixed unbounded WASM memory growth during long sessions by periodically resetting the tree-sitter parser
316- Fixed potential rendering issues caused by stale yoga layout references
317- Improved performance in non-interactive mode (`-p`) by skipping unnecessary API calls during startup
318- Improved performance by caching authentication failures for HTTP and SSE MCP servers, avoiding repeated connection attempts to servers requiring auth
319- Fixed unbounded memory growth during long-running sessions caused by Yoga WASM linear memory never shrinking
320- SDK model info now includes `supportsEffort`, `supportedEffortLevels`, and `supportsAdaptiveThinking` fields so consumers can discover model capabilities.
321- Added `ConfigChange` hook event that fires when configuration files change during a session, enabling enterprise security auditing and optional blocking of settings changes.
322- Improved startup performance by caching MCP auth failures to avoid redundant connection attempts
323- Improved startup performance by reducing HTTP calls for analytics token counting
324- Improved startup performance by batching MCP tool token counting into a single API call
325- Fixed `disableAllHooks` setting to respect managed settings hierarchy — non-managed settings can no longer disable managed hooks set by policy (#26637)
326- Fixed `--resume` session picker showing raw XML tags for sessions that start with commands like `/clear`. Now correctly falls through to the session ID fallback.
327- Improved permission prompts for path safety and working directory blocks to show the reason for the restriction instead of a bare prompt with no context
328
329## 2.1.47
330
331- Fixed FileWriteTool line counting to preserve intentional trailing blank lines instead of stripping them with `trimEnd()`.
332- Fixed Windows terminal rendering bugs caused by `os.EOL` (`\r\n`) in display code — line counts now show correct values instead of always showing 1 on Windows.
333- Improved VS Code plan preview: auto-updates as Claude iterates, enables commenting only when the plan is ready for review, and keeps the preview open when rejecting so Claude can revise.
334- Fixed a bug where bold and colored text in markdown output could shift to the wrong characters on Windows due to `\r\n` line endings.
335- Fixed compaction failing when conversation contains many PDF documents by stripping document blocks alongside images before sending to the compaction API (anthropics/claude-code#26188)
336- Improved memory usage in long-running sessions by releasing API stream buffers, agent context, and skill state after use
337- Improved startup performance by deferring SessionStart hook execution, reducing time-to-interactive by ~500ms.
338- Fixed an issue where bash tool output was silently discarded on Windows when using MSYS2 or Cygwin shells.
339- Improved performance of `@` file mentions - file suggestions now appear faster by pre-warming the index on startup and using session-based caching with background refresh.
340- Improved memory usage by trimming agent task message history after tasks complete
341- Improved memory usage during long agent sessions by eliminating O(n²) message accumulation in progress updates
342- Fixed the bash permission classifier to validate that returned match descriptions correspond to actual input rules, preventing hallucinated descriptions from incorrectly granting permissions
343- Fixed user-defined agents only loading one file on NFS/FUSE filesystems that report zero inodes (anthropics/claude-code#26044)
344- Fixed plugin agent skills silently failing to load when referenced by bare name instead of fully-qualified plugin name (anthropics/claude-code#25834)
345- Search patterns in collapsed tool results are now displayed in quotes for clarity
346- Windows: Fixed CWD tracking temp files never being cleaned up, causing them to accumulate indefinitely (anthropics/claude-code#17600)
347- Use `ctrl+f` to kill all background agents instead of double-pressing ESC. Background agents now continue running when you press ESC to cancel the main thread, giving you more control over agent lifecycle.
348- Fixed API 400 errors ("thinking blocks cannot be modified") that occurred in sessions with concurrent agents, caused by interleaved streaming content blocks preventing proper message merging.
349- Simplified teammate navigation to use only Shift+Down (with wrapping) instead of both Shift+Up and Shift+Down.
350- Fixed an issue where a single file write/edit error would abort all other parallel file write/edit operations. Independent file mutations now complete even when a sibling fails.
351- Added `last_assistant_message` field to Stop and SubagentStop hook inputs, providing the final assistant response text so hooks can access it without parsing transcript files.
352- Fixed custom session titles set via `/rename` being lost after resuming a conversation (anthropics/claude-code#23610)
353- Fixed collapsed read/search hint text overflowing on narrow terminals by truncating from the start.
354- Fixed an issue where bash commands with backslash-newline continuation lines (e.g., long commands split across multiple lines with `\`) would produce spurious empty arguments, potentially breaking command execution.
355- Fixed built-in slash commands (`/help`, `/model`, `/compact`, etc.) being hidden from the autocomplete dropdown when many user skills are installed (anthropics/claude-code#22020)
356- Fixed MCP servers not appearing in the MCP Management Dialog after deferred loading
357- Fixed session name persisting in status bar after `/clear` command (anthropics/claude-code#26082)
358- Fixed crash when a skill's `name` or `description` in SKILL.md frontmatter is a bare number (e.g., `name: 3000`) — the value is now properly coerced to a string (anthropics/claude-code#25837)
359- Fixed /resume silently dropping sessions when the first message exceeds 16KB or uses array-format content (anthropics/claude-code#25721)
360- Added `chat:newline` keybinding action for configurable multi-line input (anthropics/claude-code#26075)
361- Added `added_dirs` to the statusline JSON `workspace` section, exposing directories added via `/add-dir` to external scripts (anthropics/claude-code#26096)
362- Fixed `claude doctor` misclassifying mise and asdf-managed installations as native installs (anthropics/claude-code#26033)
363- Fixed zsh heredoc failing with "read-only file system" error in sandboxed commands (anthropics/claude-code#25990)
364- Fixed agent progress indicator showing inflated tool use count (anthropics/claude-code#26023)
365- Fixed image pasting not working on WSL2 systems where Windows copies images as BMP format (anthropics/claude-code#25935)
366- Fixed background agent results returning raw transcript data instead of the agent's final answer (anthropics/claude-code#26012)
367- Fixed Warp terminal incorrectly prompting for Shift+Enter setup when it supports it natively (anthropics/claude-code#25957)
368- Fixed CJK wide characters causing misaligned timestamps and layout elements in the TUI (anthropics/claude-code#26084)
369- Fixed custom agent `model` field in `.claude/agents/*.md` being ignored when spawning team teammates (anthropics/claude-code#26064)
370- Fixed plan mode being lost after context compaction, causing the model to switch from planning to implementation mode (anthropics/claude-code#26061)
371- Fixed `alwaysThinkingEnabled: true` in settings.json not enabling thinking mode on Bedrock and Vertex providers (anthropics/claude-code#26074)
372- Fixed `tool_decision` OTel telemetry event not being emitted in headless/SDK mode (anthropics/claude-code#26059)
373- Fixed session name being lost after context compaction — renamed sessions now preserve their custom title through compaction (anthropics/claude-code#26121)
374- Increased initial session count in resume picker from 10 to 50 for faster session discovery (anthropics/claude-code#26123)
375- Windows: fixed worktree session matching when drive letter casing differs (anthropics/claude-code#26123)
376- Fixed `/resume <session-id>` failing to find sessions whose first message exceeds 16KB (anthropics/claude-code#25920)
377- Fixed "Always allow" on multiline bash commands creating invalid permission patterns that corrupt settings (anthropics/claude-code#25909)
378- Fixed React crash (error #31) when a skill's `argument-hint` in SKILL.md frontmatter uses YAML sequence syntax (e.g., `[topic: foo | bar]`) — the value is now properly coerced to a string (anthropics/claude-code#25826)
379- Fixed crash when using `/fork` on sessions that used web search — null entries in search results from transcript deserialization are now handled gracefully (anthropics/claude-code#25811)
380- Fixed read-only git commands triggering FSEvents file watcher loops on macOS by adding --no-optional-locks flag (anthropics/claude-code#25750)
381- Fixed custom agents and skills not being discovered when running from a git worktree — project-level `.claude/agents/` and `.claude/skills/` from the main repository are now included (anthropics/claude-code#25816)
382- Fixed non-interactive subcommands like `claude doctor` and `claude plugin validate` being blocked inside nested Claude sessions (anthropics/claude-code#25803)
383- Windows: Fixed the same CLAUDE.md file being loaded twice when drive letter casing differs between paths (anthropics/claude-code#25756)
384- Fixed inline code spans in markdown being incorrectly parsed as bash commands (anthropics/claude-code#25792)
385- Fixed teammate spinners not respecting custom spinnerVerbs from settings (anthropics/claude-code#25748)
386- Fixed shell commands permanently failing after a command deletes its own working directory (anthropics/claude-code#26136)
387- Fixed hooks (PreToolUse, PostToolUse) silently failing to execute on Windows by using Git Bash instead of cmd.exe (anthropics/claude-code#25981)
388- Fixed LSP `findReferences` and other location-based operations returning results from gitignored files (e.g., `node_modules/`, `venv/`) (anthropics/claude-code#26051)
389- Moved config backup files from home directory root to `~/.claude/backups/` to reduce home directory clutter (anthropics/claude-code#26130)
390- Fixed sessions with large first prompts (>16KB) disappearing from the /resume list (anthropics/claude-code#26140)
391- Fixed shell functions with double-underscore prefixes (e.g., `__git_ps1`) not being preserved across shell sessions (anthropics/claude-code#25824)
392- Fixed spinner showing "0 tokens" counter before any tokens have been received (anthropics/claude-code#26105)
393- VSCode: Fixed conversation messages appearing dimmed while the AskUserQuestion dialog is open (anthropics/claude-code#26078)
394- Fixed background tasks failing in git worktrees due to remote URL resolution reading from worktree-specific gitdir instead of the main repository config (anthropics/claude-code#26065)
395- Fixed Right Alt key leaving visible `[25~` escape sequence residue in the input field on Windows/Git Bash terminals (anthropics/claude-code#25943)
396- The `/rename` command now updates the terminal tab title by default (anthropics/claude-code#25789)
397- Fixed Edit tool silently corrupting Unicode curly quotes (\u201c\u201d \u2018\u2019) by replacing them with straight quotes when making edits (anthropics/claude-code#26141)
398- Fixed OSC 8 hyperlinks only being clickable on the first line when link text wraps across multiple terminal lines.
