Enterprise channel 1.0.79

2026-11-08 · enterprise · enterprise pointer is 1.0.79

1.0.79 — 2026-11-08

Features

/rewind now works in place for paged sessions without reloading every turn.
Link cards now appear by default for links that cannot be opened.
/vim-mode all enables full Vim editing inside the agent prompt.
Auto-review mode now tints the entire prompt bar blue, matching plan and ask modes.
/doctor now warns when tmux holds Esc for more than 50 ms.
Vim Normal mode now appears as a section in the keyboard shortcuts cheatsheet when enabled.
End-of-turn marker now shows how many background commands are still running.

Bug Fixes

Copy buttons under code blocks remain clickable while slash or other menus are open.
Deleted sessions no longer appear in resume lists or title searches.
Delete error messages now show their full text instead of being cut off.
Text remains visible when running through an 8-colour tmux session.
Authentication error messages during sign-in no longer repeat the underlying cause.
Ctrl+E in minimal mode now moves the cursor to the end of line when editing a draft.
macOS documentation now recommends Ctrl+V to paste screenshots.
Mermaid diagrams now correctly break lines on \n and <br> instead of printing the literal characters.
MCP sign-in URL copy feedback now correctly indicates verification status.
Double-clicking a view close no longer accidentally stops a background task or subagent.
Tab completion now works immediately after startup even before skills load.
Saved feedback drafts now open correctly from the full session view.
Markdown links containing LaTeX now render correctly without entering math mode.
Ctrl+Y now toggles the prompt queue when running in WSL inside Windows Terminal.
Editing a queued prompt now shows an "editing" label and dims the row.
Cancel-turn prompt no longer crashes the TUI on very short terminals.

source

Alpha channel 1.0.79

2026-11-08 · alpha · alpha pointer is 1.0.79

1.0.79 — 2026-11-08

Features

/rewind now works in place for paged sessions without reloading every turn.
Link cards now appear by default for links that cannot be opened.
/vim-mode all enables full Vim editing inside the agent prompt.
Auto-review mode now tints the entire prompt bar blue, matching plan and ask modes.
/doctor now warns when tmux holds Esc for more than 50 ms.
Vim Normal mode now appears as a section in the keyboard shortcuts cheatsheet when enabled.
End-of-turn marker now shows how many background commands are still running.

Bug Fixes

Copy buttons under code blocks remain clickable while slash or other menus are open.
Deleted sessions no longer appear in resume lists or title searches.
Delete error messages now show their full text instead of being cut off.
Text remains visible when running through an 8-colour tmux session.
Authentication error messages during sign-in no longer repeat the underlying cause.
Ctrl+E in minimal mode now moves the cursor to the end of line when editing a draft.
macOS documentation now recommends Ctrl+V to paste screenshots.
Mermaid diagrams now correctly break lines on \n and <br> instead of printing the literal characters.
MCP sign-in URL copy feedback now correctly indicates verification status.
Double-clicking a view close no longer accidentally stops a background task or subagent.
Tab completion now works immediately after startup even before skills load.
Saved feedback drafts now open correctly from the full session view.
Markdown links containing LaTeX now render correctly without entering math mode.
Ctrl+Y now toggles the prompt queue when running in WSL inside Windows Terminal.
Editing a queued prompt now shows an "editing" label and dims the row.
Cancel-turn prompt no longer crashes the TUI on very short terminals.

source

Alpha channel 1.0.78

2026-11-06 · alpha · alpha pointer is 1.0.78

1.0.78 — 2026-11-06

Features

@ file chips can now be edited in place; typed @~/ paths resolve to the home folder.
Dragging in the transcript now selects and copies colored text across message blocks instead of whole blocks.
@ mentions now support :N line suffixes and @!hidden files.

Bug Fixes

Always-approve help text now correctly says the setting is saved to config.
Dropped-image notice no longer appears inside the user's own message in transcripts or history.
grok update --channel now gives a clear error and steps when bin/grok was installed by npm.
grok update and channel switches now work correctly for Homebrew cask installs and show the right commands.

Performance

grok export of sessions containing large shell output is now faster and uses less memory.

source

Stable channel 1.0.77

2026-11-05 · stable · stable pointer is 1.0.77

1.0.77 — 2026-11-05

Features

New [ui] kitty_keyboard = "query" setting lets Shift+Enter insert newlines in unidentified terminals.

Bug Fixes

Quote copy now preserves & in links and double-clicks no longer fold thoughts when missing text.
MCP server connections no longer restart unnecessarily on session reloads or updates.
Ctrl+L now works on more screens and copy hints match what actually copies to the clipboard.
Context usage shown in the footer no longer drops to zero or shows stale values after a rewind.
Rows in the /resume list now expand and collapse the same way they do inside a session.

source

Enterprise channel 1.0.77

2026-11-05 · enterprise · enterprise pointer is 1.0.77

1.0.77 — 2026-11-05

Features

New [ui] kitty_keyboard = "query" setting lets Shift+Enter insert newlines in unidentified terminals.

Bug Fixes

Quote copy now preserves & in links and double-clicks no longer fold thoughts when missing text.
MCP server connections no longer restart unnecessarily on session reloads or updates.
Ctrl+L now works on more screens and copy hints match what actually copies to the clipboard.
Context usage shown in the footer no longer drops to zero or shows stale values after a rewind.
Rows in the /resume list now expand and collapse the same way they do inside a session.

source

Alpha channel 1.0.77

2026-11-05 · alpha · alpha pointer is 1.0.77

1.0.77 — 2026-11-05

Features

New [ui] kitty_keyboard = "query" setting lets Shift+Enter insert newlines in unidentified terminals.

Bug Fixes

Quote copy now preserves & in links and double-clicks no longer fold thoughts when missing text.
MCP server connections no longer restart unnecessarily on session reloads or updates.
Ctrl+L now works on more screens and copy hints match what actually copies to the clipboard.
Context usage shown in the footer no longer drops to zero or shows stale values after a rewind.
Rows in the /resume list now expand and collapse the same way they do inside a session.

source

Alpha channel 1.0.75

2026-11-04 · alpha · alpha pointer is 1.0.75

1.0.75 — 2026-11-04

Features

Permission rules that don't load, such as a rule that doesn't parse or a .claude/settings.json that isn't valid JSON, show a warning when a session opens instead of being skipped silently.
Cmd+Up/Cmd+Down and Ctrl+Home/Ctrl+End move the cursor to the start or end of the whole prompt, and with Shift they select to there.
Send now shows as Enter Enter in the footer where Ctrl+Enter arrives as plain Enter, such as inside tmux with its default settings.
Esc on a /btw side question that hasn't been answered yet cancels it, instead of letting it run on and reopening the panel when the answer arrives.
Clicking a link or file that can't open, such as over SSH with no display, shows why, for example "Can't open links over SSH".
Left on an empty prompt opens the dashboard from any session or the welcome screen, not only from a session you opened from the dashboard.
The /btw question in the panel's title can be selected with the mouse and is copied to the clipboard when you release, like the answer text.
Workflow status lines say how many agents are still running instead of how many exist.
Ctrl+L in the scrollback adds the selected message or highlighted text to the prompt as a quote; open Extensions with /plugins instead.

Bug Fixes

Notes above a queued prompt, such as a model switch result or a ready worktree, stay on screen when the prompt starts instead of scrolling out of view.
Stop buttons in the dock stop a background run while its approval card is open, instead of doing nothing until you answer the card.
Rejecting a change keeps edits you saved to the file yourself, even when Grok missed the save, instead of overwriting them.
grok --resume <id> opens the session in the folder it was recorded in, wherever you launch Grok from, instead of refusing.
Windows: on older consoles such as plain cmd.exe and Windows PowerShell windows, the ▶, ⏸ and ◌ marks in todo, workflow and list rows show as characters the console can draw instead of empty boxes.
Generated image and video paths with wide characters such as Chinese, Japanese or emoji fit their row and keep the file name, and Windows paths keep it too.
The rewind confirmation no longer crashes Grok on a short terminal or when you shrink the window while it's open.
Footer and status bars in narrow terminals drop or shorten the items that don't fit, instead of painting them over each other or cutting buttons off.
Cut-off Fetch, Web Search and MCP tool output ends with the same … +N lines · <key> to view all row as code cells, naming your viewer key, and one hidden line reads +1 line.
16-color terminals show the paste label and tool accents in the muted gray other secondary text uses, and the paste label's brackets no longer vanish on the light theme.
A question card's Enter:submit button, the dashboard session popup's [✗] and collapsed workflow rows highlight under the mouse pointer like other clickable controls.
Emoji and keycap symbols such as ⚠️ or #️⃣ are measured at the width the terminal draws them, so rows no longer run past their box or lose their ….
/export to a named pipe that nothing is reading fails right away instead of freezing Grok until it is killed.
Typing a digit or symbol while the scrollback has focus, with vim keys off, moves focus to the prompt and types it there, as letters already do; ? still opens the command palette.
grok sessions delete no longer fails with "Directory not empty" when another process is still writing into the session's folder.
Dashboard rows stay in Working while a session's background subagents or workflow runs are going, and stopping the row ends them.
Code blocks inside quotes render every line under the quote's bars, quote copies turn reference links into inline links, and lines after multi-line math are numbered correctly.
Disabling a project's hook in /hooks applies only to that repository, instead of also skipping hooks with the same name in other repositories.
Turning a skill on or off in the Skills tab saves only [skills].disabled to config.toml, so an organization's managed skill folders and ignore list keep applying.
/btw starts a side question only when it begins the message; a /btw later in the text goes to the main conversation as typed.

source

Alpha channel 1.0.74

2026-11-02 · alpha · alpha pointer is 1.0.74

1.0.74 — 2026-11-02

Features

Messages between subagents appear as left-aligned rows under the bullet that name the sender and recipient, replacing the centered rows.
Shift+Enter inserts a newline in VS Code and other xterm.js-based terminals that use the Kitty keyboard protocol.
/mcps: when a browser can't open an MCP server's sign-in URL, the URL shows in a panel you can copy with c or the [copy the url] button.
[Copy] button in a finished /btw panel and its folded transcript row copies the whole answer, even when only part of it is visible.
Shell approval cards say when Auto-review blocked a command or couldn't review it, and Ctrl+F shows the reviewer's explanation.
! mode Tab completion offers commands installed as symlinks, such as Homebrew, npm, rustup and Debian alternatives binaries.
Quotes in replies get a [Copy] button that copies the whole quote as Markdown without its > markers.
Named subagent links open on a plain click, so they work over SSH from macOS where a modified click can't reach Grok.
@ paths outside the workspace: an @ path that starts with /, ~/ or ../ lists files from that folder; set [ui] at_picker_outside_workspace = false to turn it off.
Picked @ files in the prompt can be edited in place: Backspace at the end of one reopens the picker on its path, and Left or Right puts the cursor inside it to change the path.

Bug Fixes

Built-in skills, agents and workflows are refreshed when their local cache's manifest is missing or corrupt, instead of staying outdated or failing every sync.
Ctrl+Y on a question card no longer closes it without an answer; while you type an answer, it pastes back text cut with Ctrl+K.
/new tabs show the context window you just switched to before the new session replies.
Resumed sessions show a background command's completion in the turn that ran it, with its real run time instead of about 0.0s.
grok update runs when GROK_SANDBOX, --sandbox or config sets a sandbox profile, unless Grok is already running confined.
/btw on a very short terminal no longer swallows arrow keys, Page Up/Down or clicks while its panel is too small to draw.
A code block's [Copy] in the transcript copies while a queued prompt is open for editing, instead of copying nothing or closing the edit.
grok logout stays signed out when another running Grok refreshes its sign-in, instead of signing you back in on the next launch.
Opening links and files from the TUI no longer leaves a zombie process behind for each click.
A failed session delete shows the "Couldn't delete session" prefix once, so the reason is no longer cut off.
Windows: Grok exits fully after a startup error, a /minimal or /fullscreen switch, or a signal, instead of sometimes hanging.
Grok running confined in a sandbox never starts an update on its own and tells you to run grok update instead.
Quitting to install an update with Ctrl+U on the welcome screen exits within the usual exit deadline instead of sometimes hanging after "Update installed".
grok update --force-reinstall, or grok update --version with the installed version, keeps your working grok on Linux and macOS when the download fails its checks, instead of deleting it.
Windows installers check that the downloaded build runs before replacing grok.exe and agent.exe, and keep the existing install when it doesn't, such as when a proxy serves a blocked-download page.
Command-line arguments that aren't valid UTF-8 no longer crash grok at startup: paths such as --cwd accept them, and text such as -p gets a usage error.
Installers keep the installed grok when a proxy or cache returns an empty download, instead of replacing it with an empty file and reporting success.
Environment variables with non-UTF-8 bytes no longer stop Grok from starting.
Workflow run lines in the transcript end with what the run returned instead of a bare colon, /tasks shows the time it was taken, and polling an unknown task id leaves a row with its error.
Headless runs stopped by Ctrl-C or SIGTERM end their json, streaming-json or streaming-messages-json output with a closing record that keeps the run's outcome, instead of cutting off the stream.
macOS: the TUI no longer freezes when it runs inside a sandbox that blocks the window server.
grok -p --resume shows why a session couldn't load instead of always "Session does not exist", and with --output-format json a resume, continue or fork that fails before the session opens prints one JSON error.
Sessions with an unrecognized reasoning effort or unreadable bytes in their history resume instead of failing: the session runs at the model's default effort and says so once, and the unreadable lines are left out of the transcript.
Windows: on older consoles such as plain cmd.exe and Windows PowerShell windows, status marks and selection arrows in workflow views, the dashboard, and the feedback and extensions dialogs show as >, •, √ or x instead of empty boxes.
File paths with spaces in inline code, such as ~/Library/Application Support/..., link to the whole path when it exists instead of stopping at the first space.
Feedback that can't be sent leads with the outcome, such as "Couldn't send feedback. Saved to Drafts.", then gives the reason and says to retry from /feedback.
Dialog tab labels, shortcuts and settings rows that don't fit end in … instead of being cut off.
Prompt history search highlights matched letters in the same color as other fuzzy lists, and the counts on the / and @ menu borders are dimmed like other border captions.
Dashboard: the peek header reads Responding while a reply streams, and the Esc footer hint reads new agent like the other lowercase hints.
Custom agent files saved with a UTF-8 byte order mark, as some Windows editors write them, load instead of being skipped.
Running markers in workflow views and MCP servers that are still starting use the theme's running color, like other progress indicators.
Minimal mode: "Sign-in failed" shows in the error color instead of the warning color.
Context window: a larger window picked with /context-window or set with context_window in a [model."<id>"] table stays in effect, so compaction no longer starts early after the session sits idle, after a failed request such as a rate limit, or until you restart.

Performance

Long sessions use less memory: finished Markdown blocks far off screen release their rendered layout, and search still finds their text.
Typing no longer lags when ffmpeg isn't installed and PATH holds slow directories, such as Windows drives under WSL.
Long pastes that arrive as keystrokes, such as pastes on native Windows or in terminals without bracketed paste, no longer freeze the prompt for seconds while the plan-mode tip is on.
Pasting long single-line text such as logs, JSON or base64 into the prompt no longer freezes the TUI for seconds.
File watching on Linux no longer keeps a CPU core busy for about a minute after startup in large repositories.
Lower CPU use while subagents run: the TUI repaints only for animations on screen, about 7 times a second instead of 30 when no running row is visible.
grok --resume <id> opens much faster on large session stores, reading only the resumed session instead of every session.

source

Enterprise channel 1.0.73

2026-10-28 · enterprise · enterprise pointer is 1.0.73

1.0.73 — 2026-10-28

Features

Page Up/Page Down, Home and End move the selection in /skills and other pickers.
Messages between two subagents appear in the main session's transcript as centered rows once delivered, such as find callers → scan src/.
Drag-selecting text in a /btw answer scrolls the answer while you hold the pointer on or past the panel's edge.
Automatic updates stop retrying a new version that fails to start on this machine until a newer version is out, and each launch says to run grok update to see why.
Gruvbox dark theme (gruvbox-dark-hard), selectable with /theme.
Start in multiline in /settings ([ui] multiline_default) starts new sessions and the dashboard in multiline mode, where Enter inserts a newline.

Bug Fixes

Windows: quitting Grok ends grok.exe once the terminal is restored, so the shell no longer sometimes waits with no prompt.
Mouse wheel in a /btw answer reaches the last lines on short terminals.
Several Send nows in a row no longer label the turns they stop "Turn cancelled by user" or jump the transcript view.
--restore-code: when the code restore fails, your uncommitted and untracked files stay in the working tree instead of in a stash.
Windows: shortcuts typed with Shift-held symbols or AZERTY digit keys work.
Queued images show as green chips in the queue pane, as in the prompt.
Typing while the transcript has focus moves any character, including non-ASCII text, to the prompt, and leaves a pending card or plan approval as it is.
Worktree creation notes stay on screen when a queued prompt starts.
Session load errors show the whole reason instead of cutting out the middle of a long message.
Lines with | that are not a Markdown table wrap like other text instead of being cut off at the terminal edge.
Transcript selection stays on the same row when a row above it is removed.
Kitty graphics protocol: images Grok hides are freed in the terminal, so they no longer use its memory.
--restore-code restores its own stash even when another worktree changes the stash list meanwhile, and keeps your changes stashed when a post-checkout hook fails after the checkout.

Performance

Startup no longer waits about 100 ms for a sign-in file when you use an API key or have not signed in yet.

source

Alpha channel 1.0.73

2026-10-28 · alpha · alpha pointer is 1.0.73

1.0.73 — 2026-10-28

Features

Page Up/Page Down, Home and End move the selection in /skills and other pickers.
Messages between two subagents appear in the main session's transcript as centered rows once delivered, such as find callers → scan src/.
Drag-selecting text in a /btw answer scrolls the answer while you hold the pointer on or past the panel's edge.
Automatic updates stop retrying a new version that fails to start on this machine until a newer version is out, and each launch says to run grok update to see why.
Gruvbox dark theme (gruvbox-dark-hard), selectable with /theme.
Start in multiline in /settings ([ui] multiline_default) starts new sessions and the dashboard in multiline mode, where Enter inserts a newline.

Bug Fixes

Windows: quitting Grok ends grok.exe once the terminal is restored, so the shell no longer sometimes waits with no prompt.
Mouse wheel in a /btw answer reaches the last lines on short terminals.
Several Send nows in a row no longer label the turns they stop "Turn cancelled by user" or jump the transcript view.
--restore-code: when the code restore fails, your uncommitted and untracked files stay in the working tree instead of in a stash.
Windows: shortcuts typed with Shift-held symbols or AZERTY digit keys work.
Queued images show as green chips in the queue pane, as in the prompt.
Typing while the transcript has focus moves any character, including non-ASCII text, to the prompt, and leaves a pending card or plan approval as it is.
Worktree creation notes stay on screen when a queued prompt starts.
Session load errors show the whole reason instead of cutting out the middle of a long message.
Lines with | that are not a Markdown table wrap like other text instead of being cut off at the terminal edge.
Transcript selection stays on the same row when a row above it is removed.
Kitty graphics protocol: images Grok hides are freed in the terminal, so they no longer use its memory.
--restore-code restores its own stash even when another worktree changes the stash list meanwhile, and keeps your changes stashed when a post-checkout hook fails after the checkout.