399
400## 2.1.46
401
402- Fixed orphaned CC processes after terminal disconnect on macOS
403- Added support for using claude.ai MCP connectors in Claude Code
404
405## 2.1.45
406
407- Added support for Claude Sonnet 4.6
408- Added support for reading `enabledPlugins` and `extraKnownMarketplaces` from `--add-dir` directories
409- Added `spinnerTipsOverride` setting to customize spinner tips — configure `tips` with an array of custom tip strings, and optionally set `excludeDefault: true` to show only your custom tips instead of the built-in ones
410- Added `SDKRateLimitInfo` and `SDKRateLimitEvent` types to the SDK, enabling consumers to receive rate limit status updates including utilization, reset times, and overage information
411- Fixed Agent Teams teammates failing on Bedrock, Vertex, and Foundry by propagating API provider environment variables to tmux-spawned processes (anthropics/claude-code#23561)
412- Fixed sandbox "operation not permitted" errors when writing temporary files on macOS by using the correct per-user temp directory (anthropics/claude-code#21654)
413- Fixed Task tool (backgrounded agents) crashing with a `ReferenceError` on completion (anthropics/claude-code#22087)
414- Fixed autocomplete suggestions not being accepted on Enter when images are pasted in the input
415- Fixed skills invoked by subagents incorrectly appearing in main session context after compaction
416- Fixed excessive `.claude.json.backup` files accumulating on every startup
417- Fixed plugin-provided commands, agents, and hooks not being available immediately after installation without requiring a restart
418- Improved startup performance by removing eager loading of session history for stats caching
419- Improved memory usage for shell commands that produce large output — RSS no longer grows unboundedly with command output size
420- Improved collapsed read/search groups to show the current file or search pattern being processed beneath the summary line while active
421- [VSCode] Improved permission destination choice (project/user/session) to persist across sessions
422
423## 2.1.44
424
425- Fixed ENAMETOOLONG errors for deeply-nested directory paths
426- Fixed auth refresh errors
427
428## 2.1.43
429
430- Fixed AWS auth refresh hanging indefinitely by adding a 3-minute timeout
431- Fixed spurious warnings for non-agent markdown files in `.claude/agents/` directory
432- Fixed structured-outputs beta header being sent unconditionally on Vertex/Bedrock
433
434## 2.1.42
435
436- Improved startup performance by deferring Zod schema construction
437- Improved prompt cache hit rates by moving date out of system prompt
438- Added one-time Opus 4.6 effort callout for eligible users
439- Fixed /resume showing interrupt messages as session titles
440- Fixed image dimension limit errors to suggest /compact
441
442## 2.1.41
443
444- Added guard against launching Claude Code inside another Claude Code session
445- Fixed Agent Teams using wrong model identifier for Bedrock, Vertex, and Foundry customers
446- Fixed a crash when MCP tools return image content during streaming
447- Fixed /resume session previews showing raw XML tags instead of readable command names
448- Improved model error messages for Bedrock/Vertex/Foundry users with fallback suggestions
449- Fixed plugin browse showing misleading "Space to Toggle" hint for already-installed plugins
450- Fixed hook blocking errors (exit code 2) not showing stderr to the user
451- Added `speed` attribute to OTel events and trace spans for fast mode visibility
452- Added `claude auth login`, `claude auth status`, and `claude auth logout` CLI subcommands
453- Added Windows ARM64 (win32-arm64) native binary support
454- Improved `/rename` to auto-generate session name from conversation context when called without arguments
455- Improved narrow terminal layout for prompt footer
456- Fixed file resolution failing for @-mentions with anchor fragments (e.g., `@README.md#installation`)
457- Fixed FileReadTool blocking the process on FIFOs, `/dev/stdin`, and large files
458- Fixed background task notifications not being delivered in streaming Agent SDK mode
459- Fixed cursor jumping to end on each keystroke in classifier rule input
460- Fixed markdown link display text being dropped for raw URL
461- Fixed auto-compact failure error notifications being shown to users
462- Fixed permission wait time being included in subagent elapsed time display
463- Fixed proactive ticks firing while in plan mode
464- Fixed clear stale permission rules when settings change on disk
465- Fixed hook blocking errors showing stderr content in UI
466
467## 2.1.39
468
469- Improved terminal rendering performance
470- Fixed fatal errors being swallowed instead of displayed
471- Fixed process hanging after session close
472- Fixed character loss at terminal screen boundary
473- Fixed blank lines in verbose transcript view
474
475## 2.1.38
476
477- Fixed VS Code terminal scroll-to-top regression introduced in 2.1.37
478- Fixed Tab key queueing slash commands instead of autocompleting
479- Fixed bash permission matching for commands using environment variable wrappers
480- Fixed text between tool uses disappearing when not using streaming
481- Fixed duplicate sessions when resuming in VS Code extension
482- Improved heredoc delimiter parsing to prevent command smuggling
483- Blocked writes to `.claude/skills` directory in sandbox mode
484
485## 2.1.37
486
487- Fixed an issue where /fast was not immediately available after enabling /extra-usage
488
489## 2.1.36
490
491- Fast mode is now available for Opus 4.6. Learn more at https://code.claude.com/docs/en/fast-mode
492
493## 2.1.34
494
495- Fixed a crash when agent teams setting changed between renders
496- Fixed a bug where commands excluded from sandboxing (via `sandbox.excludedCommands` or `dangerouslyDisableSandbox`) could bypass the Bash ask permission rule when `autoAllowBashIfSandboxed` was enabled
497
498## 2.1.33
499
500- Fixed agent teammate sessions in tmux to send and receive messages
501- Fixed warnings about agent teams not being available on your current plan
502- Added `TeammateIdle` and `TaskCompleted` hook events for multi-agent workflows
503- Added support for restricting which sub-agents can be spawned via `Task(agent_type)` syntax in agent "tools" frontmatter
504- Added `memory` frontmatter field support for agents, enabling persistent memory with `user`, `project`, or `local` scope
505- Added plugin name to skill descriptions and `/skills` menu for better discoverability
506- Fixed an issue where submitting a new message while the model was in extended thinking would interrupt the thinking phase
507- Fixed an API error that could occur when aborting mid-stream, where whitespace text combined with a thinking block would bypass normalization and produce an invalid request
508- Fixed API proxy compatibility issue where 404 errors on streaming endpoints no longer triggered non-streaming fallback
509- Fixed an issue where proxy settings configured via `settings.json` environment variables were not applied to WebFetch and other HTTP requests on the Node.js build
510- Fixed `/resume` session picker showing raw XML markup instead of clean titles for sessions started with slash commands
511- Improved error messages for API connection failures — now shows specific cause (e.g., ECONNREFUSED, SSL errors) instead of generic "Connection error"
512- Errors from invalid managed settings are now surfaced
513- VSCode: Added support for remote sessions, allowing OAuth users to browse and resume sessions from claude.ai
514- VSCode: Added git branch and message count to the session picker, with support for searching by branch name
515- VSCode: Fixed scroll-to-bottom under-scrolling on initial session load and session switch
516
517## 2.1.32
518
519- Claude Opus 4.6 is now available!
520- Added research preview agent teams feature for multi-agent collaboration (token-intensive feature, requires setting CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1)
521- Claude now automatically records and recalls memories as it works
522- Added "Summarize from here" to the message selector, allowing partial conversation summarization.
523- Skills defined in `.claude/skills/` within additional directories (`--add-dir`) are now loaded automatically.
524- Fixed `@` file completion showing incorrect relative paths when running from a subdirectory
525- Updated --resume to re-use --agent value specified in previous conversation by default.
526- Fixed: Bash tool no longer throws "Bad substitution" errors when heredocs contain JavaScript template literals like `${index + 1}`, which previously interrupted tool execution
527- Skill character budget now scales with context window (2% of context), so users with larger context windows can see more skill descriptions without truncation
528- Fixed Thai/Lao spacing vowels (สระ า, ำ) not rendering correctly in the input field
529- VSCode: Fixed slash commands incorrectly being executed when pressing Enter with preceding text in the input field
530- VSCode: Added spinner when loading past conversations list
531
532## 2.1.31
533
534- Added session resume hint on exit, showing how to continue your conversation later
535- Added support for full-width (zenkaku) space input from Japanese IME in checkbox selection
536- Fixed PDF too large errors permanently locking up sessions, requiring users to start a new conversation
537- Fixed bash commands incorrectly reporting failure with "Read-only file system" errors when sandbox mode was enabled
538- Fixed a crash that made sessions unusable after entering plan mode when project config in `~/.claude.json` was missing default fields
539- Fixed `temperatureOverride` being silently ignored in the streaming API path, causing all streaming requests to use the default temperature (1) regardless of the configured override
540- Fixed LSP shutdown/exit compatibility with strict language servers that reject null params
541- Improved system prompts to more clearly guide the model toward using dedicated tools (Read, Edit, Glob, Grep) instead of bash equivalents (`cat`, `sed`, `grep`, `find`), reducing unnecessary bash command usage
542- Improved PDF and request size error messages to show actual limits (100 pages, 20MB)
543- Reduced layout jitter in the terminal when the spinner appears and disappears during streaming
544- Removed misleading Anthropic API pricing from model selector for third-party provider (Bedrock, Vertex, Foundry) users
545
546## 2.1.30
547
548- Added `pages` parameter to the Read tool for PDFs, allowing specific page ranges to be read (e.g., `pages: "1-5"`). Large PDFs (>10 pages) now return a lightweight reference when `@` mentioned instead of being inlined into context.
549- Added pre-configured OAuth client credentials for MCP servers that don't support Dynamic Client Registration (e.g., Slack). Use `--client-id` and `--client-secret` with `claude mcp add`.
550- Added `/debug` for Claude to help troubleshoot the current session
551- Added support for additional `git log` and `git show` flags in read-only mode (e.g., `--topo-order`, `--cherry-pick`, `--format`, `--raw`)
552- Added token count, tool uses, and duration metrics to Task tool results
553- Added reduced motion mode to the config
554- Fixed phantom "(no content)" text blocks appearing in API conversation history, reducing token waste and potential model confusion
555- Fixed prompt cache not correctly invalidating when tool descriptions or input schemas changed, only when tool names changed
556- Fixed 400 errors that could occur after running `/login` when the conversation contained thinking blocks
557- Fixed a hang when resuming sessions with corrupted transcript files containing `parentUuid` cycles
558- Fixed rate limit message showing incorrect "/upgrade" suggestion for Max 20x users when extra-usage is unavailable
559- Fixed permission dialogs stealing focus while actively typing
560- Fixed subagents not being able to access SDK-provided MCP tools because they were not synced to the shared application state
561- Fixed a regression where Windows users with a `.bashrc` file could not run bash commands
562- Improved memory usage for `--resume` (68% reduction for users with many sessions) by replacing the session index with lightweight stat-based loading and progressive enrichment
563- Improved `TaskStop` tool to display the stopped command/task description in the result line instead of a generic "Task stopped" message
564- Changed `/model` to execute immediately instead of being queued
565- [VSCode] Added multiline input support to the "Other" text input in question dialogs (use Shift+Enter for new lines)
566- [VSCode] Fixed duplicate sessions appearing in the session list when starting a new conversation
567
568## 2.1.29
569
570- Fixed startup performance issues when resuming sessions that have `saved_hook_context`
571
572## 2.1.27
573
574- Added tool call failures and denials to debug logs
575- Fixed context management validation error for gateway users, ensuring `CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1` avoids the error
576- Added `--from-pr` flag to resume sessions linked to a specific GitHub PR number or URL
577- Sessions are now automatically linked to PRs when created via `gh pr create`
578- Fixed /context command not displaying colored output
579- Fixed status bar duplicating background task indicator when PR status was shown
580- Windows: Fixed bash command execution failing for users with `.bashrc` files
581- Windows: Fixed console windows flashing when spawning child processes
582- VSCode: Fixed OAuth token expiration causing 401 errors after extended sessions
583
584## 2.1.25
585
586- Fixed beta header validation error for gateway users on Bedrock and Vertex, ensuring `CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1` avoids the error
587
588## 2.1.23
589
590- Added customizable spinner verbs setting (`spinnerVerbs`)
591- Fixed mTLS and proxy connectivity for users behind corporate proxies or using client certificates
592- Fixed per-user temp directory isolation to prevent permission conflicts on shared systems
593- Fixed a race condition that could cause 400 errors when prompt caching scope was enabled
594- Fixed pending async hooks not being cancelled when headless streaming sessions ended
595- Fixed tab completion not updating the input field when accepting a suggestion
596- Fixed ripgrep search timeouts silently returning empty results instead of reporting errors
597- Improved terminal rendering performance with optimized screen data layout
598- Changed Bash commands to show timeout duration alongside elapsed time
599- Changed merged pull requests to show a purple status indicator in the prompt footer
600- [IDE] Fixed model options displaying incorrect region strings for Bedrock users in headless mode
601
602## 2.1.22
603
604- Fixed structured outputs for non-interactive (-p) mode
605
606## 2.1.21
607
608- Added support for full-width (zenkaku) number input from Japanese IME in option selection prompts
609- Fixed shell completion cache files being truncated on exit
610- Fixed API errors when resuming sessions that were interrupted during tool execution
611- Fixed auto-compact triggering too early on models with large output token limits
612- Fixed task IDs potentially being reused after deletion
613- Fixed file search not working in VS Code extension on Windows
614- Improved read/search progress indicators to show "Reading…" while in progress and "Read" when complete
615- Improved Claude to prefer file operation tools (Read, Edit, Write) over bash equivalents (cat, sed, awk)
616- [VSCode] Added automatic Python virtual environment activation, ensuring `python` and `pip` commands use the correct interpreter (configurable via `claudeCode.usePythonEnvironment` setting)
617- [VSCode] Fixed message action buttons having incorrect background colors
618
619## 2.1.20
620
621- Added arrow key history navigation in vim normal mode when cursor cannot move further
622- Added external editor shortcut (Ctrl+G) to the help menu for better discoverability
623- Added PR review status indicator to the prompt footer, showing the current branch's PR state (approved, changes requested, pending, or draft) as a colored dot with a clickable link
624- Added support for loading `CLAUDE.md` files from additional directories specified via `--add-dir` flag (requires setting `CLAUDE_CODE_ADDITIONAL_DIRECTORIES_CLAUDE_MD=1`)
625- Added ability to delete tasks via the `TaskUpdate` tool
626- Fixed session compaction issues that could cause resume to load full history instead of the compact summary
627- Fixed agents sometimes ignoring user messages sent while actively working on a task
628- Fixed wide character (emoji, CJK) rendering artifacts where trailing columns were not cleared when replaced by narrower characters
629- Fixed JSON parsing errors when MCP tool responses contain special Unicode characters