Performance

Startup no longer waits about 100 ms for a sign-in file when you use an API key or have not signed in yet.

source

Alpha channel 1.0.72

2026-10-27 · alpha · alpha pointer is 1.0.72

1.0.72 — 2026-10-27

Features

Relative file paths in the agent's output, such as src/app/main.rs, open on click when the file exists under the session's folder, as absolute paths already do.
Subagent and background task ids mentioned in the transcript open that subagent's view or the task's output on click.
Pull-request numbers such as #123 in the agent's replies open that pull request on the session repository's forge.
Subagents named in a reply show as the subagent's label instead of its raw id once the reply finishes, as a link that opens its transcript.
/workflow runs detail view: → or a click focuses a phase's agent list, ↑/↓ pick an agent, Enter opens it, and ← returns to the phases.
Feedback draft rows show on their own instead of inside a group of tool rows, and opening one shows the /feedback Drafts tab with that draft selected.

Bug Fixes

Warp: clicking a URL that wraps across rows opens the whole URL instead of only its first row.
/delete confirmation starts on Cancel, so pressing Enter no longer deletes the session; press 1 to delete.
Max thoughts width (/settings, [ui].max_thoughts_width) sets how wide thinking blocks wrap; it previously had no effect.
Dashboard shows the new folder on the dispatch box's top border right after a Ctrl+L or /cd pick.
Custom models on the Responses API with a non-xAI base_url no longer fail every turn with an HTTP 400 schema error for use_tool.
/feedback Drafts tab keeps the selected draft when the list reloads, such as after leaving Drafts and coming back.
MCP tools built with FastMCP or the official MCP SDK send their result to the model once instead of twice, so each call uses about half the context.
Plan review notes accept every key, so a note that starts with a no longer approves the plan.
MCP servers over stdio no longer lose part of a long reply line, such as a large tool list, when a read is interrupted.

source

Alpha channel 1.0.71

2026-10-26 · alpha · alpha pointer is 1.0.71

1.0.71 — 2026-10-26

Features

ui.single_ctrl_c_quit = true lets a single Ctrl+C quit when no draft or unfinished work would be lost. It is also under Editor in /settings; the default still asks for a second press.
Image chips in the prompt use the same green as sent images in the transcript.
Hovering an image chip in the transcript shows the image that was sent, or why it was not attached.
Hooks accept "failClosed": true: when such a PreToolUse or UserPromptSubmit hook times out or crashes, the tool call is denied or the prompt is blocked instead of going ahead.
Vim editing in the prompt (experimental): set [keys] vim = "all" in config.toml for Insert and Normal modes with vim motions, operators, counts and undo. The prompt's info line shows NORMAL or INSERT.
Inline images display in Noctty on Windows.
Answers to the agent's questions show one row per question and answer, and end in … only when the text is cut.
Messages from other agents appear as compact centered rows, and consecutive messages fold into one row.
Messages the agent sends to subagents appear as the same centered rows (you → <subagent>) and fold together with the messages it receives.
Prompt vim cursor is a block in Normal mode and a bar in Insert mode; ui.cursor_blink sets whether they blink.
Minimal mode starts its info row with the prompt's vim mode (NORMAL or INSERT), and NORMAL is shown in bold.

Bug Fixes

Image chips in your messages stay whole on one row when the message wraps.
Images the server rejects are removed from the conversation, so the notice that images were left out no longer repeats, and grows, on every later turn.
Feedback draft rows show the draft's state: drafted, declined, cancelled or failed.
A slash menu closed with Esc no longer reopens by itself when the command list refreshes, such as right after startup.
Turn failed rows no longer repeat a generic "Request failed:" before the error.

source

Stable channel 1.0.70

2026-10-23 · stable · stable pointer is 1.0.70

1.0.70 — 2026-10-23

Features

Answered questions now have a blank line between them, except in compact mode (/compact-mode).
GROK_DISABLE_EXPERIMENTS turns off every [[experiments]] entry in your configuration; [experiments] enabled = false in any configuration file does the same.

Bug Fixes

/plugins now counts only the plugin skills that load, and says how many were skipped because they link outside the plugin folder.

source

Alpha channel 1.0.70

2026-10-23 · alpha · alpha pointer is 1.0.70

1.0.70 — 2026-10-23

Features

Answered questions now have a blank line between them, except in compact mode (/compact-mode).
GROK_DISABLE_EXPERIMENTS turns off every [[experiments]] entry in your configuration; [experiments] enabled = false in any configuration file does the same.

Bug Fixes

/plugins now counts only the plugin skills that load, and says how many were skipped because they link outside the plugin folder.

source

Enterprise channel 1.0.69

2026-10-23 · enterprise · enterprise pointer is 1.0.69

1.0.69 — 2026-10-23

Features

Dashboard directory grouping now shows a header with the folder path and agent count above each folder's agents.
Dashboard location (branch and folder) now sits on the dispatch box's top border instead of its own header row.
/resume on the dashboard now lists sessions from every folder, with the current folder and its checkouts first.
Welcome menu now lists Resume session (Ctrl+R) above New worktree (Ctrl+W).

Bug Fixes

[ui] simple_mode is retired and no longer has any effect; sessions always open with the prompt focused.
grok doctor and /doctor now point out a leftover [ui] vim_mode key (now [keys] vim) or the retired [ui] simple_mode key in your config.

Performance

Quitting Grok returns you to your shell sooner because it no longer waits out fixed intervals.

source

Alpha channel 1.0.69

2026-10-23 · alpha · alpha pointer is 1.0.69

1.0.69 — 2026-10-23

Features

Dashboard directory grouping now shows a header with the folder path and agent count above each folder's agents.
Dashboard location (branch and folder) now sits on the dispatch box's top border instead of its own header row.
/resume on the dashboard now lists sessions from every folder, with the current folder and its checkouts first.
Welcome menu now lists Resume session (Ctrl+R) above New worktree (Ctrl+W).

Bug Fixes

[ui] simple_mode is retired and no longer has any effect; sessions always open with the prompt focused.
grok doctor and /doctor now point out a leftover [ui] vim_mode key (now [keys] vim) or the retired [ui] simple_mode key in your config.

Performance

Quitting Grok returns you to your shell sooner because it no longer waits out fixed intervals.

source

Alpha channel 1.0.68

2026-10-22 · alpha · alpha pointer is 1.0.68

1.0.68 — 2026-10-22

Features

Worktree create and fork now list files that could not be copied and any mode changes that were lost.
Vim navigation is now one setting, [keys] vim (off or nav), also set with /vim-mode; the older [ui] vim_mode still works.
Terminal theme now draws a subtle band behind your prompts using the colors your terminal reports.
Mode-switch toast now shows the mode name (Agent, Plan, Ask) in the mode's own color.
Change-directory picker (Ctrl+L on the dashboard) opens with the path field unfocused, like the /resume search.
Image placeholders in the transcript now show whether the actual image was sent or only the text was typed.

Bug Fixes

Worktree create now fails cleanly instead of falling back when it cannot copy tracked files or runs out of disk space.
Running Grok sessions are no longer broken when an update deletes their binary.
Dashboard model picks now keep their selected context window when spawning the next session.
Send Now prompts and rewound turns now correctly restore and mark any images that were originally sent.
Resumed sessions now show the correct sent or struck-through state for image placeholders.

source

Stable channel 1.0.67

2026-10-21 · stable · stable pointer is 1.0.67

1.0.67 — 2026-10-21

Features

Worktrees also use their main checkout's saved "always allow" answers, so you are not asked again for a tool you already allowed there. Set [features] worktree_inherits_main_approvals = false to turn this off; the main checkout's "never allow" answers apply either way.
Prompt suggestions follow vim mode: with [ui].vim_mode on, Tab moves focus to the scrollback instead of accepting the suggestion. → or Ctrl+→ accepts a suggestion, and Esc dismisses it.

Bug Fixes

Headless grok -p runs exit when their work is done instead of hanging when a background task or subagent finishes and its output is read in the same turn.
The theme picker in Settings restores your previous theme when you leave it any way other than Enter, as /theme already does.
Quitting a session with hooks configured is faster: the session-end Stop hook runs at the same time as the SessionEnd hooks and gets at most 1.5 seconds.

source

Enterprise channel 1.0.67

2026-10-21 · enterprise · enterprise pointer is 1.0.67

1.0.67 — 2026-10-21

Features

Worktrees also use their main checkout's saved "always allow" answers, so you are not asked again for a tool you already allowed there. Set [features] worktree_inherits_main_approvals = false to turn this off; the main checkout's "never allow" answers apply either way.
Prompt suggestions follow vim mode: with [ui].vim_mode on, Tab moves focus to the scrollback instead of accepting the suggestion. → or Ctrl+→ accepts a suggestion, and Esc dismisses it.

Bug Fixes

Headless grok -p runs exit when their work is done instead of hanging when a background task or subagent finishes and its output is read in the same turn.
The theme picker in Settings restores your previous theme when you leave it any way other than Enter, as /theme already does.
Quitting a session with hooks configured is faster: the session-end Stop hook runs at the same time as the SessionEnd hooks and gets at most 1.5 seconds.

source

Alpha channel 1.0.67

2026-10-21 · alpha · alpha pointer is 1.0.67

1.0.67 — 2026-10-21

Features

Worktrees also use their main checkout's saved "always allow" answers, so you are not asked again for a tool you already allowed there. Set [features] worktree_inherits_main_approvals = false to turn this off; the main checkout's "never allow" answers apply either way.
Prompt suggestions follow vim mode: with [ui].vim_mode on, Tab moves focus to the scrollback instead of accepting the suggestion. → or Ctrl+→ accepts a suggestion, and Esc dismisses it.

Bug Fixes

Headless grok -p runs exit when their work is done instead of hanging when a background task or subagent finishes and its output is read in the same turn.
The theme picker in Settings restores your previous theme when you leave it any way other than Enter, as /theme already does.
Quitting a session with hooks configured is faster: the session-end Stop hook runs at the same time as the SessionEnd hooks and gets at most 1.5 seconds.

source

Enterprise channel 1.0.66

2026-10-20 · enterprise · enterprise pointer is 1.0.66

1.0.66 — 2026-10-20

Bug Fixes

/theme restores your previous theme when you leave the picker any way other than Enter, such as Esc, Ctrl+U, Tab then Esc, or opening Settings or the dashboard.
Selected and highlighted rows stay readable on every theme because reverse-video text is never dimmed. This fixes faint descriptions in selected menu rows on the terminal theme in terminals such as Kitty.

Performance

@ file completion shows results about 5 ms after you type instead of about 40 ms.

source

Alpha channel 1.0.66

2026-10-20 · alpha · alpha pointer is 1.0.66

1.0.66 — 2026-10-20

Bug Fixes

/theme restores your previous theme when you leave the picker any way other than Enter, such as Esc, Ctrl+U, Tab then Esc, or opening Settings or the dashboard.
Selected and highlighted rows stay readable on every theme because reverse-video text is never dimmed. This fixes faint descriptions in selected menu rows on the terminal theme in terminals such as Kitty.

Performance

@ file completion shows results about 5 ms after you type instead of about 40 ms.

source

Alpha channel 1.0.65

2026-10-20 · alpha · alpha pointer is 1.0.65

1.0.65 — 2026-10-20

Features

/resume search finds sessions from other folders, tolerates typos, and also matches the commands, file paths, and replies in a session.
--permission-mode plan starts the session in plan mode, as documented, instead of the default mode.
The dashboard's Ctrl+L folder picker lists up to 50 recent folders instead of 10.
grok --help gives each command and option a one-line summary, and grok help -h and grok help --help print the help command's own help instead of an error.

Bug Fixes

/resume search keeps a single typed character literal and matches each typed word at the start of a word, so log fix no longer matches "Blog prefix cleanup".
DISABLE_TELEMETRY and DISABLE_ERROR_REPORTING turn telemetry and crash reports off for any non-empty value, including 0 and false. A GROK_TELEMETRY_ENABLED or GROK_ERROR_REPORTING value Grok can't read also counts as off.

source

Alpha channel 1.0.64

2026-10-19 · alpha · alpha pointer is 1.0.64

1.0.64 — 2026-10-19

Breaking Changes

Untrusted folders run only your organization's managed hooks. User, project, plugin, and agent hooks wait until you trust the folder. Headless sessions treat user hooks as trusted.

Features

Worktree names typed with Ctrl+W keep Unicode letters such as café, and a name or ref that starts with - is refused.
Auto mode runs a shell command without review only when every command in it matches an allowlist entry; anything else asks first. Permission rules read each word the way bash passes it to the program.
/resume loads older sessions when you reach the last row, and shows how many sessions are loaded.
Auto mode runs make --version, cmake --version, and just --version without asking.
HTTP connections, including remote MCP servers, accept gzip, brotli, zstd, and deflate compressed responses.
Plugin install suggestions above the prompt are off when disable_plugins = true is set.

Bug Fixes

New worktrees keep tracked files that match .gitignore, and leave out files excluded only by .git/info/exclude or core.excludesFile.
A new session that fails to start puts what you typed back in the prompt you return to, instead of losing it.
Web search works from a session's first turn, and in subagents, when the search model signs in through an external auth provider.
grok -w, grok worktree create, worktree forks, and subagent worktrees refuse up front with the sandbox's message when the sandbox profile does not allow creating worktrees.
Worktree options are hidden in /fork, /new, and the Welcome screen when the sandbox profile does not allow creating worktrees.
A worktree name you type that is already taken is refused instead of getting a numbered suffix, and a name whose folder you deleted by hand can be used again.
External OpenTelemetry export keeps sending logs and metrics when OTEL_EXPORTER_OTLP_COMPRESSION is set to gzip or zstd.
**[model.*] tables** without a context window take the window and API of another table that sends the same model, instead of the 200,000-token fallback.
Grok on macOS without a GUI session, such as over SSH, no longer hangs on its first modifier-key check.
Custom models on the Responses API keep the turn going when a web search call starts without an action.
Shell commands run correctly on zsh when your startup files alias grep or replace run-help.
Voice dictation preview grows the prompt as you speak and keeps the newest words in view.

source

Alpha channel 1.0.63

2026-10-17 · alpha · alpha pointer is 1.0.63

1.0.63 — 2026-10-17

Breaking Changes

Memory always uses the concise MEMORY.md index of topic titles and paths, and /memory lists topics by how often Grok reads them. The [memory_v2] compact_index_enabled setting no longer has an effect.

Features

Rex terminal support reports each open session and each running subagent as its own record, so Rex shows what each one is doing. Set children = false under [ui.notifications.program_status] to report only the main status.
grok worktree list and grok worktree show mark a worktree whose folder is missing or whose source repository moved, with a hint on how to repair it.
Command hooks accept an args list: Grok runs command as one program with those arguments and no shell, and fills hook variables such as $GROK_SESSION_ID in each argument.

Bug Fixes

Plan reviews that open while you are typing keep your text in the review's notes, so a typed letter no longer triggers a plan shortcut such as approve.
Plan reviews at the end of a turn wait until you close an open /feedback dialog, question, tool prompt, or queued prompt edit.
grok -w and Ctrl+W explain that no worktree was made when you run them outside a git repository.
/fork keeps the parent session's model and effort when Grok was started with -m or --effort.
Custom models on Chat Completions endpoints outside xAI no longer fail turns with a tool schema error when MCP servers are connected.
Workflow run views hide each agent's live status text on narrower terminals, so agent names are no longer cut off.

source

Stable channel 1.0.62

2026-10-16 · stable · stable pointer is 1.0.62

1.0.62 — 2026-10-16

Bug Fixes

Headless mode (grok -p) waits for the follow-up turn that a finished background task starts before exiting, and a --background-wait-timeout that stops pending work exits with an error instead of success.
Cold /resume from the dashboard overlay keeps the overlay attached, so you can still back out with Left or Ctrl+\.
Enter on a highlighted slash command with optional arguments runs the command instead of leaving a trailing space in the composer.
CLI commands such as grok models, grok sessions and grok export no longer crash when their output pipe closes early, such as when piped to head, or when the output device is full.
grok update and grok login no longer crash when stdout or stderr is closed or unwritable.

1.0.61 — 2026-10-16

Features

Resume picker finds sessions whose titles contain every word you type, in any order.
/resume <id or name> opens the session directly when the ID or title matches exactly one session, instead of always opening the picker.

Bug Fixes

Dashboard peek text can be selected by dragging over it with the mouse, and the selection is copied to the clipboard when you release.
Workflow run views scroll toward newer agents when you wheel down the agent roster, instead of snapping back.
Hook spawn failures on Windows name the program that failed to start, such as PowerShell or Git Bash, instead of blaming the hook's script.

source

Enterprise channel 1.0.62

2026-10-16 · enterprise · enterprise pointer is 1.0.62

1.0.62 — 2026-10-16

Bug Fixes

Headless mode (grok -p) waits for the follow-up turn that a finished background task starts before exiting, and a --background-wait-timeout that stops pending work exits with an error instead of success.
Cold /resume from the dashboard overlay keeps the overlay attached, so you can still back out with Left or Ctrl+\.
Enter on a highlighted slash command with optional arguments runs the command instead of leaving a trailing space in the composer.
CLI commands such as grok models, grok sessions and grok export no longer crash when their output pipe closes early, such as when piped to head, or when the output device is full.
grok update and grok login no longer crash when stdout or stderr is closed or unwritable.

1.0.61 — 2026-10-16

Features

Resume picker finds sessions whose titles contain every word you type, in any order.
/resume <id or name> opens the session directly when the ID or title matches exactly one session, instead of always opening the picker.

Bug Fixes

Dashboard peek text can be selected by dragging over it with the mouse, and the selection is copied to the clipboard when you release.
Workflow run views scroll toward newer agents when you wheel down the agent roster, instead of snapping back.
Hook spawn failures on Windows name the program that failed to start, such as PowerShell or Git Bash, instead of blaming the hook's script.

source

Alpha channel 1.0.62

2026-10-16 · alpha · alpha pointer is 1.0.62

1.0.62 — 2026-10-16

Bug Fixes

Headless mode (grok -p) waits for the follow-up turn that a finished background task starts before exiting, and a --background-wait-timeout that stops pending work exits with an error instead of success.
Cold /resume from the dashboard overlay keeps the overlay attached, so you can still back out with Left or Ctrl+\.
Enter on a highlighted slash command with optional arguments runs the command instead of leaving a trailing space in the composer.
CLI commands such as grok models, grok sessions and grok export no longer crash when their output pipe closes early, such as when piped to head, or when the output device is full.
grok update and grok login no longer crash when stdout or stderr is closed or unwritable.

source

Alpha channel 1.0.61

2026-10-16 · alpha · alpha pointer is 1.0.61

1.0.61 — 2026-10-16

Features

Resume picker finds sessions whose titles contain every word you type, in any order.
/resume <id or name> opens the session directly when the ID or title matches exactly one session, instead of always opening the picker.