630- Fixed up/down arrow keys in multi-line and wrapped text input to prioritize cursor movement over history navigation
631- Fixed draft prompt being lost when pressing UP arrow to navigate command history
632- Fixed ghost text flickering when typing slash commands mid-input
633- Fixed marketplace source removal not properly deleting settings
634- Fixed duplicate output in some commands like `/context`
635- Fixed task list sometimes showing outside the main conversation view
636- Fixed syntax highlighting for diffs occurring within multiline constructs like Python docstrings
637- Fixed crashes when cancelling tool use
638- Improved `/sandbox` command UI to show dependency status with installation instructions when dependencies are missing
639- Improved thinking status text with a subtle shimmer animation
640- Improved task list to dynamically adjust visible items based on terminal height
641- Improved fork conversation hint to show how to resume the original session
642- Changed collapsed read/search groups to show present tense ("Reading", "Searching for") while in progress, and past tense ("Read", "Searched for") when complete
643- Changed `ToolSearch` results to appear as a brief notification instead of inline in the conversation
644- Changed the `/commit-push-pr` skill to automatically post PR URLs to Slack channels when configured via MCP tools
645- Changed the `/copy` command to be available to all users
646- Changed background agents to prompt for tool permissions before launching
647- Changed permission rules like `Bash(*)` to be accepted and treated as equivalent to `Bash`
648- Changed config backups to be timestamped and rotated (keeping 5 most recent) to prevent data loss
649
650## 2.1.19
651
652- Added env var `CLAUDE_CODE_ENABLE_TASKS`, set to `false` to keep the old system temporarily
653- Added shorthand `$0`, `$1`, etc. for accessing individual arguments in custom commands
654- Fixed crashes on processors without AVX instruction support
655- Fixed dangling Claude Code processes when terminal is closed by catching EIO errors from `process.exit()` and using SIGKILL as fallback
656- Fixed `/rename` and `/tag` not updating the correct session when resuming from a different directory (e.g., git worktrees)
657- Fixed resuming sessions by custom title not working when run from a different directory
658- Fixed pasted text content being lost when using prompt stash (Ctrl+S) and restore
659- Fixed agent list displaying "Sonnet (default)" instead of "Inherit (default)" for agents without an explicit model setting
660- Fixed backgrounded hook commands not returning early, potentially causing the session to wait on a process that was intentionally backgrounded
661- Fixed file write preview omitting empty lines
662- Changed skills without additional permissions or hooks to be allowed without requiring approval
663- Changed indexed argument syntax from `$ARGUMENTS.0` to `$ARGUMENTS[0]` (bracket syntax)
664- [SDK] Added replay of `queued_command` attachment messages as `SDKUserMessageReplay` events when `replayUserMessages` is enabled
665- [VSCode] Enabled session forking and rewind functionality for all users
666
667## 2.1.18
668
669- Added customizable keyboard shortcuts. Configure keybindings per context, create chord sequences, and personalize your workflow. Run `/keybindings` to get started. Learn more at https://code.claude.com/docs/en/keybindings
670
671## 2.1.17
672
673- Fixed crashes on processors without AVX instruction support
674
675## 2.1.16
676
677- Added new task management system, including new capabilities like dependency tracking
678- [VSCode] Added native plugin management support
679- [VSCode] Added ability for OAuth users to browse and resume remote Claude sessions from the Sessions dialog
680- Fixed out-of-memory crashes when resuming sessions with heavy subagent usage
681- Fixed an issue where the "context remaining" warning was not hidden after running `/compact`
682- Fixed session titles on the resume screen not respecting the user's language setting
683- [IDE] Fixed a race condition on Windows where the Claude Code sidebar view container would not appear on start
684
685## 2.1.15
686
687- Added deprecation notification for npm installations - run `claude install` or see https://docs.anthropic.com/en/docs/claude-code/getting-started for more options
688- Improved UI rendering performance with React Compiler
689- Fixed the "Context left until auto-compact" warning not disappearing after running `/compact`
690- Fixed MCP stdio server timeout not killing child process, which could cause UI freezes
691
692## 2.1.14
693
694- Added history-based autocomplete in bash mode (`!`) - type a partial command and press Tab to complete from your bash command history
695- Added search to installed plugins list - type to filter by name or description
696- Added support for pinning plugins to specific git commit SHAs, allowing marketplace entries to install exact versions
697- Fixed a regression where the context window blocking limit was calculated too aggressively, blocking users at ~65% context usage instead of the intended ~98%
698- Fixed memory issues that could cause crashes when running parallel subagents
699- Fixed memory leak in long-running sessions where stream resources were not cleaned up after shell commands completed
700- Fixed `@` symbol incorrectly triggering file autocomplete suggestions in bash mode
701- Fixed `@`-mention menu folder click behavior to navigate into directories instead of selecting them
702- Fixed `/feedback` command generating invalid GitHub issue URLs when description is very long
703- Fixed `/context` command to show the same token count and percentage as the status line in verbose mode
704- Fixed an issue where `/config`, `/context`, `/model`, and `/todos` command overlays could close unexpectedly
705- Fixed slash command autocomplete selecting wrong command when typing similar commands (e.g., `/context` vs `/compact`)
706- Fixed inconsistent back navigation in plugin marketplace when only one marketplace is configured
707- Fixed iTerm2 progress bar not clearing properly on exit, preventing lingering indicators and bell sounds
708- Improved backspace to delete pasted text as a single token instead of one character at a time
709- [VSCode] Added `/usage` command to display current plan usage
710
711## 2.1.12
712
713- Fixed message rendering bug
714
715## 2.1.11
716
717- Fixed excessive MCP connection requests for HTTP/SSE transports
718
719## 2.1.10
720
721- Added new `Setup` hook event that can be triggered via `--init`, `--init-only`, or `--maintenance` CLI flags for repository setup and maintenance operations
722- Added keyboard shortcut 'c' to copy OAuth URL when browser doesn't open automatically during login
723- Fixed a crash when running bash commands containing heredocs with JavaScript template literals like `${index + 1}`
724- Improved startup to capture keystrokes typed before the REPL is fully ready
725- Improved file suggestions to show as removable attachments instead of inserting text when accepted
726- [VSCode] Added install count display to plugin listings
727- [VSCode] Added trust warning when installing plugins
728
729## 2.1.9
730
731- Added `auto:N` syntax for configuring the MCP tool search auto-enable threshold, where N is the context window percentage (0-100)
732- Added `plansDirectory` setting to customize where plan files are stored
733- Added external editor support (Ctrl+G) in AskUserQuestion "Other" input field
734- Added session URL attribution to commits and PRs created from web sessions
735- Added support for `PreToolUse` hooks to return `additionalContext` to the model
736- Added `${CLAUDE_SESSION_ID}` string substitution for skills to access the current session ID
737- Fixed long sessions with parallel tool calls failing with an API error about orphan tool_result blocks
738- Fixed MCP server reconnection hanging when cached connection promise never resolves
739- Fixed Ctrl+Z suspend not working in terminals using Kitty keyboard protocol (Ghostty, iTerm2, kitty, WezTerm)
740
741## 2.1.7
742
743- Added `showTurnDuration` setting to hide turn duration messages (e.g., "Cooked for 1m 6s")
744- Added ability to provide feedback when accepting permission prompts
745- Added inline display of agent's final response in task notifications, making it easier to see results without reading the full transcript file
746- Fixed security vulnerability where wildcard permission rules could match compound commands containing shell operators
747- Fixed false "file modified" errors on Windows when cloud sync tools, antivirus scanners, or Git touch file timestamps without changing content
748- Fixed orphaned tool_result errors when sibling tools fail during streaming execution
749- Fixed context window blocking limit being calculated using the full context window instead of the effective context window (which reserves space for max output tokens)
750- Fixed spinner briefly flashing when running local slash commands like `/model` or `/theme`
751- Fixed terminal title animation jitter by using fixed-width braille characters
752- Fixed plugins with git submodules not being fully initialized when installed
753- Fixed bash commands failing on Windows when temp directory paths contained characters like `t` or `n` that were misinterpreted as escape sequences
754- Improved typing responsiveness by reducing memory allocation overhead in terminal rendering
755- Enabled MCP tool search auto mode by default for all users. When MCP tool descriptions exceed 10% of the context window, they are automatically deferred and discovered via the MCPSearch tool instead of being loaded upfront. This reduces context usage for users with many MCP tools configured. Users can disable this by adding `MCPSearch` to `disallowedTools` in their settings.
756- Changed OAuth and API Console URLs from console.anthropic.com to platform.claude.com
757- [VSCode] Fixed `claudeProcessWrapper` setting passing the wrapper path instead of the Claude binary path
758
759## 2.1.6
760
761- Added search functionality to `/config` command for quickly filtering settings
762- Added Updates section to `/doctor` showing auto-update channel and available npm versions (stable/latest)
763- Added date range filtering to `/stats` command - press `r` to cycle between Last 7 days, Last 30 days, and All time
764- Added automatic discovery of skills from nested `.claude/skills` directories when working with files in subdirectories
765- Added `context_window.used_percentage` and `context_window.remaining_percentage` fields to status line input for easier context window display
766- Added an error display when the editor fails during Ctrl+G
767- Fixed permission bypass via shell line continuation that could allow blocked commands to execute
768- Fixed false "File has been unexpectedly modified" errors when file watchers touch files without changing content
769- Fixed text styling (bold, colors) getting progressively misaligned in multi-line responses
770- Fixed the feedback panel closing unexpectedly when typing 'n' in the description field
771- Fixed rate limit warning appearing at low usage after weekly reset (now requires 70% usage)
772- Fixed rate limit options menu incorrectly auto-opening when resuming a previous session
773- Fixed numpad keys outputting escape sequences instead of characters in Kitty keyboard protocol terminals
774- Fixed Option+Return not inserting newlines in Kitty keyboard protocol terminals
775- Fixed corrupted config backup files accumulating in the home directory (now only one backup is created per config file)
776- Fixed `mcp list` and `mcp get` commands leaving orphaned MCP server processes
777- Fixed visual artifacts in ink2 mode when nodes become hidden via `display:none`
778- Improved the external CLAUDE.md imports approval dialog to show which files are being imported and from where
779- Improved the `/tasks` dialog to go directly to task details when there's only one background task running
780- Improved @ autocomplete with icons for different suggestion types and single-line formatting
781- Updated "Help improve Claude" setting fetch to refresh OAuth and retry when it fails due to a stale OAuth token
782- Changed task notification display to cap at 3 lines with overflow summary when multiple background tasks complete simultaneously
783- Changed terminal title to "Claude Code" on startup for better window identification
784- Removed ability to @-mention MCP servers to enable/disable - use `/mcp enable <name>` instead
785- [VSCode] Fixed usage indicator not updating after manual compact
786
787## 2.1.5
788
789- Added `CLAUDE_CODE_TMPDIR` environment variable to override the temp directory used for internal temp files, useful for environments with custom temp directory requirements
790
791## 2.1.4
792
793- Added `CLAUDE_CODE_DISABLE_BACKGROUND_TASKS` environment variable to disable all background task functionality including auto-backgrounding and the Ctrl+B shortcut
794- Fixed "Help improve Claude" setting fetch to refresh OAuth and retry when it fails due to a stale OAuth token
795
796## 2.1.3
797
798- Merged slash commands and skills, simplifying the mental model with no change in behavior
799- Added release channel (`stable` or `latest`) toggle to `/config`
800- Added detection and warnings for unreachable permission rules, with warnings in `/doctor` and after saving rules that include the source of each rule and actionable fix guidance
801- Fixed plan files persisting across `/clear` commands, now ensuring a fresh plan file is used after clearing a conversation
802- Fixed false skill duplicate detection on filesystems with large inodes (e.g., ExFAT) by using 64-bit precision for inode values
803- Fixed mismatch between background task count in status bar and items shown in tasks dialog
804- Fixed sub-agents using the wrong model during conversation compaction
805- Fixed web search in sub-agents using incorrect model
806- Fixed trust dialog acceptance when running from the home directory not enabling trust-requiring features like hooks during the session
807- Improved terminal rendering stability by preventing uncontrolled writes from corrupting cursor state
808- Improved slash command suggestion readability by truncating long descriptions to 2 lines
809- Changed tool hook execution timeout from 60 seconds to 10 minutes
810- [VSCode] Added clickable destination selector for permission requests, allowing you to choose where settings are saved (this project, all projects, shared with team, or session only)
811
812## 2.1.2
813
814- Added source path metadata to images dragged onto the terminal, helping Claude understand where images originated
815- Added clickable hyperlinks for file paths in tool output in terminals that support OSC 8 (like iTerm)
816- Added support for Windows Package Manager (winget) installations with automatic detection and update instructions
817- Added Shift+Tab keyboard shortcut in plan mode to quickly select "auto-accept edits" option
818- Added `FORCE_AUTOUPDATE_PLUGINS` environment variable to allow plugin autoupdate even when the main auto-updater is disabled
819- Added `agent_type` to SessionStart hook input, populated if `--agent` is specified
820- Fixed a command injection vulnerability in bash command processing where malformed input could execute arbitrary commands
821- Fixed a memory leak where tree-sitter parse trees were not being freed, causing WASM memory to grow unbounded over long sessions
822- Fixed binary files (images, PDFs, etc.) being accidentally included in memory when using `@include` directives in CLAUDE.md files
823- Fixed updates incorrectly claiming another installation is in progress
824- Fixed crash when socket files exist in watched directories (defense-in-depth for EOPNOTSUPP errors)
825- Fixed remote session URL and teleport being broken when using `/tasks` command
826- Fixed MCP tool names being exposed in analytics events by sanitizing user-specific server configurations
827- Improved Option-as-Meta hint on macOS to show terminal-specific instructions for native CSIu terminals like iTerm2, Kitty, and WezTerm
828- Improved error message when pasting images over SSH to suggest using `scp` instead of the unhelpful clipboard shortcut hint
829- Improved permission explainer to not flag routine dev workflows (git fetch/rebase, npm install, tests, PRs) as medium risk
830- Changed large bash command outputs to be saved to disk instead of truncated, allowing Claude to read the full content
831- Changed large tool outputs to be persisted to disk instead of truncated, providing full output access via file references
832- Changed `/plugins` installed tab to unify plugins and MCPs with scope-based grouping
833- Deprecated Windows managed settings path `C:\ProgramData\ClaudeCode\managed-settings.json` - administrators should migrate to `C:\Program Files\ClaudeCode\managed-settings.json`