Bug Fixes

Dashboard peek text can be selected by dragging over it with the mouse, and the selection is copied to the clipboard when you release.
Workflow run views scroll toward newer agents when you wheel down the agent roster, instead of snapping back.
Hook spawn failures on Windows name the program that failed to start, such as PowerShell or Git Bash, instead of blaming the hook's script.

source

Stable channel 1.0.60

2026-10-15 · stable · stable pointer is 1.0.60

1.0.60 — 2026-10-15

Features

grok --resume <title> lists matching sessions newest first with when each was last updated, and suggests similar titles when nothing matches.
Shift+Y in the fullscreen markdown viewer copies the code block under the cursor.

Bug Fixes

grok --resume <title> with a title that matches no local session reports the miss right away instead of trying a remote restore and asking you to sign in.
Esc on the first question in a session opened from the dashboard moves focus to the scrollback, as it does in a standalone session, instead of returning to the dashboard.

1.0.59 — 2026-10-14

Features

Dashboard /loop without a prompt shows its usage instead of starting a new session.
grok --resume <title> matches a title whatever separates its words, so fix-login-bug finds "Fix login bug".
[Copy] buttons appear under code blocks in finished replies, labeled with the code's language or file name.

Bug Fixes

Dashboard peek replies waiting on a clipboard paste are canceled with a toast when you move the peek to another row or press Esc, instead of being sent later.
grok worktree rm no longer suggests -f for a worktree that -f would also refuse to remove.
Advanced PowerShell inline hooks that read $input run with a warning instead of failing.
Hooks that return continue: false from PreToolUse, PostToolUse, or UserPromptSubmit stop the turn instead of being ignored.
Grok no longer crashes at startup when the SENTRY_DSN environment variable holds an invalid value.
Copy rows under code blocks, tables, and diagrams display correctly after switching between minimal and fullscreen modes.
Copying a code block or a whole reply keeps its original tabs instead of turning them into spaces.

1.0.58 — 2026-10-14

Bug Fixes

Minor bug fixes.

1.0.57 — 2026-10-14

Bug Fixes

Minor bug fixes.

1.0.56 — 2026-10-13

Bug Fixes

Minor bug fixes.

1.0.55 — 2026-10-13

Bug Fixes

Minor bug fixes.

1.0.54 — 2026-10-12

Features

/resume takes an optional session id or title and opens the picker filtered to matching sessions.

Bug Fixes

Pasted Windows line endings reach the model as newlines, markdown images no longer show a stray "!", and dollar amounts that are not math stay plain text.
Write approval cards show a binary or unreadable file as unreadable instead of garbled text.
Prompts you send while a session is resuming at startup are queued and sent once it loads instead of staying in the composer.
A corrupt active-sessions registry no longer stops new sessions from registering: grok keeps a backup, salvages the session rows it can read, and warns you.
Wrapped URLs in shell command descriptions link to the full address on every row.
Worktree forks (grok -w -r or /fork into a worktree) run in and show the worktree's path, so a shell command no longer edits the original checkout.
Links are clickable in herdr panes.
Memory search shows the matching part of a long line instead of cutting it off before the match.
/memory skips a symlinked note instead of failing the whole listing.
Startup is faster when TMUX names a tmux server that has stopped.
Failed MCP tool rows wrap the error to the screen width and keep its line breaks.
[skills].ignore matches a symlinked SKILL.md by the folder that holds it.
Failed file reads expand to show why they failed.
Send now (Ctrl+Enter) on a slash command runs or queues it as a command instead of sending it as a prompt.

Performance

The idle welcome screen uses less CPU.
Prompt undo history is capped by size, so large pastes you delete no longer stay in memory.
Typing in a prompt that holds large pastes is faster.
Long sessions use less memory: long replies scrolled off screen release their render caches.

1.0.53 — 2026-10-12

Features

Wrapped shell commands in tool and background-task rows copy back as one line.
/hooks lists load errors from the main agent's inline hooks and from plugins.
--tools and --disallowed-tools also apply in the interactive TUI.
/goal pauses after three continuations in a row with no tool call; run /goal resume to continue.
Quitting warns when it would stop a running workflow and asks you to confirm with a second quit.
Multi-file edits show how many other files the edit changed.

Bug Fixes

Worktrees created from a repository subfolder report the repository root as their source in grok worktree list and grok worktree show.
/feedback stays open while you type when a permission request, question, or MCP prompt arrives.
Persistent shell runs your command even when restoring its saved state fails.
/fork is refused while an approval request is open, so the approval keys keep answering that request.
grok --help and grok --version no longer create a .grok folder.
Wrapped URLs in tool headers link to the full address on every row.
search_replace refuses an old_string that matches again at an overlapping position, so it no longer edits the wrong match.
read_file reports the correct line count for files that end with a newline and no longer returns an empty line past the end.
/workflow stop and /workflow pause no longer pick a run from a loose fuzzy match, and headless -p runs print each refused approval request to stderr.
Workflow run views list agents outside a declared phase and count only finished agents as done.
Parallel MCP tool calls that ask for input each show their own prompt.
/feedback Drafts load the valid drafts when one saved draft is invalid.
grok sessions list and /resume keep working when one session directory is unreadable.
grok plugin install interrupted mid-clone on Linux no longer leaves a partial folder that blocks the retry.
Hooks you disabled stay disabled when the disabled-hooks file cannot be read.
Edit diffs mark removed and added lines with - and + when color is off.
Session usage keeps every turn when two windows write to the same session.
Failed directory listings expand to show why they failed.
Stdio MCP servers expand a leading ~ in their command and arguments to your home directory.
Editing a queued command such as /compact is refused with a message, so your new text is kept.
/docs opens the exact guide or picker you type.
grok inspect warns when a known [features] key in config.toml has a non-boolean value.
Send now (Ctrl+Enter) no longer answers the stop-subagents panel or an MCP prompt card.

Performance

@-mentions reuse the file list until the folder tree changes, so suggestions appear faster.

source

Alpha channel 1.0.60

2026-10-15 · alpha · alpha pointer is 1.0.60

1.0.60 — 2026-10-15

Features

grok --resume <title> lists matching sessions newest first with when each was last updated, and suggests similar titles when nothing matches.
Shift+Y in the fullscreen markdown viewer copies the code block under the cursor.

Bug Fixes

grok --resume <title> with a title that matches no local session reports the miss right away instead of trying a remote restore and asking you to sign in.
Esc on the first question in a session opened from the dashboard moves focus to the scrollback, as it does in a standalone session, instead of returning to the dashboard.

source

Alpha channel 1.0.59

2026-10-14 · alpha · alpha pointer is 1.0.59

1.0.59 — 2026-10-14

Features

Dashboard /loop without a prompt shows its usage instead of starting a new session.
grok --resume <title> matches a title whatever separates its words, so fix-login-bug finds "Fix login bug".
[Copy] buttons appear under code blocks in finished replies, labeled with the code's language or file name.

Bug Fixes

Dashboard peek replies waiting on a clipboard paste are canceled with a toast when you move the peek to another row or press Esc, instead of being sent later.
grok worktree rm no longer suggests -f for a worktree that -f would also refuse to remove.
Advanced PowerShell inline hooks that read $input run with a warning instead of failing.
Hooks that return continue: false from PreToolUse, PostToolUse, or UserPromptSubmit stop the turn instead of being ignored.
Grok no longer crashes at startup when the SENTRY_DSN environment variable holds an invalid value.
Copy rows under code blocks, tables, and diagrams display correctly after switching between minimal and fullscreen modes.
Copying a code block or a whole reply keeps its original tabs instead of turning them into spaces.

source

Alpha channel 1.0.54

2026-10-12 · alpha · alpha pointer is 1.0.54

1.0.54 — 2026-10-12

Features

/resume takes an optional session id or title and opens the picker filtered to matching sessions.

Bug Fixes

Pasted Windows line endings reach the model as newlines, markdown images no longer show a stray "!", and dollar amounts that are not math stay plain text.
Write approval cards show a binary or unreadable file as unreadable instead of garbled text.
Prompts you send while a session is resuming at startup are queued and sent once it loads instead of staying in the composer.
A corrupt active-sessions registry no longer stops new sessions from registering: grok keeps a backup, salvages the session rows it can read, and warns you.
Wrapped URLs in shell command descriptions link to the full address on every row.
Worktree forks (grok -w -r or /fork into a worktree) run in and show the worktree's path, so a shell command no longer edits the original checkout.
Links are clickable in herdr panes.
Memory search shows the matching part of a long line instead of cutting it off before the match.
/memory skips a symlinked note instead of failing the whole listing.
Startup is faster when TMUX names a tmux server that has stopped.
Failed MCP tool rows wrap the error to the screen width and keep its line breaks.
[skills].ignore matches a symlinked SKILL.md by the folder that holds it.
Failed file reads expand to show why they failed.
Send now (Ctrl+Enter) on a slash command runs or queues it as a command instead of sending it as a prompt.

Performance

The idle welcome screen uses less CPU.
Prompt undo history is capped by size, so large pastes you delete no longer stay in memory.
Typing in a prompt that holds large pastes is faster.
Long sessions use less memory: long replies scrolled off screen release their render caches.

source

Enterprise channel 1.0.53

2026-10-12 · enterprise · enterprise pointer is 1.0.53

1.0.53 — 2026-10-12

Features

Wrapped shell commands in tool and background-task rows copy back as one line.
/hooks lists load errors from the main agent's inline hooks and from plugins.
--tools and --disallowed-tools also apply in the interactive TUI.
/goal pauses after three continuations in a row with no tool call; run /goal resume to continue.
Quitting warns when it would stop a running workflow and asks you to confirm with a second quit.
Multi-file edits show how many other files the edit changed.

Bug Fixes

Worktrees created from a repository subfolder report the repository root as their source in grok worktree list and grok worktree show.
/feedback stays open while you type when a permission request, question, or MCP prompt arrives.
Persistent shell runs your command even when restoring its saved state fails.
/fork is refused while an approval request is open, so the approval keys keep answering that request.
grok --help and grok --version no longer create a .grok folder.
Wrapped URLs in tool headers link to the full address on every row.
search_replace refuses an old_string that matches again at an overlapping position, so it no longer edits the wrong match.
read_file reports the correct line count for files that end with a newline and no longer returns an empty line past the end.
/workflow stop and /workflow pause no longer pick a run from a loose fuzzy match, and headless -p runs print each refused approval request to stderr.
Workflow run views list agents outside a declared phase and count only finished agents as done.
Parallel MCP tool calls that ask for input each show their own prompt.
/feedback Drafts load the valid drafts when one saved draft is invalid.
grok sessions list and /resume keep working when one session directory is unreadable.
grok plugin install interrupted mid-clone on Linux no longer leaves a partial folder that blocks the retry.
Hooks you disabled stay disabled when the disabled-hooks file cannot be read.
Edit diffs mark removed and added lines with - and + when color is off.
Session usage keeps every turn when two windows write to the same session.
Failed directory listings expand to show why they failed.
Stdio MCP servers expand a leading ~ in their command and arguments to your home directory.
Editing a queued command such as /compact is refused with a message, so your new text is kept.
/docs opens the exact guide or picker you type.
grok inspect warns when a known [features] key in config.toml has a non-boolean value.
Send now (Ctrl+Enter) no longer answers the stop-subagents panel or an MCP prompt card.

Performance

@-mentions reuse the file list until the folder tree changes, so suggestions appear faster.

source

Alpha channel 1.0.53

2026-10-12 · alpha · alpha pointer is 1.0.53

1.0.53 — 2026-10-12

Features

Wrapped shell commands in tool and background-task rows copy back as one line.
/hooks lists load errors from the main agent's inline hooks and from plugins.
--tools and --disallowed-tools also apply in the interactive TUI.
/goal pauses after three continuations in a row with no tool call; run /goal resume to continue.
Quitting warns when it would stop a running workflow and asks you to confirm with a second quit.
Multi-file edits show how many other files the edit changed.

Bug Fixes

Worktrees created from a repository subfolder report the repository root as their source in grok worktree list and grok worktree show.
/feedback stays open while you type when a permission request, question, or MCP prompt arrives.
Persistent shell runs your command even when restoring its saved state fails.
/fork is refused while an approval request is open, so the approval keys keep answering that request.
grok --help and grok --version no longer create a .grok folder.
Wrapped URLs in tool headers link to the full address on every row.
search_replace refuses an old_string that matches again at an overlapping position, so it no longer edits the wrong match.
read_file reports the correct line count for files that end with a newline and no longer returns an empty line past the end.
/workflow stop and /workflow pause no longer pick a run from a loose fuzzy match, and headless -p runs print each refused approval request to stderr.
Workflow run views list agents outside a declared phase and count only finished agents as done.
Parallel MCP tool calls that ask for input each show their own prompt.
/feedback Drafts load the valid drafts when one saved draft is invalid.
grok sessions list and /resume keep working when one session directory is unreadable.
grok plugin install interrupted mid-clone on Linux no longer leaves a partial folder that blocks the retry.
Hooks you disabled stay disabled when the disabled-hooks file cannot be read.
Edit diffs mark removed and added lines with - and + when color is off.
Session usage keeps every turn when two windows write to the same session.
Failed directory listings expand to show why they failed.
Stdio MCP servers expand a leading ~ in their command and arguments to your home directory.
Editing a queued command such as /compact is refused with a message, so your new text is kept.
/docs opens the exact guide or picker you type.
grok inspect warns when a known [features] key in config.toml has a non-boolean value.
Send now (Ctrl+Enter) no longer answers the stop-subagents panel or an MCP prompt card.

Performance

@-mentions reuse the file list until the folder tree changes, so suggestions appear faster.

source

Stable channel 1.0.52

2026-10-11 · stable · stable pointer is 1.0.52

1.0.52 — 2026-10-11

Features

Rex terminal support: Grok now reports its status (working, waiting for you, done, or failed) to Rex through OSC 7501, so Rex tabs show what each session is doing.
Tool call density improvements: expanded tool call groups now stay expanded as new calls join them, and Ctrl+E expands or collapses thinking faster in long transcripts.
Project hooks placed in a repository subfolder are now loaded when you launch inside that folder.
HTTP hooks to localhost now work, and bad hook URLs are reported at load time.
Command status-line scripts now see the current turn's prompt ID and start time while the turn is active.

Bug Fixes

Grep searches that time out or fail now show an error instead of appearing as empty results.
/model switches that fail no longer change the footer or saved default model.
Export and copy commands no longer overwrite an existing file; they refuse with a clear message.
Shell commands no longer keep running after you quit grok or it receives a signal.
Approval cards for background subagents no longer disappear when the parent turn ends.
grok worktree rm and gc no longer delete a worktree that an open session is using.
grok inspect no longer warns about valid pins that only requirements.toml accepts.
Pasted images that exceed the 50 MB send limit are now rejected with a clear message instead of silently dropped.
grok -p --effort now prints a warning when the model ignores the flag and refuses invalid effort levels.
Permission prompts now show which session needs approval.
grok worktree rm now refuses to delete git repositories, .git folders, or any directory grok did not create.
Grep and list_dir now report a clear error when the search root or directory cannot be read instead of silently returning no results.
Cancelled terminal commands now show the actual command that was stopped.
Wrapped URLs in prompts and quotes now link to the complete address instead of being cut at the wrap.
@mentions and image placeholders now respect Read deny rules and are omitted from the prompt.
Restarting after a model switch no longer reverts to the previous model.
Long tool turns that include PostToolUse hook output now compact correctly instead of hitting the prompt length limit.
Headless prompts that start with a byte-order mark no longer send the mark to the model or break JSON prompt files.
When a permission rule blocks a tool call, any context returned by a PreToolUse hook still reaches the model.
An invalid shell_environment_policy table in config.toml now causes grok to exit with a clear error instead of silently using the full environment.
MCP tool replies containing invalid characters no longer cause the call to time out.
Context overflow errors now trigger automatic compaction instead of failing the turn.
Inline tags in agent replies no longer break line layout, and /workflow stop now picks the correct live run.
Auto-compaction now stops after three context refills in a row instead of looping forever.
Minimal mode now shows the permission prompt instead of a hidden plan or cancel panel when both are present.
Project rules files that resolve outside the git root or workspace folder are now skipped.
[permission] rules now load alongside compact deny/allow/ask lists in config.toml.
Unrecognized permission keys or [permissions] sections now appear as skipped entries in grok inspect.
Approval cards now close the block viewer so the card is visible.
Shell redirects to wildcard or variable paths now correctly respect deny rules in Auto mode.
The in-session resume picker now matches the ungrouped layout shown on the welcome screen when configured.
HTTP MCP tools that return a JSON-RPC error are now executed only once.
StopFailure hook matchers containing commas or spaces are now treated as regex patterns.

Performance

Multiple Notification hook chains now start in parallel instead of waiting for the previous chain to finish.

1.0.51 — 2026-10-09

Breaking Changes

Cancel and edit-resend on long sessions no longer unexpectedly revert workspace files.
Memory settings simplified: rollout and file_writes_enabled options removed; memory is on or off.
Removed the /share command and related session sharing features from the CLI and SDK.

Features

Inline viewports can now be restored at a previously saved screen area without extra cursor queries.
Headless grok -p now waits for background goals to finish and reports their outcome.
/imagine and similar commands now run the matching bundled skill when the required tools are unavailable.
New /density command and Tool call density setting let you choose how much detail tool calls show in the transcript.
Web fetch approval cards now include "always allow" and "never allow" domain options that persist for the folder.
Slash command completion now updates live when skills are added or changed during a session.
Request form now securely collects and fills sensitive details like passwords or card numbers without sending values to the model.
Chrome profiles can now be persisted across restarts using durable state directories.
Workspace server sessions now share one Chrome instance for cookies and storage.
/feedback Drafts tab now supports G / Home and End to jump to the first or last draft.
Pinned sessions on the dashboard now stay in the order you pinned or reordered them.
Shift+Enter, Ctrl+., Ctrl+/ and Ctrl+[ / ] now work inside tmux when extended-keys are enabled.
grok -p --output-format json now reports the exact worktree folder used (useful after -w name collisions).
Bare hook commands such as my-guard now run from PATH when not found next to the hook file.
Headless JSON error output now includes the worktree folder when a session was started.
grok sessions list and /resume now show worktree sessions started from the same subfolder.
/exit (and aliases) in a dashboard details view now closes only that session instead of quitting Grok.
Editors offering file access now receive a one-time transcript notice that the agent reads and writes disk files.
--background-wait-timeout now also stops an active headless goal after the timeout and exits with an error.
Interactive sessions now include the Grok Build user guide location so the agent can answer usage questions.
New /workflow commands let you list, stop, and resume background workflow runs without sending them to the model.
Background workflow runs now appear in the tasks pane with a stop button.
Hooks that ask about web searches or file deletes now show the hook name on the approval card.
config.toml now accepts an [update] section without breaking other commands.
Hook load errors listed in /hooks can now be expanded to see the full message.
grok worktree commands now work correctly inside Jujutsu repositories.
Images pasted with skill commands like /pr-workflow are now sent to the model.
Pressing Ctrl+C on a goal now pauses the goal instead of letting it continue.
grok -p now runs auth_provider_command when no login is stored.
Reopening a chat with an active goal now continues the goal automatically.
Browser tabs can now be restored on demand instead of all at once when a session resumes.
grok agent stdio now supports listing sessions via ACP and returns consistent reply envelopes.
Answering an approval with a message now tells the model it is a follow-up.
--allowed-tools is now accepted as an alias for --allow in headless mode.