834- [SDK] Changed minimum zod peer dependency to ^4.0.0
835- [VSCode] Fixed usage display not updating after manual compact
836
837## 2.1.0
838
839- Added automatic skill hot-reload - skills created or modified in `~/.claude/skills` or `.claude/skills` are now immediately available without restarting the session
840- Added support for running skills and slash commands in a forked sub-agent context using `context: fork` in skill frontmatter
841- Added support for `agent` field in skills to specify agent type for execution
842- Added `language` setting to configure Claude's response language (e.g., language: "japanese")
843- Changed Shift+Enter to work out of the box in iTerm2, WezTerm, Ghostty, and Kitty without modifying terminal configs
844- Added `respectGitignore` support in `settings.json` for per-project control over @-mention file picker behavior
845- Added `IS_DEMO` environment variable to hide email and organization from the UI, useful for streaming or recording sessions
846- Fixed security issue where sensitive data (OAuth tokens, API keys, passwords) could be exposed in debug logs
847- Fixed files and skills not being properly discovered when resuming sessions with `-c` or `--resume`
848- Fixed pasted content being lost when replaying prompts from history using up arrow or Ctrl+R search
849- Fixed Esc key with queued prompts to only move them to input without canceling the running task
850- Reduced permission prompts for complex bash commands
851- Fixed command search to prioritize exact and prefix matches on command names over fuzzy matches in descriptions
852- Fixed PreToolUse hooks to allow `updatedInput` when returning `ask` permission decision, enabling hooks to act as middleware while still requesting user consent
853- Fixed plugin path resolution for file-based marketplace sources
854- Fixed LSP tool being incorrectly enabled when no LSP servers were configured
855- Fixed background tasks failing with "git repository not found" error for repositories with dots in their names
856- Fixed Claude in Chrome support for WSL environments
857- Fixed Windows native installer silently failing when executable creation fails
858- Improved CLI help output to display options and subcommands in alphabetical order for easier navigation
859- Added wildcard pattern matching for Bash tool permissions using `*` at any position in rules (e.g., `Bash(npm *)`, `Bash(* install)`, `Bash(git * main)`)
860- Added unified Ctrl+B backgrounding for both bash commands and agents - pressing Ctrl+B now backgrounds all running foreground tasks simultaneously
861- Added support for MCP `list_changed` notifications, allowing MCP servers to dynamically update their available tools, prompts, and resources without requiring reconnection
862- Added `/teleport` and `/remote-env` slash commands for claude.ai subscribers, allowing them to resume and configure remote sessions
863- Added support for disabling specific agents using `Task(AgentName)` syntax in settings.json permissions or the `--disallowedTools` CLI flag
864- Added hooks support to agent frontmatter, allowing agents to define PreToolUse, PostToolUse, and Stop hooks scoped to the agent's lifecycle
865- Added hooks support for skill and slash command frontmatter
866- Added new Vim motions: `;` and `,` to repeat f/F/t/T motions, `y` operator for yank with `yy`/`Y`, `p`/`P` for paste, text objects (`iw`, `aw`, `iW`, `aW`, `i"`, `a"`, `i'`, `a'`, `i(`, `a(`, `i[`, `a[`, `i{`, `a{`), `>>` and `<<` for indent/dedent, and `J` to join lines
867- Added `/plan` command shortcut to enable plan mode directly from the prompt
868- Added slash command autocomplete support when `/` appears anywhere in input, not just at the beginning
869- Added `--tools` flag support in interactive mode to restrict which built-in tools Claude can use during interactive sessions
870- Added `CLAUDE_CODE_FILE_READ_MAX_OUTPUT_TOKENS` environment variable to override the default file read token limit
871- Added support for `once: true` config for hooks
872- Added support for YAML-style lists in frontmatter `allowed-tools` field for cleaner skill declarations
873- Added support for prompt and agent hook types from plugins (previously only command hooks were supported)
874- Added Cmd+V support for image paste in iTerm2 (maps to Ctrl+V)
875- Added left/right arrow key navigation for cycling through tabs in dialogs
876- Added real-time thinking block display in Ctrl+O transcript mode
877- Added filepath to full output in background bash task details dialog
878- Added Skills as a separate category in the context visualization
879- Fixed OAuth token refresh not triggering when server reports token expired but local expiration check disagrees
880- Fixed session persistence getting stuck after transient server errors by recovering from 409 conflicts when the entry was actually stored
881- Fixed session resume failures caused by orphaned tool results during concurrent tool execution
882- Fixed a race condition where stale OAuth tokens could be read from the keychain cache during concurrent token refresh attempts
883- Fixed AWS Bedrock subagents not inheriting EU/APAC cross-region inference model configuration, causing 403 errors when IAM permissions are scoped to specific regions
884- Fixed API context overflow when background tasks produce large output by truncating to 30K chars with file path reference
885- Fixed a hang when reading FIFO files by skipping symlink resolution for special file types
886- Fixed terminal keyboard mode not being reset on exit in Ghostty, iTerm2, Kitty, and WezTerm
887- Fixed Alt+B and Alt+F (word navigation) not working in iTerm2, Ghostty, Kitty, and WezTerm
888- Fixed `${CLAUDE_PLUGIN_ROOT}` not being substituted in plugin `allowed-tools` frontmatter, which caused tools to incorrectly require approval
889- Fixed files created by the Write tool using hardcoded 0o600 permissions instead of respecting the system umask
890- Fixed commands with `$()` command substitution failing with parse errors
891- Fixed multi-line bash commands with backslash continuations being incorrectly split and flagged for permissions
892- Fixed bash command prefix extraction to correctly identify subcommands after global options (e.g., `git -C /path log` now correctly matches `Bash(git log:*)` rules)
893- Fixed slash commands passed as CLI arguments (e.g., `claude /context`) not being executed properly
894- Fixed pressing Enter after Tab-completing a slash command selecting a different command instead of submitting the completed one
895- Fixed slash command argument hint flickering and inconsistent display when typing commands with arguments
896- Fixed Claude sometimes redundantly invoking the Skill tool when running slash commands directly
897- Fixed skill token estimates in `/context` to accurately reflect frontmatter-only loading
898- Fixed subagents sometimes not inheriting the parent's model by default
899- Fixed model picker showing incorrect selection for Bedrock/Vertex users using `--model haiku`
900- Fixed duplicate Bash commands appearing in permission request option labels
901- Fixed noisy output when background tasks complete - now shows clean completion message instead of raw output
902- Fixed background task completion notifications to appear proactively with bullet point
903- Fixed forked slash commands showing "AbortError" instead of "Interrupted" message when cancelled
904- Fixed cursor disappearing after dismissing permission dialogs
905- Fixed `/hooks` menu selecting wrong hook type when scrolling to a different option
906- Fixed images in queued prompts showing as "[object Object]" when pressing Esc to cancel
907- Fixed images being silently dropped when queueing messages while backgrounding a task
908- Fixed large pasted images failing with "Image was too large" error
909- Fixed extra blank lines in multiline prompts containing CJK characters (Japanese, Chinese, Korean)
910- Fixed ultrathink keyword highlighting being applied to wrong characters when user prompt text wraps to multiple lines
911- Fixed collapsed "Reading X files…" indicator incorrectly switching to past tense when thinking blocks appear mid-stream
912- Fixed Bash read commands (like `ls` and `cat`) not being counted in collapsed read/search groups, causing groups to incorrectly show "Read 0 files"
913- Fixed spinner token counter to properly accumulate tokens from subagents during execution
914- Fixed memory leak in git diff parsing where sliced strings retained large parent strings
915- Fixed race condition where LSP tool could return "no server available" during startup
916- Fixed feedback submission hanging indefinitely when network requests timeout
917- Fixed search mode in plugin discovery and log selector views exiting when pressing up arrow
918- Fixed hook success message showing trailing colon when hook has no output
919- Multiple optimizations to improve startup performance
920- Improved terminal rendering performance when using native installer or Bun, especially for text with emoji, ANSI codes, and Unicode characters
921- Improved performance when reading Jupyter notebooks with many cells
922- Improved reliability for piped input like `cat refactor.md | claude`
923- Improved reliability for AskQuestion tool
924- Improved sed in-place edit commands to render as file edits with diff preview
925- Improved Claude to automatically continue when response is cut off due to output token limit, instead of showing an error message
926- Improved compaction reliability
927- Improved subagents (Task tool) to continue working after permission denial, allowing them to try alternative approaches
928- Improved skills to show progress while executing, displaying tool uses as they happen
929- Improved skills from `/skills/` directories to be visible in the slash command menu by default (opt-out with `user-invocable: false` in frontmatter)
930- Improved skill suggestions to prioritize recently and frequently used skills
931- Improved spinner feedback when waiting for the first response token
932- Improved token count display in spinner to include tokens from background agents
933- Improved incremental output for async agents to give the main thread more control and visibility
934- Improved permission prompt UX with Tab hint moved to footer, cleaner Yes/No input labels with contextual placeholders
935- Improved the Claude in Chrome notification with shortened help text and persistent display until dismissed
936- Improved macOS screenshot paste reliability with TIFF format support
937- Improved `/stats` output
938- Updated Atlassian MCP integration to use a more reliable default configuration (streamable HTTP)
939- Changed "Interrupted" message color from red to grey for a less alarming appearance
940- Removed permission prompt when entering plan mode - users can now enter plan mode without approval
941- Removed underline styling from image reference links
942- [SDK] Changed minimum zod peer dependency to ^4.0.0
943- [VSCode] Added currently selected model name to the context menu
944- [VSCode] Added descriptive labels on auto-accept permission button (e.g., "Yes, allow npm for this project" instead of "Yes, and don't ask again")
945- [VSCode] Fixed paragraph breaks not rendering in markdown content
946- [VSCode] Fixed scrolling in the extension inadvertently scrolling the parent iframe
947- [Windows] Fixed issue with improper rendering
948
949## 2.0.76
950
951- Fixed issue with macOS code-sign warning when using Claude in Chrome integration
952
953## 2.0.75
954
955- Minor bugfixes
956
957## 2.0.74
958
959- Added LSP (Language Server Protocol) tool for code intelligence features like go-to-definition, find references, and hover documentation
960- Added `/terminal-setup` support for Kitty, Alacritty, Zed, and Warp terminals
961- Added ctrl+t shortcut in `/theme` to toggle syntax highlighting on/off
962- Added syntax highlighting info to theme picker
963- Added guidance for macOS users when Alt shortcuts fail due to terminal configuration
964- Fixed skill `allowed-tools` not being applied to tools invoked by the skill
965- Fixed Opus 4.5 tip incorrectly showing when user was already using Opus
966- Fixed a potential crash when syntax highlighting isn't initialized correctly
967- Fixed visual bug in `/plugins discover` where list selection indicator showed while search box was focused
968- Fixed macOS keyboard shortcuts to display 'opt' instead of 'alt'
969- Improved `/context` command visualization with grouped skills and agents by source, slash commands, and sorted token count
970- [Windows] Fixed issue with improper rendering
971- [VSCode] Added gift tag pictogram for year-end promotion message
972
973## 2.0.73
974
975- Added clickable `[Image #N]` links that open attached images in the default viewer
976- Added alt-y yank-pop to cycle through kill ring history after ctrl-y yank
977- Added search filtering to the plugin discover screen (type to filter by name, description, or marketplace)
978- Added support for custom session IDs when forking sessions with `--session-id` combined with `--resume` or `--continue` and `--fork-session`
979- Fixed slow input history cycling and race condition that could overwrite text after message submission
980- Improved `/theme` command to open theme picker directly
981- Improved theme picker UI
982- Improved search UX across resume session, permissions, and plugins screens with a unified SearchBox component
983- [VSCode] Added tab icon badges showing pending permissions (blue) and unread completions (orange)
984
985## 2.0.72
986
987- Added Claude in Chrome (Beta) feature that works with the Chrome extension (https://claude.ai/chrome) to let you control your browser directly from Claude Code
988- Reduced terminal flickering
989- Added scannable QR code to mobile app tip for quick app downloads
990- Added loading indicator when resuming conversations for better feedback
991- Fixed `/context` command not respecting custom system prompts in non-interactive mode
992- Fixed order of consecutive Ctrl+K lines when pasting with Ctrl+Y
993- Improved @ mention file suggestion speed (~3× faster in git repositories)
994- Improved file suggestion performance in repos with `.ignore` or `.rgignore` files
995- Improved settings validation errors to be more prominent
996- Changed thinking toggle from Tab to Alt+T to avoid accidental triggers
997
998## 2.0.71
999
1000- Added /config toggle to enable/disable prompt suggestions
1001- Added `/settings` as an alias for the `/config` command
1002- Fixed @ file reference suggestions incorrectly triggering when cursor is in the middle of a path
1003- Fixed MCP servers from `.mcp.json` not loading when using `--dangerously-skip-permissions`
1004- Fixed permission rules incorrectly rejecting valid bash commands containing shell glob patterns (e.g., `ls *.txt`, `for f in *.png`)
1005- Bedrock: Environment variable `ANTHROPIC_BEDROCK_BASE_URL` is now respected for token counting and inference profile listing
1006- New syntax highlighting engine for native build
1007
1008## 2.0.70
1009
1010- Added Enter key to accept and submit prompt suggestions immediately (tab still accepts for editing)
1011- Added wildcard syntax `mcp__server__*` for MCP tool permissions to allow or deny all tools from a server
1012- Added auto-update toggle for plugin marketplaces, allowing per-marketplace control over automatic updates
1013- Added `current_usage` field to status line input, enabling accurate context window percentage calculations
1014- Fixed input being cleared when processing queued commands while the user was typing
1015- Fixed prompt suggestions replacing typed input when pressing Tab
1016- Fixed diff view not updating when terminal is resized
1017- Improved memory usage by 3x for large conversations
1018- Improved resolution of stats screenshots copied to clipboard (Ctrl+S) for crisper images
1019- Removed # shortcut for quick memory entry (tell Claude to edit your CLAUDE.md instead)
1020- Fix thinking mode toggle in /config not persisting correctly
1021- Improve UI for file creation permission dialog
1022
1023## 2.0.69
1024
1025- Minor bugfixes
1026
1027## 2.0.68
1028
1029- Fixed IME (Input Method Editor) support for languages like Chinese, Japanese, and Korean by correctly positioning the composition window at the cursor
1030- Fixed a bug where disallowed MCP tools were visible to the model
1031- Fixed an issue where steering messages could be lost while a subagent is working
1032- Fixed Option+Arrow word navigation treating entire CJK (Chinese, Japanese, Korean) text sequences as a single word instead of navigating by word boundaries
1033- Improved plan mode exit UX: show simplified yes/no dialog when exiting with empty or missing plan instead of throwing an error
1034- Add support for enterprise managed settings. Contact your Anthropic account team to enable this feature.
1035
1036## 2.0.67
1037
1038- Thinking mode is now enabled by default for Opus 4.5
1039- Thinking mode configuration has moved to /config
1040- Added search functionality to `/permissions` command with `/` keyboard shortcut for filtering rules by tool name
1041- Show reason why autoupdater is disabled in `/doctor`
1042- Fixed false "Another process is currently updating Claude" error when running `claude update` while another instance is already on the latest version
1043- Fixed MCP servers from `.mcp.json` being stuck in pending state when running in non-interactive mode (`-p` flag or piped input)
1044- Fixed scroll position resetting after deleting a permission rule in `/permissions`
1045- Fixed word deletion (opt+delete) and word navigation (opt+arrow) not working correctly with non-Latin text such as Cyrillic, Greek, Arabic, Hebrew, Thai, and Chinese
1046- Fixed `claude install --force` not bypassing stale lock files
1047- Fixed consecutive @~/ file references in CLAUDE.md being incorrectly parsed due to markdown strikethrough interference
1048- Windows: Fixed plugin MCP servers failing due to colons in log directory paths
1049
1050## 2.0.65
1051
1052- Added ability to switch models while writing a prompt using alt+p (linux, windows), option+p (macos).
1053- Added context window information to status line input
1054- Added `fileSuggestion` setting for custom `@` file search commands
1055- Added `CLAUDE_CODE_SHELL` environment variable to override automatic shell detection (useful when login shell differs from actual working shell)
1056- Fixed prompt not being saved to history when aborting a query with Escape
1057- Fixed Read tool image handling to identify format from bytes instead of file extension
1058
1059## 2.0.64
1060
1061- Made auto-compacting instant
1062- Agents and bash commands can run asynchronously and send messages to wake up the main agent
1063- /stats now provides users with interesting CC stats, such as favorite model, usage graph, usage streak
1064- Added named session support: use `/rename` to name sessions, `/resume <name>` in REPL or `claude --resume <name>` from the terminal to resume them
1065- Added support for .claude/rules/`. See https://code.claude.com/docs/en/memory for details.
1066- Added image dimension metadata when images are resized, enabling accurate coordinate mappings for large images
1067- Fixed auto-loading .env when using native installer
1068- Fixed `--system-prompt` being ignored when using `--continue` or `--resume` flags
1069- Improved `/resume` screen with grouped forked sessions and keyboard shortcuts for preview (P) and rename (R)
1070- VSCode: Added copy-to-clipboard button on code blocks and bash tool inputs
1071- VSCode: Fixed extension not working on Windows ARM64 by falling back to x64 binary via emulation
1072- Bedrock: Improve efficiency of token counting
1073- Bedrock: Add support for `aws login` AWS Management Console credentials
1074- Unshipped AgentOutputTool and BashOutputTool, in favor of a new unified TaskOutputTool
1075
1076## 2.0.62
1077
1078- Added "(Recommended)" indicator for multiple-choice questions, with the recommended option moved to the top of the list
1079- Added `attribution` setting to customize commit and PR bylines (deprecates `includeCoAuthoredBy`)
1080- Fixed duplicate slash commands appearing when ~/.claude is symlinked to a project directory
1081- Fixed slash command selection not working when multiple commands share the same name
1082- Fixed an issue where skill files inside symlinked skill directories could become circular symlinks
1083- Fixed running versions getting removed because lock file incorrectly going stale
1084- Fixed IDE diff tab not closing when rejecting file changes
1085
1086## 2.0.61
1087
1088- Reverted VSCode support for multiple terminal clients due to responsiveness issues.
1089
1090## 2.0.60
1091
1092- Added background agent support. Agents run in the background while you work
1093- Added --disable-slash-commands CLI flag to disable all slash commands
1094- Added model name to "Co-Authored-By" commit messages
1095- Enabled "/mcp enable [server-name]" or "/mcp disable [server-name]" to quickly toggle all servers
1096- Updated Fetch to skip summarization for pre-approved websites
1097- VSCode: Added support for multiple terminal clients connecting to the IDE server simultaneously
1098
1099## 2.0.59
1100
1101- Added --agent CLI flag to override the agent setting for the current session
1102- Added `agent` setting to configure main thread with a specific agent's system prompt, tool restrictions, and model
1103- VS Code: Fixed .claude.json config file being read from incorrect location
1104
1105## 2.0.58
1106
1107- Pro users now have access to Opus 4.5 as part of their subscription!
1108- Fixed timer duration showing "11m 60s" instead of "12m 0s"
1109- Windows: Managed settings now prefer `C:\Program Files\ClaudeCode` if it exists. Support for `C:\ProgramData\ClaudeCode` will be removed in a future version.
1110
1111## 2.0.57
1112
1113- Added feedback input when rejecting plans, allowing users to tell Claude what to change
1114- VSCode: Added streaming message support for real-time response display
1115
1116## 2.0.56
1117
1118- Added setting to enable/disable terminal progress bar (OSC 9;4)
1119- VSCode Extension: Added support for VS Code's secondary sidebar (VS Code 1.97+), allowing Claude Code to be displayed in the right sidebar while keeping the file explorer on the left. Requires setting sidebar as Preferred Location in the config.
1120
1121## 2.0.55
1122
1123- Fixed proxy DNS resolution being forced on by default. Now opt-in via `CLAUDE_CODE_PROXY_RESOLVES_HOSTS=true` environment variable
1124- Fixed keyboard navigation becoming unresponsive when holding down arrow keys in memory location selector
1125- Improved AskUserQuestion tool to auto-submit single-select questions on the last question, eliminating the extra review screen for simple question flows
1126- Improved fuzzy matching for `@` file suggestions with faster, more accurate results
1127
1128## 2.0.54
1129
1130- Hooks: Enable PermissionRequest hooks to process 'always allow' suggestions and apply permission updates
1131- Fix issue with excessive iTerm notifications
1132
1133## 2.0.52
1134
1135- Fixed duplicate message display when starting Claude with a command line argument
1136- Fixed `/usage` command progress bars to fill up as usage increases (instead of showing remaining percentage)
1137- Fixed image pasting not working on Linux systems running Wayland (now falls back to wl-paste when xclip is unavailable)
1138- Permit some uses of `$!` in bash commands
1139
1140## 2.0.51
1141
1142- Added Opus 4.5! https://www.anthropic.com/news/claude-opus-4-5
1143- Introducing Claude Code for Desktop: https://claude.com/download
1144- To give you room to try out our new model, we've updated usage limits for Claude Code users. See the Claude Opus 4.5 blog for full details
1145- Pro users can now purchase extra usage for access to Opus 4.5 in Claude Code
1146- Plan Mode now builds more precise plans and executes more thoroughly
1147- Usage limit notifications now easier to understand
1148- Switched `/usage` back to "% used"
1149- Fixed handling of thinking errors
1150- Fixed performance regression
1151
1152## 2.0.50
1153
1154- Fixed bug preventing calling MCP tools that have nested references in their input schemas
1155- Silenced a noisy but harmless error during upgrades
1156- Improved ultrathink text display
1157- Improved clarity of 5-hour session limit warning message
1158
1159## 2.0.49
1160
1161- Added readline-style ctrl-y for pasting deleted text
1162- Improved clarity of usage limit warning message
1163- Fixed handling of subagent permissions
1164
1165## 2.0.47
1166
1167- Improved error messages and validation for `claude --teleport`
1168- Improved error handling in `/usage`
1169- Fixed race condition with history entry not getting logged at exit
1170- Fixed Vertex AI configuration not being applied from `settings.json`
1171
1172## 2.0.46
1173
1174- Fixed image files being reported with incorrect media type when format cannot be detected from metadata
1175
1176## 2.0.45
1177
1178- Added support for Microsoft Foundry! See https://code.claude.com/docs/en/azure-ai-foundry
1179- Added `PermissionRequest` hook to automatically approve or deny tool permission requests with custom logic
1180- Send background tasks to Claude Code on the web by starting a message with `&`
1181
1182## 2.0.43
1183
1184- Added `permissionMode` field for custom agents
1185- Added `tool_use_id` field to `PreToolUseHookInput` and `PostToolUseHookInput` types
1186- Added skills frontmatter field to declare skills to auto-load for subagents
1187- Added the `SubagentStart` hook event
1188- Fixed nested `CLAUDE.md` files not loading when @-mentioning files
1189- Fixed duplicate rendering of some messages in the UI
1190- Fixed some visual flickers
1191- Fixed NotebookEdit tool inserting cells at incorrect positions when cell IDs matched the pattern `cell-N`
1192
1193## 2.0.42
1194
1195- Added `agent_id` and `agent_transcript_path` fields to `SubagentStop` hooks.
1196
1197## 2.0.41
1198
1199- Added `model` parameter to prompt-based stop hooks, allowing users to specify a custom model for hook evaluation
1200- Fixed slash commands from user settings being loaded twice, which could cause rendering issues
1201- Fixed incorrect labeling of user settings vs project settings in command descriptions
1202- Fixed crash when plugin command hooks timeout during execution
1203- Fixed: Bedrock users no longer see duplicate Opus entries in the /model picker when using `--model haiku`
1204- Fixed broken security documentation links in trust dialogs and onboarding
1205- Fixed issue where pressing ESC to close the diff modal would also interrupt the model
1206- ctrl-r history search landing on a slash command no longer cancels the search
1207- SDK: Support custom timeouts for hooks
1208- Allow more safe git commands to run without approval
1209- Plugins: Added support for sharing and installing output styles
1210- Teleporting a session from web will automatically set the upstream branch
1211
1212## 2.0.37
1213
1214- Fixed how idleness is computed for notifications
1215- Hooks: Added matcher values for Notification hook events
1216- Output Styles: Added `keep-coding-instructions` option to frontmatter
1217
1218## 2.0.36
1219
1220- Fixed: DISABLE_AUTOUPDATER environment variable now properly disables package manager update notifications
1221- Fixed queued messages being incorrectly executed as bash commands
1222- Fixed input being lost when typing while a queued message is processed
1223
1224## 2.0.35
1225
1226- Improve fuzzy search results when searching commands
1227- Improved VS Code extension to respect `chat.fontSize` and `chat.fontFamily` settings throughout the entire UI, and apply font changes immediately without requiring reload
1228- Added `CLAUDE_CODE_EXIT_AFTER_STOP_DELAY` environment variable to automatically exit SDK mode after a specified idle duration, useful for automated workflows and scripts
1229- Migrated `ignorePatterns` from project config to deny permissions in the localSettings.