Bug Fixes

Slash command suggestions no longer make the text after the cursor disappear while typing.
Running workflows in the dock now show the same pulsing indicator as in the main scrollback.
Mode switch tool calls now appear in the transcript with the target mode name.
Always-approve mode now automatically approves web requests without extra prompts.
The follow indicator no longer appears when only blank padding sits below the viewport.
Search failures now keep the original pattern instead of showing an empty Search "".
Sessions that open without a model list now use the public Auto id and succeed on every account.
Background wake turns that request plan changes or mode switches are now handled correctly.
Model-call retries now show progress status instead of a generic waiting message.
Foreground subagent activity now appears correctly after resume.
Very large pasted prompts are now offloaded to a file instead of being sent inline.
Sign-in failures now show a clear plain-English message with the next step instead of raw errors.
Logging out now stops any running turns on local sessions instead of leaving them active.
The send-now shortcut no longer accidentally enables always-approve mode while a permission card is open.
/find now keeps match counts and navigation accurate while replies or command output stream in.
Images from tools are now validated so bad files no longer break the current turn or future replays.
Screenshots and other image artifacts saved by the browser tool are now stored as real image files.
Rate-limit errors now show the expected message; very small images are rejected early.
Send-now (Ctrl+Enter) no longer accidentally answers an open question card.
Opening the queue with Up now scrolls so the selected follow-up is visible.
Command output that contains clear-screen codes is no longer discarded.
Dashboard cursor and overlay counter now correctly track the active grouping view.
Headless JSON output now reports the correct stop reason when the turn limit is reached and removes duplicate error prefixes.
Session picker footer no longer lists the same shortcut twice.
Dragging images from a file manager onto terminals like GNOME Terminal or xfce4-terminal now attaches them correctly.
Plan comments added while a revision runs are now preserved and sent when the plan is approved.
Permission prompts that fail now show a clearer message without blaming the interface.
--disallowed-tools now accepts the exact names shown to the model (run_terminal_command, spawn_subagent, etc.).
Long pastes without bracketed-paste markers are now read completely instead of stalling after ~1 KB.
A feedback send now shows "Sending feedback…" immediately; failed sends no longer repeat the lead phrase.
Typing, arrows or Enter while an approval card appears no longer accidentally approve the action or enable always-approve mode.
Error hints for legacy session import now appear in full instead of being cut off by the pager.
Skills in the folder where grok started now correctly show as (local).
Worktree creation no longer hangs when the repo contains named pipes or sockets.
--no-subagents flag now works correctly in headless / -p sessions.
Resumed sessions now correctly show aborted subagent tasks as cancelled instead of failed.
Large parallel shell batches no longer drop calls with "hook state lost" errors.
Write approval cards now show the existing file contents that will be replaced.
Slash command completion now shows the correct name for files under commands/ instead of always showing /commands.
PreToolUse hooks now validate rewrites against the tool schema and deny invalid ones before running the tool.
Ctrl+C during a turn now cancels the turn even when a modal such as /mcps or the block viewer is open.
grok -w now opens the session at the worktree root (with a note) when the original launch folder does not exist in the new worktree.
/cd and the location picker now accept quoted or escaped folder paths.
Web search hooks now fire under the correct web_search name and carry query.
Large file writes or many screenshots no longer end the session.
Glob tool hooks now match under list_dir and carry the correct directory.
/find now scrolls to the exact wrapped row containing the current match.
Downloads via FetchMcpResource are now refused in plan mode and resolve paths consistently.
Pinned dashboard sessions now survive restarts and layout changes instead of disappearing from the saved list.
Slash commands like /imagine now work regardless of capitalization and correctly substitute skills when tools are missing.
Turning off bypass mode without naming a session now correctly restores prompts in all open sessions.
/delete with extra text now shows an error instead of deleting the current session.
Failed /feedback status messages now wrap inside the dialog instead of clipping at the border.
Number-key picks in the dashboard peek no longer leak to the wrong agent after Up/Down navigation.
Worktrees created with -w are now correctly tracked and resumable even when GROK_HOME is reached through a symlink.
grok worktree rm and gc now clean up names after manually deleting a worktree folder so -w NAME works again.
Hook scripts that read stdin with read now receive a complete line.
A policy that disables non-managed hooks now applies to the next session instead of only after restart.
/resume and grok sessions list no longer show worktree sessions from a sibling submodule or separate-git-dir repo.
A hook listed under two different matchers now runs only once per tool call.
Typing a hidden command such as /cd in a session and pressing Enter now runs it instead of accepting a near-match skill.
@-file completion no longer accepts a match from an earlier query when you type quickly.
Tool hooks now run one at a time per session and stop promptly on cancel or exit.
PostToolUse hooks now receive a correct whole-millisecond duration.
PreToolUse hooks for shell commands now see the model's block_until_ms value.
MCP tool calls now show hook permission prompts and cards name the hook that asked.
Typing while a permission or ask card is visible no longer selects or answers the card.
grok worktree list now shows unfinished copies left by an interrupted create.
Auto mode now asks before deleting files that the classifier has not already allowed.
Fixed intermittent List/Read/Open failures during concurrent v2 hardening migrations.
grok -p -s with a reused session id now shows the plain refusal message instead of JSON-RPC noise.
Settings saves no longer re-insert the built-in default fork model into config.toml.
Fixed several pager rows that showed wrong paths, truncated names, or hid failure/cancel states.
--system-prompt now correctly appends rules from --append-system-prompt instead of dropping them.
/new now inherits always-approve mode from the tab it was run from instead of another tab.
Multi-select questions now let you toggle several answers with number keys before submitting.
Slash-command notices typed while a reply streams no longer hide the streaming reply.
@ file search no longer accepts keys or moves the highlight while results are still updating.
Worktree db rebuild now restores names so rm and show by name work again.
A prompt blocked by a hook no longer titles the session.
Background subagents now respect saved approvals and --always-approve after the parent turn finishes.
Per-session system prompt overrides now correctly use only the session's own rules.
Connection stalled errors now show a clearer message with network advice.
Invalid model provider tables now produce a clear startup error instead of silent fallback.
Narrow feedback dialog now shows full draft titles and readable confirm prompts.
/workflow stop|resume <name> now correctly targets the live run instead of an ended first run.
Dialogs like /feedback no longer disappear on very short terminal windows.
Delete operations now respect the same permission rules as shell rm commands.
Hooks can no longer rewrite a web-search call to restrict domains.
Workspace server now starts with per-session private Chrome by default.
Notification hooks for permission prompts and task completion now run reliably before a headless session exits.
Headless goal runs no longer print a spurious "Goal paused" message after completion.
Legacy UI density settings now combine correctly across config tiers and produce expected folding.
Approval answers sent after a turn ends no longer affect the turn.
Background shell commands now fail immediately when the working directory does not exist.
grok worktree rm without -f now safely keeps worktrees when the unsaved-work check fails or times out.
The dashboard slash menu now shows skills from the home session even when opened early.
grok worktree rm and gc now correctly remove worktrees whose only untracked content is empty folders.
Editing a queued prompt now prevents accidental removal with a stale client view.
The model is now told when memory is turned off or on for the session.
Hook results in the UI now correctly distinguish between notes and tool outcomes.
Permission cards for MCP tools can no longer be tricked into showing fake option rows.
The agent stdio mode now returns clearer errors for malformed client requests and consistent recap replies.
Scroll logging and acceleration calculations no longer produce backwards timestamps under load.
Hooks and tool commands started from the TUI can no longer read or write the interactive terminal.
Delete operations now correctly match Bash(rm …) ask/deny rules even when the path contains .. or the cwd is a symlink.
Web search and image approval cards now only offer always-approve mode to clients that support it.
Closing a session in the dashboard no longer leaves it listed under Inactive.
Running /compact on an empty session no longer resets the reported context window.
With goals disabled, a headless /goal line is now treated as a normal prompt instead of failing.
Background shell commands no longer flip back to running when late output arrives.
Only slash commands at the start of a message trigger skills.
Browser tabs restore and close more reliably with fewer races.
Fast model names in the footer no longer show double parentheses.
SubagentStart and Notification hook activity is no longer reported to the UI.
Voice error messages now display with correct punctuation.
disable_zdr_incompatible_tools changes now apply to new turns without restarting the worker.
Project skills are now labeled correctly for nested projects.

Performance

Opening the resume picker is now much faster on machines with many worktrees.
grok worktree commands start faster by skipping unnecessary model catalog work at boot.

source

Enterprise channel 1.0.52

2026-10-11 · enterprise · enterprise pointer is 1.0.52

1.0.52 — 2026-10-11

Features

Rex terminal support: Grok now reports its status (working, waiting for you, done, or failed) to Rex through OSC 7501, so Rex tabs show what each session is doing.
Tool call density improvements: expanded tool call groups now stay expanded as new calls join them, and Ctrl+E expands or collapses thinking faster in long transcripts.
Project hooks placed in a repository subfolder are now loaded when you launch inside that folder.
HTTP hooks to localhost now work, and bad hook URLs are reported at load time.
Command status-line scripts now see the current turn's prompt ID and start time while the turn is active.