1230- Fixed menu navigation getting stuck on items with empty string or other falsy values (e.g., in the `/hooks` menu)
1231
1232## 2.0.34
1233
1234- VSCode Extension: Added setting to configure the initial permission mode for new conversations
1235- Improved file path suggestion performance with native Rust-based fuzzy finder
1236- Fixed infinite token refresh loop that caused MCP servers with OAuth (e.g., Slack) to hang during connection
1237- Fixed memory crash when reading or writing large files (especially base64-encoded images)
1238
1239## 2.0.33
1240
1241- Native binary installs now launch quicker.
1242- Fixed `claude doctor` incorrectly detecting Homebrew vs npm-global installations by properly resolving symlinks
1243- Fixed `claude mcp serve` exposing tools with incompatible outputSchemas
1244
1245## 2.0.32
1246
1247- Un-deprecate output styles based on community feedback
1248- Added `companyAnnouncements` setting for displaying announcements on startup
1249- Fixed hook progress messages not updating correctly during PostToolUse hook execution
1250
1251## 2.0.31
1252
1253- Windows: native installation uses shift+tab as shortcut for mode switching, instead of alt+m
1254- Vertex: add support for Web Search on supported models
1255- VSCode: Adding the respectGitIgnore configuration to include .gitignored files in file searches (defaults to true)
1256- Fixed a bug with subagents and MCP servers related to "Tool names must be unique" error
1257- Fixed issue causing `/compact` to fail with `prompt_too_long` by making it respect existing compact boundaries
1258- Fixed plugin uninstall not removing plugins
1259
1260## 2.0.30
1261
1262- Added helpful hint to run `security unlock-keychain` when encountering API key errors on macOS with locked keychain
1263- Added `allowUnsandboxedCommands` sandbox setting to disable the dangerouslyDisableSandbox escape hatch at policy level
1264- Added `disallowedTools` field to custom agent definitions for explicit tool blocking
1265- Added prompt-based stop hooks
1266- VSCode: Added respectGitIgnore configuration to include .gitignored files in file searches (defaults to true)
1267- Enabled SSE MCP servers on native build
1268- Deprecated output styles. Review options in `/output-style` and use --system-prompt-file, --system-prompt, --append-system-prompt, CLAUDE.md, or plugins instead
1269- Removed support for custom ripgrep configuration, resolving an issue where Search returns no results and config discovery fails
1270- Fixed Explore agent creating unwanted .md investigation files during codebase exploration
1271- Fixed a bug where `/context` would sometimes fail with "max_tokens must be greater than thinking.budget_tokens" error message
1272- Fixed `--mcp-config` flag to correctly override file-based MCP configurations
1273- Fixed bug that saved session permissions to local settings
1274- Fixed MCP tools not being available to sub-agents
1275- Fixed hooks and plugins not executing when using --dangerously-skip-permissions flag
1276- Fixed delay when navigating through typeahead suggestions with arrow keys
1277- VSCode: Restored selection indicator in input footer showing current file or code selection status
1278
1279## 2.0.28
1280
1281- Plan mode: introduced new Plan subagent
1282- Subagents: claude can now choose to resume subagents
1283- Subagents: claude can dynamically choose the model used by its subagents
1284- SDK: added --max-budget-usd flag
1285- Discovery of custom slash commands, subagents, and output styles no longer respects .gitignore
1286- Stop `/terminal-setup` from adding backslash to `Shift + Enter` in VS Code
1287- Add branch and tag support for git-based plugins and marketplaces using fragment syntax (e.g., `owner/repo#branch`)
1288- Fixed a bug where macOS permission prompts would show up upon initial launch when launching from home directory
1289- Various other bug fixes
1290
1291## 2.0.27
1292
1293- New UI for permission prompts
1294- Added current branch filtering and search to session resume screen for easier navigation
1295- Fixed directory @-mention causing "No assistant message found" error
1296- VSCode Extension: Add config setting to include .gitignored files in file searches
1297- VSCode Extension: Bug fixes for unrelated 'Warmup' conversations, and configuration/settings occasionally being reset to defaults
1298
1299## 2.0.25
1300
1301- Removed legacy SDK entrypoint. Please migrate to @anthropic-ai/claude-agent-sdk for future SDK updates: https://platform.claude.com/docs/en/agent-sdk/migration-guide
1302
1303## 2.0.24
1304
1305- Fixed a bug where project-level skills were not loading when --setting-sources 'project' was specified
1306- Claude Code Web: Support for Web -> CLI teleport
1307- Sandbox: Releasing a sandbox mode for the BashTool on Linux & Mac
1308- Bedrock: Display awsAuthRefresh output when auth is required
1309
1310## 2.0.22
1311
1312- Fixed content layout shift when scrolling through slash commands
1313- IDE: Add toggle to enable/disable thinking.
1314- Fix bug causing duplicate permission prompts with parallel tool calls
1315- Add support for enterprise managed MCP allowlist and denylist
1316
1317## 2.0.21
1318
1319- Support MCP `structuredContent` field in tool responses
1320- Added an interactive question tool
1321- Claude will now ask you questions more often in plan mode
1322- Added Haiku 4.5 as a model option for Pro users
1323- Fixed an issue where queued commands don't have access to previous messages' output
1324
1325## 2.0.20
1326
1327- Added support for Claude Skills
1328
1329## 2.0.19
1330
1331- Auto-background long-running bash commands instead of killing them. Customize with BASH_DEFAULT_TIMEOUT_MS
1332- Fixed a bug where Haiku was unnecessarily called in print mode
1333
1334## 2.0.17
1335
1336- Added Haiku 4.5 to model selector!
1337- Haiku 4.5 automatically uses Sonnet in plan mode, and Haiku for execution (i.e. SonnetPlan by default)
1338- 3P (Bedrock and Vertex) are not automatically upgraded yet. Manual upgrading can be done through setting `ANTHROPIC_DEFAULT_HAIKU_MODEL`
1339- Introducing the Explore subagent. Powered by Haiku it'll search through your codebase efficiently to save context!
1340- OTEL: support HTTP_PROXY and HTTPS_PROXY
1341- `CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC` now disables release notes fetching
1342
1343## 2.0.15
1344
1345- Fixed bug with resuming where previously created files needed to be read again before writing
1346- Fixed bug with `-p` mode where @-mentioned files needed to be read again before writing
1347
1348## 2.0.14
1349
1350- Fix @-mentioning MCP servers to toggle them on/off
1351- Improve permission checks for bash with inline env vars
1352- Fix ultrathink + thinking toggle
1353- Reduce unnecessary logins
1354- Document --system-prompt
1355- Several improvements to rendering
1356- Plugins UI polish
1357
1358## 2.0.13
1359
1360- Fixed `/plugin` not working on native build
1361
1362## 2.0.12
1363
1364- **Plugin System Released**: Extend Claude Code with custom commands, agents, hooks, and MCP servers from marketplaces
1365- `/plugin install`, `/plugin enable/disable`, `/plugin marketplace` commands for plugin management
1366- Repository-level plugin configuration via `extraKnownMarketplaces` for team collaboration
1367- `/plugin validate` command for validating plugin structure and configuration
1368- Plugin announcement blog post at https://www.anthropic.com/news/claude-code-plugins
1369- Plugin documentation available at https://code.claude.com/docs/en/plugins
1370- Comprehensive error messages and diagnostics via `/doctor` command
1371- Avoid flickering in `/model` selector
1372- Improvements to `/help`
1373- Avoid mentioning hooks in `/resume` summaries
1374- Changes to the "verbose" setting in `/config` now persist across sessions
1375
1376## 2.0.11
1377
1378- Reduced system prompt size by 1.4k tokens
1379- IDE: Fixed keyboard shortcuts and focus issues for smoother interaction
1380- Fixed Opus fallback rate limit errors appearing incorrectly
1381- Fixed /add-dir command selecting wrong default tab
1382
1383## 2.0.10
1384
1385- Rewrote terminal renderer for buttery smooth UI
1386- Enable/disable MCP servers by @mentioning, or in /mcp
1387- Added tab completion for shell commands in bash mode
1388- PreToolUse hooks can now modify tool inputs
1389- Press Ctrl-G to edit your prompt in your system's configured text editor
1390- Fixes for bash permission checks with environment variables in the command
1391
1392## 2.0.9
1393
1394- Fix regression where bash backgrounding stopped working
1395
1396## 2.0.8
1397
1398- Update Bedrock default Sonnet model to `global.anthropic.claude-sonnet-4-5-20250929-v1:0`
1399- IDE: Add drag-and-drop support for files and folders in chat
1400- /context: Fix counting for thinking blocks
1401- Improve message rendering for users with light themes on dark terminals
1402- Remove deprecated .claude.json allowedTools, ignorePatterns, env, and todoFeatureEnabled config options (instead, configure these in your settings.json)
1403
1404## 2.0.5
1405
1406- IDE: Fix IME unintended message submission with Enter and Tab
1407- IDE: Add "Open in Terminal" link in login screen
1408- Fix unhandled OAuth expiration 401 API errors
1409- SDK: Added SDKUserMessageReplay.isReplay to prevent duplicate messages
1410
1411## 2.0.1
1412
1413- Skip Sonnet 4.5 default model setting change for Bedrock and Vertex
1414- Various bug fixes and presentation improvements
1415
1416## 2.0.0
1417
1418- New native VS Code extension
1419- Fresh coat of paint throughout the whole app
1420- /rewind a conversation to undo code changes
1421- /usage command to see plan limits
1422- Tab to toggle thinking (sticky across sessions)
1423- Ctrl-R to search history
1424- Unshipped claude config command
1425- Hooks: Reduced PostToolUse 'tool_use' ids were found without 'tool_result' blocks errors
1426- SDK: The Claude Code SDK is now the Claude Agent SDK
1427- Add subagents dynamically with `--agents` flag
1428
1429## 1.0.126
1430
1431- Enable /context command for Bedrock and Vertex
1432- Add mTLS support for HTTP-based OpenTelemetry exporters
1433
1434## 1.0.124
1435
1436- Set `CLAUDE_BASH_NO_LOGIN` environment variable to 1 or true to to skip login shell for BashTool
1437- Fix Bedrock and Vertex environment variables evaluating all strings as truthy
1438- No longer inform Claude of the list of allowed tools when permission is denied
1439- Fixed security vulnerability in Bash tool permission checks
1440- Improved VSCode extension performance for large files
1441
1442## 1.0.123
1443
1444- Bash permission rules now support output redirections when matching (e.g., `Bash(python:*)` matches `python script.py > output.txt`)
1445- Fixed thinking mode triggering on negation phrases like "don't think"
1446- Fixed rendering performance degradation during token streaming
1447- Added SlashCommand tool, which enables Claude to invoke your slash commands. https://code.claude.com/docs/en/slash-commands#SlashCommand-tool
1448- Enhanced BashTool environment snapshot logging
1449- Fixed a bug where resuming a conversation in headless mode would sometimes enable thinking unnecessarily
1450- Migrated --debug logging to a file, to enable easy tailing & filtering
1451
1452## 1.0.120
1453
1454- Fix input lag during typing, especially noticeable with large prompts
1455- Improved VSCode extension command registry and sessions dialog user experience
1456- Enhanced sessions dialog responsiveness and visual feedback
1457- Fixed IDE compatibility issue by removing worktree support check
1458- Fixed security vulnerability where Bash tool permission checks could be bypassed using prefix matching
1459
1460## 1.0.119
1461
1462- Fix Windows issue where process visually freezes on entering interactive mode
1463- Support dynamic headers for MCP servers via headersHelper configuration
1464- Fix thinking mode not working in headless sessions
1465- Fix slash commands now properly update allowed tools instead of replacing them
1466
1467## 1.0.117
1468
1469- Add Ctrl-R history search to recall previous commands like bash/zsh
1470- Fix input lag while typing, especially on Windows
1471- Add sed command to auto-allowed commands in acceptEdits mode
1472- Fix Windows PATH comparison to be case-insensitive for drive letters
1473- Add permissions management hint to /add-dir output
1474
1475## 1.0.115
1476
1477- Improve thinking mode display with enhanced visual effects
1478- Type /t to temporarily disable thinking mode in your prompt
1479- Improve path validation for glob and grep tools
1480- Show condensed output for post-tool hooks to reduce visual clutter
1481- Fix visual feedback when loading state completes
1482- Improve UI consistency for permission request dialogs
1483
1484## 1.0.113
1485
1486- Deprecated piped input in interactive mode
1487- Move Ctrl+R keybinding for toggling transcript to Ctrl+O
1488
1489## 1.0.112
1490
1491- Transcript mode (Ctrl+R): Added the model used to generate each assistant message
1492- Addressed issue where some Claude Max users were incorrectly recognized as Claude Pro users
1493- Hooks: Added systemMessage support for SessionEnd hooks
1494- Added `spinnerTipsEnabled` setting to disable spinner tips
1495- IDE: Various improvements and bug fixes
1496
1497## 1.0.111
1498
1499- /model now validates provided model names
1500- Fixed Bash tool crashes caused by malformed shell syntax parsing
1501
1502## 1.0.110
1503
1504- /terminal-setup command now supports WezTerm
1505- MCP: OAuth tokens now proactively refresh before expiration
1506- Fixed reliability issues with background Bash processes
1507
1508## 1.0.109
1509
1510- SDK: Added partial message streaming support via `--include-partial-messages` CLI flag
1511
1512## 1.0.106
1513
1514- Windows: Fixed path permission matching to consistently use POSIX format (e.g., `Read(//c/Users/...)`)
1515
1516## 1.0.97
1517
1518- Settings: /doctor now validates permission rule syntax and suggests corrections
1519
1520## 1.0.94
1521
1522- Vertex: add support for global endpoints for supported models
1523- /memory command now allows direct editing of all imported memory files
1524- SDK: Add custom tools as callbacks
1525- Added /todos command to list current todo items
1526
1527## 1.0.93
1528
1529- Windows: Add alt + v shortcut for pasting images from clipboard
1530- Support NO_PROXY environment variable to bypass proxy for specified hostnames and IPs
1531
1532## 1.0.90
1533
1534- Settings file changes take effect immediately - no restart required
1535
1536## 1.0.88
1537
1538- Fixed issue causing "OAuth authentication is currently not supported"
1539- Status line input now includes `exceeds_200k_tokens`
1540- Fixed incorrect usage tracking in /cost.
1541- Introduced `ANTHROPIC_DEFAULT_SONNET_MODEL` and `ANTHROPIC_DEFAULT_OPUS_MODEL` for controlling model aliases opusplan, opus, and sonnet.
1542- Bedrock: Updated default Sonnet model to Sonnet 4
1543
1544## 1.0.86
1545
1546- Added /context to help users self-serve debug context issues
1547- SDK: Added UUID support for all SDK messages
1548- SDK: Added `--replay-user-messages` to replay user messages back to stdout
1549
1550## 1.0.85
1551
1552- Status line input now includes session cost info
1553- Hooks: Introduced SessionEnd hook
1554
1555## 1.0.84
1556
1557- Fix tool_use/tool_result id mismatch error when network is unstable
1558- Fix Claude sometimes ignoring real-time steering when wrapping up a task
1559- @-mention: Add ~/.claude/\* files to suggestions for easier agent, output style, and slash command editing
1560- Use built-in ripgrep by default; to opt out of this behavior, set USE_BUILTIN_RIPGREP=0
1561
1562## 1.0.83
1563
1564- @-mention: Support files with spaces in path
1565- New shimmering spinner
1566
1567## 1.0.82
1568
1569- SDK: Add request cancellation support
1570- SDK: New additionalDirectories option to search custom paths, improved slash command processing
1571- Settings: Validation prevents invalid fields in .claude/settings.json files
1572- MCP: Improve tool name consistency
1573- Bash: Fix crash when Claude tries to automatically read large files
1574
1575## 1.0.81
1576
1577- Released output styles, including new built-in educational output styles "Explanatory" and "Learning". Docs: https://code.claude.com/docs/en/output-styles
1578- Agents: Fix custom agent loading when agent files are unparsable
1579
1580## 1.0.80
1581
1582- UI improvements: Fix text contrast for custom subagent colors and spinner rendering issues
1583
1584## 1.0.77
1585
1586- Bash tool: Fix heredoc and multiline string escaping, improve stderr redirection handling
1587- SDK: Add session support and permission denial tracking
1588- Fix token limit errors in conversation summarization
1589- Opus Plan Mode: New setting in `/model` to run Opus only in plan mode, Sonnet otherwise
1590
1591## 1.0.73
1592
1593- MCP: Support multiple config files with `--mcp-config file1.json file2.json`
1594- MCP: Press Esc to cancel OAuth authentication flows
1595- Bash: Improved command validation and reduced false security warnings
1596- UI: Enhanced spinner animations and status line visual hierarchy
1597- Linux: Added support for Alpine and musl-based distributions (requires separate ripgrep installation)
1598
1599## 1.0.72
1600
1601- Ask permissions: have Claude Code always ask for confirmation to use specific tools with /permissions
1602
1603## 1.0.71
1604
1605- Background commands: (Ctrl-b) to run any Bash command in the background so Claude can keep working (great for dev servers, tailing logs, etc.)
1606- Customizable status line: add your terminal prompt to Claude Code with /statusline
1607
1608## 1.0.70
1609
1610- Performance: Optimized message rendering for better performance with large contexts
1611- Windows: Fixed native file search, ripgrep, and subagent functionality
1612- Added support for @-mentions in slash command arguments
1613
1614## 1.0.69
1615
1616- Upgraded Opus to version 4.1
1617
1618## 1.0.68
1619
1620- Fix incorrect model names being used for certain commands like `/pr-comments`
1621- Windows: improve permissions checks for allow / deny tools and project trust. This may create a new project entry in `.claude.json` - manually merge the history field if desired.
1622- Windows: improve sub-process spawning to eliminate "No such file or directory" when running commands like pnpm
1623- Enhanced /doctor command with CLAUDE.md and MCP tool context for self-serve debugging
1624- SDK: Added canUseTool callback support for tool confirmation
1625- Added `disableAllHooks` setting
1626- Improved file suggestions performance in large repos
1627
1628## 1.0.65
1629
1630- IDE: Fixed connection stability issues and error handling for diagnostics
1631- Windows: Fixed shell environment setup for users without .bashrc files
1632
1633## 1.0.64
1634
1635- Agents: Added model customization support - you can now specify which model an agent should use
1636- Agents: Fixed unintended access to the recursive agent tool
1637- Hooks: Added systemMessage field to hook JSON output for displaying warnings and context
1638- SDK: Fixed user input tracking across multi-turn conversations
1639- Added hidden files to file search and @-mention suggestions
1640
1641## 1.0.63
1642
1643- Windows: Fixed file search, @agent mentions, and custom slash commands functionality
1644
1645## 1.0.62
1646
1647- Added @-mention support with typeahead for custom agents. @<your-custom-agent> to invoke it
1648- Hooks: Added SessionStart hook for new session initialization
1649- /add-dir command now supports typeahead for directory paths
1650- Improved network connectivity check reliability
1651
1652## 1.0.61
1653
1654- Transcript mode (Ctrl+R): Changed Esc to exit transcript mode rather than interrupt
1655- Settings: Added `--settings` flag to load settings from a JSON file
1656- Settings: Fixed resolution of settings files paths that are symlinks
1657- OTEL: Fixed reporting of wrong organization after authentication changes
1658- Slash commands: Fixed permissions checking for allowed-tools with Bash
1659- IDE: Added support for pasting images in VSCode MacOS using ⌘+V
1660- IDE: Added `CLAUDE_CODE_AUTO_CONNECT_IDE=false` for disabling IDE auto-connection
1661- Added `CLAUDE_CODE_SHELL_PREFIX` for wrapping Claude and user-provided shell commands run by Claude Code
1662
1663## 1.0.60
1664
1665- You can now create custom subagents for specialized tasks! Run /agents to get started
1666
1667## 1.0.59
1668
1669- SDK: Added tool confirmation support with canUseTool callback
1670- SDK: Allow specifying env for spawned process
1671- Hooks: Exposed PermissionDecision to hooks (including "ask")
1672- Hooks: UserPromptSubmit now supports additionalContext in advanced JSON output
1673- Fixed issue where some Max users that specified Opus would still see fallback to Sonnet
1674
1675## 1.0.58
1676
1677- Added support for reading PDFs
1678- MCP: Improved server health status display in 'claude mcp list'
1679- Hooks: Added CLAUDE_PROJECT_DIR env var for hook commands
1680
1681## 1.0.57
1682
1683- Added support for specifying a model in slash commands
1684- Improved permission messages to help Claude understand allowed tools
1685- Fix: Remove trailing newlines from bash output in terminal wrapping
1686
1687## 1.0.56
1688
1689- Windows: Enabled shift+tab for mode switching on versions of Node.js that support terminal VT mode
1690- Fixes for WSL IDE detection
1691- Fix an issue causing awsRefreshHelper changes to .aws directory not to be picked up
1692
1693## 1.0.55
1694
1695- Clarified knowledge cutoff for Opus 4 and Sonnet 4 models
1696- Windows: fixed Ctrl+Z crash
1697- SDK: Added ability to capture error logging
1698- Add --system-prompt-file option to override system prompt in print mode
1699
1700## 1.0.54
1701
1702- Hooks: Added UserPromptSubmit hook and the current working directory to hook inputs
1703- Custom slash commands: Added argument-hint to frontmatter
1704- Windows: OAuth uses port 45454 and properly constructs browser URL
1705- Windows: mode switching now uses alt + m, and plan mode renders properly
1706- Shell: Switch to in-memory shell snapshot to fix file-related errors
1707
1708## 1.0.53
1709
1710- Updated @-mention file truncation from 100 lines to 2000 lines
1711- Add helper script settings for AWS token refresh: awsAuthRefresh (for foreground operations like aws sso login) and awsCredentialExport (for background operation with STS-like response).
1712
1713## 1.0.52
1714
1715- Added support for MCP server instructions
1716
1717## 1.0.51
1718
1719- Added support for native Windows (requires Git for Windows)
1720- Added support for Bedrock API keys through environment variable AWS_BEARER_TOKEN_BEDROCK
1721- Settings: /doctor can now help you identify and fix invalid setting files
1722- `--append-system-prompt` can now be used in interactive mode, not just --print/-p.
1723- Increased auto-compact warning threshold from 60% to 80%
1724- Fixed an issue with handling user directories with spaces for shell snapshots
1725- OTEL resource now includes os.type, os.version, host.arch, and wsl.version (if running on Windows Subsystem for Linux)
1726- Custom slash commands: Fixed user-level commands in subdirectories
1727- Plan mode: Fixed issue where rejected plan from sub-task would get discarded
1728
1729## 1.0.48
1730
1731- Fixed a bug in v1.0.45 where the app would sometimes freeze on launch
1732- Added progress messages to Bash tool based on the last 5 lines of command output
1733- Added expanding variables support for MCP server configuration
1734- Moved shell snapshots from /tmp to ~/.claude for more reliable Bash tool calls
1735- Improved IDE extension path handling when Claude Code runs in WSL
1736- Hooks: Added a PreCompact hook
1737- Vim mode: Added c, f/F, t/T
1738
1739## 1.0.45
1740
1741- Redesigned Search (Grep) tool with new tool input parameters and features
1742- Disabled IDE diffs for notebook files, fixing "Timeout waiting after 1000ms" error
1743- Fixed config file corruption issue by enforcing atomic writes
1744- Updated prompt input undo to Ctrl+\_ to avoid breaking existing Ctrl+U behavior, matching zsh's undo shortcut
1745- Stop Hooks: Fixed transcript path after /clear and fixed triggering when loop ends with tool call
1746- Custom slash commands: Restored namespacing in command names based on subdirectories. For example, .claude/commands/frontend/component.md is now /frontend:component, not /component.
1747
1748## 1.0.44
1749
1750- New /export command lets you quickly export a conversation for sharing
1751- MCP: resource_link tool results are now supported
1752- MCP: tool annotations and tool titles now display in /mcp view
1753- Changed Ctrl+Z to suspend Claude Code. Resume by running `fg`. Prompt input undo is now Ctrl+U.
1754
1755## 1.0.43
1756
1757- Fixed a bug where the theme selector was saving excessively
1758- Hooks: Added EPIPE system error handling
1759
1760## 1.0.42
1761
1762- Added tilde (`~`) expansion support to `/add-dir` command
1763
1764## 1.0.41
1765
1766- Hooks: Split Stop hook triggering into Stop and SubagentStop
1767- Hooks: Enabled optional timeout configuration for each command
1768- Hooks: Added "hook_event_name" to hook input
1769- Fixed a bug where MCP tools would display twice in tool list
1770- New tool parameters JSON for Bash tool in `tool_decision` event
1771
1772## 1.0.40
1773
1774- Fixed a bug causing API connection errors with UNABLE_TO_GET_ISSUER_CERT_LOCALLY if `NODE_EXTRA_CA_CERTS` was set
1775
1776## 1.0.39
1777
1778- New Active Time metric in OpenTelemetry logging
1779
1780## 1.0.38
1781
1782- Released hooks. Special thanks to community input in https://github.com/anthropics/claude-code/issues/712. Docs: https://code.claude.com/docs/en/hooks
1783
1784## 1.0.37
1785
1786- Remove ability to set `Proxy-Authorization` header via ANTHROPIC_AUTH_TOKEN or apiKeyHelper
1787
1788## 1.0.36
1789
1790- Web search now takes today's date into context
1791- Fixed a bug where stdio MCP servers were not terminating properly on exit
1792
1793## 1.0.35
1794
1795- Added support for MCP OAuth Authorization Server discovery
1796
1797## 1.0.34
1798
1799- Fixed a memory leak causing a MaxListenersExceededWarning message to appear
1800
1801## 1.0.33
1802
1803- Improved logging functionality with session ID support
1804- Added prompt input undo functionality (Ctrl+Z and vim 'u' command)
1805- Improvements to plan mode
1806
1807## 1.0.32
1808
1809- Updated loopback config for litellm
1810- Added forceLoginMethod setting to bypass login selection screen
1811
1812## 1.0.31
1813
1814- Fixed a bug where ~/.claude.json would get reset when file contained invalid JSON
1815
1816## 1.0.30
1817
1818- Custom slash commands: Run bash output, @-mention files, enable thinking with thinking keywords
1819- Improved file path autocomplete with filename matching
1820- Added timestamps in Ctrl-r mode and fixed Ctrl-c handling
1821- Enhanced jq regex support for complex filters with pipes and select
1822
1823## 1.0.29
1824
1825- Improved CJK character support in cursor navigation and rendering
1826
1827## 1.0.28
1828
1829- Slash commands: Fix selector display during history navigation
1830- Resizes images before upload to prevent API size limit errors
1831- Added XDG_CONFIG_HOME support to configuration directory
1832- Performance optimizations for memory usage
1833- New attributes (terminal.type, language) in OpenTelemetry logging
1834
1835## 1.0.27
1836
1837- Streamable HTTP MCP servers are now supported
1838- Remote MCP servers (SSE and HTTP) now support OAuth
1839- MCP resources can now be @-mentioned
1840- /resume slash command to switch conversations within Claude Code
1841
1842## 1.0.25
1843
1844- Slash commands: moved "project" and "user" prefixes to descriptions
1845- Slash commands: improved reliability for command discovery
1846- Improved support for Ghostty
1847- Improved web search reliability
1848
1849## 1.0.24
1850
1851- Improved /mcp output
1852- Fixed a bug where settings arrays got overwritten instead of merged
1853
1854## 1.0.23
1855
1856- Released TypeScript SDK: import @anthropic-ai/claude-code to get started
1857- Released Python SDK: pip install claude-code-sdk to get started
1858
1859## 1.0.22
1860
1861- SDK: Renamed `total_cost` to `total_cost_usd`
1862
1863## 1.0.21
1864
1865- Improved editing of files with tab-based indentation
1866- Fix for tool_use without matching tool_result errors
1867- Fixed a bug where stdio MCP server processes would linger after quitting Claude Code
1868
1869## 1.0.18
1870
1871- Added --add-dir CLI argument for specifying additional working directories
1872- Added streaming input support without require -p flag
1873- Improved startup performance and session storage performance
1874- Added CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR environment variable to freeze working directory for bash commands
1875- Added detailed MCP server tools display (/mcp)
1876- MCP authentication and permission improvements
1877- Added auto-reconnection for MCP SSE connections on disconnect
1878- Fixed issue where pasted content was lost when dialogs appeared
1879
1880## 1.0.17
1881
1882- We now emit messages from sub-tasks in -p mode (look for the parent_tool_use_id property)
1883- Fixed crashes when the VS Code diff tool is invoked multiple times quickly
1884- MCP server list UI improvements
1885- Update Claude Code process title to display "claude" instead of "node"
1886
1887## 1.0.11
1888
1889- Claude Code can now also be used with a Claude Pro subscription
1890- Added /upgrade for smoother switching to Claude Max plans
1891- Improved UI for authentication from API keys and Bedrock/Vertex/external auth tokens
1892- Improved shell configuration error handling
1893- Improved todo list handling during compaction
1894
1895## 1.0.10
1896
1897- Added markdown table support
1898- Improved streaming performance
1899
1900## 1.0.8
1901
1902- Fixed Vertex AI region fallback when using CLOUD_ML_REGION
1903- Increased default otel interval from 1s -> 5s
1904- Fixed edge cases where MCP_TIMEOUT and MCP_TOOL_TIMEOUT weren't being respected
1905- Fixed a regression where search tools unnecessarily asked for permissions
1906- Added support for triggering thinking non-English languages
1907- Improved compacting UI
1908
1909## 1.0.7
1910
1911- Renamed /allowed-tools -> /permissions
1912- Migrated allowedTools and ignorePatterns from .claude.json -> settings.json
1913- Deprecated claude config commands in favor of editing settings.json
1914- Fixed a bug where --dangerously-skip-permissions sometimes didn't work in --print mode
1915- Improved error handling for /install-github-app
1916- Bugfixes, UI polish, and tool reliability improvements
1917
1918## 1.0.6
1919
1920- Improved edit reliability for tab-indented files
1921- Respect CLAUDE_CONFIG_DIR everywhere
1922- Reduced unnecessary tool permission prompts
1923- Added support for symlinks in @file typeahead
1924- Bugfixes, UI polish, and tool reliability improvements
1925
1926## 1.0.4
1927
1928- Fixed a bug where MCP tool errors weren't being parsed correctly
1929
1930## 1.0.1
1931
1932- Added `DISABLE_INTERLEAVED_THINKING` to give users the option to opt out of interleaved thinking.
1933- Improved model references to show provider-specific names (Sonnet 3.7 for Bedrock, Sonnet 4 for Console)
1934- Updated documentation links and OAuth process descriptions
1935
1936## 1.0.0
1937
1938- Claude Code is now generally available
1939- Introducing Sonnet 4 and Opus 4 models
1940
1941## 0.2.125
1942
1943- Breaking change: Bedrock ARN passed to `ANTHROPIC_MODEL` or `ANTHROPIC_SMALL_FAST_MODEL` should no longer contain an escaped slash (specify `/` instead of `%2F`)
1944- Removed `DEBUG=true` in favor of `ANTHROPIC_LOG=debug`, to log all requests
1945
1946## 0.2.117
1947
1948- Breaking change: --print JSON output now returns nested message objects, for forwards-compatibility as we introduce new metadata fields
1949- Introduced settings.cleanupPeriodDays
1950- Introduced CLAUDE_CODE_API_KEY_HELPER_TTL_MS env var
1951- Introduced --debug mode
1952
1953## 0.2.108
1954
1955- You can now send messages to Claude while it works to steer Claude in real-time
1956- Introduced BASH_DEFAULT_TIMEOUT_MS and BASH_MAX_TIMEOUT_MS env vars
1957- Fixed a bug where thinking was not working in -p mode
1958- Fixed a regression in /cost reporting
1959- Deprecated MCP wizard interface in favor of other MCP commands
1960- Lots of other bugfixes and improvements
1961
1962## 0.2.107
1963
1964- CLAUDE.md files can now import other files. Add @path/to/file.md to ./CLAUDE.md to load additional files on launch
1965
1966## 0.2.106
1967
1968- MCP SSE server configs can now specify custom headers
1969- Fixed a bug where MCP permission prompt didn't always show correctly
1970
1971## 0.2.105
1972
1973- Claude can now search the web
1974- Moved system & account status to /status
1975- Added word movement keybindings for Vim
1976- Improved latency for startup, todo tool, and file edits
1977
1978## 0.2.102
1979
1980- Improved thinking triggering reliability
1981- Improved @mention reliability for images and folders
1982- You can now paste multiple large chunks into one prompt
1983
1984## 0.2.100
1985
1986- Fixed a crash caused by a stack overflow error
1987- Made db storage optional; missing db support disables --continue and --resume
1988
1989## 0.2.98
1990
1991- Fixed an issue where auto-compact was running twice
1992
1993## 0.2.96
1994
1995- Claude Code can now also be used with a Claude Max subscription (https://claude.ai/upgrade)
1996
1997## 0.2.93
1998
1999- Resume conversations from where you left off from with "claude --continue" and "claude --resume"
2000- Claude now has access to a Todo list that helps it stay on track and be more organized
2001
2002## 0.2.82
2003
2004- Added support for --disallowedTools
2005- Renamed tools for consistency: LSTool -> LS, View -> Read, etc.
2006
2007## 0.2.75
2008
2009- Hit Enter to queue up additional messages while Claude is working
2010- Drag in or copy/paste image files directly into the prompt
2011- @-mention files to directly add them to context
2012- Run one-off MCP servers with `claude --mcp-config <path-to-file>`
2013- Improved performance for filename auto-complete
2014
2015## 0.2.74
2016
2017- Added support for refreshing dynamically generated API keys (via apiKeyHelper), with a 5 minute TTL
2018- Task tool can now perform writes and run bash commands
2019
2020## 0.2.72
2021
2022- Updated spinner to indicate tokens loaded and tool usage
2023
2024## 0.2.70
2025
2026- Network commands like curl are now available for Claude to use
2027- Claude can now run multiple web queries in parallel
2028- Pressing ESC once immediately interrupts Claude in Auto-accept mode
2029
2030## 0.2.69
2031
2032- Fixed UI glitches with improved Select component behavior
2033- Enhanced terminal output display with better text truncation logic
2034
2035## 0.2.67
2036
2037- Shared project permission rules can be saved in .claude/settings.json
2038
2039## 0.2.66
2040
2041- Print mode (-p) now supports streaming output via --output-format=stream-json
2042- Fixed issue where pasting could trigger memory or bash mode unexpectedly
2043
2044## 0.2.63
2045
2046- Fixed an issue where MCP tools were loaded twice, which caused tool call errors
2047
2048## 0.2.61
2049
2050- Navigate menus with vim-style keys (j/k) or bash/emacs shortcuts (Ctrl+n/p) for faster interaction
2051- Enhanced image detection for more reliable clipboard paste functionality
2052- Fixed an issue where ESC key could crash the conversation history selector
2053
2054## 0.2.59
2055
2056- Copy+paste images directly into your prompt
2057- Improved progress indicators for bash and fetch tools
2058- Bugfixes for non-interactive mode (-p)
2059
2060## 0.2.54
2061
2062- Quickly add to Memory by starting your message with '#'
2063- Press ctrl+r to see full output for long tool results
2064- Added support for MCP SSE transport
2065
2066## 0.2.53
2067
2068- New web fetch tool lets Claude view URLs that you paste in
2069- Fixed a bug with JPEG detection
2070
2071## 0.2.50
2072
2073- New MCP "project" scope now allows you to add MCP servers to .mcp.json files and commit them to your repository
2074
2075## 0.2.49
2076
2077- Previous MCP server scopes have been renamed: previous "project" scope is now "local" and "global" scope is now "user"
2078
2079## 0.2.47
2080
2081- Press Tab to auto-complete file and folder names
2082- Press Shift + Tab to toggle auto-accept for file edits
2083- Automatic conversation compaction for infinite conversation length (toggle with /config)
2084
2085## 0.2.44
2086
2087- Ask Claude to make a plan with thinking mode: just say 'think' or 'think harder' or even 'ultrathink'
2088
2089## 0.2.41
2090
2091- MCP server startup timeout can now be configured via MCP_TIMEOUT environment variable
2092- MCP server startup no longer blocks the app from starting up
2093
2094## 0.2.37
2095
2096- New /release-notes command lets you view release notes at any time
2097- `claude config add/remove` commands now accept multiple values separated by commas or spaces
2098
2099## 0.2.36
2100
2101- Import MCP servers from Claude Desktop with `claude mcp add-from-claude-desktop`
2102- Add MCP servers as JSON strings with `claude mcp add-json <n> <json>`
2103
2104## 0.2.34
2105
2106- Vim bindings for text input - enable with /vim or /config
2107
2108## 0.2.32
2109
2110- Interactive MCP setup wizard: Run "claude mcp add" to add MCP servers with a step-by-step interface
2111- Fix for some PersistentShell issues
2112
2113## 0.2.31
2114
2115- Custom slash commands: Markdown files in .claude/commands/ directories now appear as custom slash commands to insert prompts into your conversation
2116- MCP debug mode: Run with --mcp-debug flag to get more information about MCP server errors
2117
2118## 0.2.30
2119
2120- Added ANSI color theme for better terminal compatibility
2121- Fixed issue where slash command arguments weren't being sent properly
2122- (Mac-only) API keys are now stored in macOS Keychain
2123
2124## 0.2.26
2125
2126- New /approved-tools command for managing tool permissions
2127- Word-level diff display for improved code readability
2128- Fuzzy matching for slash commands
2129
2130## 0.2.21
2131
2132- Fuzzy matching for /commands