Bug Fixes

Grep searches that time out or fail now show an error instead of appearing as empty results.
/model switches that fail no longer change the footer or saved default model.
Export and copy commands no longer overwrite an existing file; they refuse with a clear message.
Shell commands no longer keep running after you quit grok or it receives a signal.
Approval cards for background subagents no longer disappear when the parent turn ends.
grok worktree rm and gc no longer delete a worktree that an open session is using.
grok inspect no longer warns about valid pins that only requirements.toml accepts.
Pasted images that exceed the 50 MB send limit are now rejected with a clear message instead of silently dropped.
grok -p --effort now prints a warning when the model ignores the flag and refuses invalid effort levels.
Permission prompts now show which session needs approval.
grok worktree rm now refuses to delete git repositories, .git folders, or any directory grok did not create.
Grep and list_dir now report a clear error when the search root or directory cannot be read instead of silently returning no results.
Cancelled terminal commands now show the actual command that was stopped.
Wrapped URLs in prompts and quotes now link to the complete address instead of being cut at the wrap.
@mentions and image placeholders now respect Read deny rules and are omitted from the prompt.
Restarting after a model switch no longer reverts to the previous model.
Long tool turns that include PostToolUse hook output now compact correctly instead of hitting the prompt length limit.
Headless prompts that start with a byte-order mark no longer send the mark to the model or break JSON prompt files.
When a permission rule blocks a tool call, any context returned by a PreToolUse hook still reaches the model.
An invalid shell_environment_policy table in config.toml now causes grok to exit with a clear error instead of silently using the full environment.
MCP tool replies containing invalid characters no longer cause the call to time out.
Context overflow errors now trigger automatic compaction instead of failing the turn.
Inline tags in agent replies no longer break line layout, and /workflow stop now picks the correct live run.
Auto-compaction now stops after three context refills in a row instead of looping forever.
Minimal mode now shows the permission prompt instead of a hidden plan or cancel panel when both are present.
Project rules files that resolve outside the git root or workspace folder are now skipped.
[permission] rules now load alongside compact deny/allow/ask lists in config.toml.
Unrecognized permission keys or [permissions] sections now appear as skipped entries in grok inspect.
Approval cards now close the block viewer so the card is visible.
Shell redirects to wildcard or variable paths now correctly respect deny rules in Auto mode.
The in-session resume picker now matches the ungrouped layout shown on the welcome screen when configured.
HTTP MCP tools that return a JSON-RPC error are now executed only once.
StopFailure hook matchers containing commas or spaces are now treated as regex patterns.

Performance

Multiple Notification hook chains now start in parallel instead of waiting for the previous chain to finish.

source

Alpha channel 1.0.52

2026-10-11 · alpha · alpha pointer is 1.0.52

1.0.52 — 2026-10-11

Features

Rex terminal support: Grok now reports its status (working, waiting for you, done, or failed) to Rex through OSC 7501, so Rex tabs show what each session is doing.
Tool call density improvements: expanded tool call groups now stay expanded as new calls join them, and Ctrl+E expands or collapses thinking faster in long transcripts.
Project hooks placed in a repository subfolder are now loaded when you launch inside that folder.
HTTP hooks to localhost now work, and bad hook URLs are reported at load time.
Command status-line scripts now see the current turn's prompt ID and start time while the turn is active.

Bug Fixes

Grep searches that time out or fail now show an error instead of appearing as empty results.
/model switches that fail no longer change the footer or saved default model.
Export and copy commands no longer overwrite an existing file; they refuse with a clear message.
Shell commands no longer keep running after you quit grok or it receives a signal.
Approval cards for background subagents no longer disappear when the parent turn ends.
grok worktree rm and gc no longer delete a worktree that an open session is using.
grok inspect no longer warns about valid pins that only requirements.toml accepts.
Pasted images that exceed the 50 MB send limit are now rejected with a clear message instead of silently dropped.
grok -p --effort now prints a warning when the model ignores the flag and refuses invalid effort levels.
Permission prompts now show which session needs approval.
grok worktree rm now refuses to delete git repositories, .git folders, or any directory grok did not create.
Grep and list_dir now report a clear error when the search root or directory cannot be read instead of silently returning no results.
Cancelled terminal commands now show the actual command that was stopped.
Wrapped URLs in prompts and quotes now link to the complete address instead of being cut at the wrap.
@mentions and image placeholders now respect Read deny rules and are omitted from the prompt.
Restarting after a model switch no longer reverts to the previous model.
Long tool turns that include PostToolUse hook output now compact correctly instead of hitting the prompt length limit.
Headless prompts that start with a byte-order mark no longer send the mark to the model or break JSON prompt files.
When a permission rule blocks a tool call, any context returned by a PreToolUse hook still reaches the model.
An invalid shell_environment_policy table in config.toml now causes grok to exit with a clear error instead of silently using the full environment.
MCP tool replies containing invalid characters no longer cause the call to time out.
Context overflow errors now trigger automatic compaction instead of failing the turn.
Inline tags in agent replies no longer break line layout, and /workflow stop now picks the correct live run.
Auto-compaction now stops after three context refills in a row instead of looping forever.
Minimal mode now shows the permission prompt instead of a hidden plan or cancel panel when both are present.
Project rules files that resolve outside the git root or workspace folder are now skipped.
[permission] rules now load alongside compact deny/allow/ask lists in config.toml.
Unrecognized permission keys or [permissions] sections now appear as skipped entries in grok inspect.
Approval cards now close the block viewer so the card is visible.
Shell redirects to wildcard or variable paths now correctly respect deny rules in Auto mode.
The in-session resume picker now matches the ungrouped layout shown on the welcome screen when configured.
HTTP MCP tools that return a JSON-RPC error are now executed only once.
StopFailure hook matchers containing commas or spaces are now treated as regex patterns.

Performance

Multiple Notification hook chains now start in parallel instead of waiting for the previous chain to finish.

source

Enterprise channel 1.0.51

2026-10-09 · enterprise · enterprise pointer is 1.0.51

1.0.51 — 2026-10-09

Breaking Changes

Cancel and edit-resend on long sessions no longer unexpectedly revert workspace files.
Memory settings simplified: rollout and file_writes_enabled options removed; memory is on or off.
Removed the /share command and related session sharing features from the CLI and SDK.

Features

Inline viewports can now be restored at a previously saved screen area without extra cursor queries.
Headless grok -p now waits for background goals to finish and reports their outcome.
/imagine and similar commands now run the matching bundled skill when the required tools are unavailable.
New /density command and Tool call density setting let you choose how much detail tool calls show in the transcript.
Web fetch approval cards now include "always allow" and "never allow" domain options that persist for the folder.
Slash command completion now updates live when skills are added or changed during a session.
Request form now securely collects and fills sensitive details like passwords or card numbers without sending values to the model.
Chrome profiles can now be persisted across restarts using durable state directories.
Workspace server sessions now share one Chrome instance for cookies and storage.
/feedback Drafts tab now supports G / Home and End to jump to the first or last draft.
Pinned sessions on the dashboard now stay in the order you pinned or reordered them.
Shift+Enter, Ctrl+., Ctrl+/ and Ctrl+[ / ] now work inside tmux when extended-keys are enabled.
grok -p --output-format json now reports the exact worktree folder used (useful after -w name collisions).
Bare hook commands such as my-guard now run from PATH when not found next to the hook file.
Headless JSON error output now includes the worktree folder when a session was started.
grok sessions list and /resume now show worktree sessions started from the same subfolder.
/exit (and aliases) in a dashboard details view now closes only that session instead of quitting Grok.
Editors offering file access now receive a one-time transcript notice that the agent reads and writes disk files.
--background-wait-timeout now also stops an active headless goal after the timeout and exits with an error.
Interactive sessions now include the Grok Build user guide location so the agent can answer usage questions.
New /workflow commands let you list, stop, and resume background workflow runs without sending them to the model.
Background workflow runs now appear in the tasks pane with a stop button.
Hooks that ask about web searches or file deletes now show the hook name on the approval card.
config.toml now accepts an [update] section without breaking other commands.
Hook load errors listed in /hooks can now be expanded to see the full message.
grok worktree commands now work correctly inside Jujutsu repositories.
Images pasted with skill commands like /pr-workflow are now sent to the model.
Pressing Ctrl+C on a goal now pauses the goal instead of letting it continue.
grok -p now runs auth_provider_command when no login is stored.
Reopening a chat with an active goal now continues the goal automatically.
Browser tabs can now be restored on demand instead of all at once when a session resumes.
grok agent stdio now supports listing sessions via ACP and returns consistent reply envelopes.
Answering an approval with a message now tells the model it is a follow-up.
--allowed-tools is now accepted as an alias for --allow in headless mode.

Bug Fixes

Slash command suggestions no longer make the text after the cursor disappear while typing.
Running workflows in the dock now show the same pulsing indicator as in the main scrollback.
Mode switch tool calls now appear in the transcript with the target mode name.
Always-approve mode now automatically approves web requests without extra prompts.
The follow indicator no longer appears when only blank padding sits below the viewport.
Search failures now keep the original pattern instead of showing an empty Search "".
Sessions that open without a model list now use the public Auto id and succeed on every account.
Background wake turns that request plan changes or mode switches are now handled correctly.
Model-call retries now show progress status instead of a generic waiting message.
Foreground subagent activity now appears correctly after resume.
Very large pasted prompts are now offloaded to a file instead of being sent inline.
Sign-in failures now show a clear plain-English message with the next step instead of raw errors.
Logging out now stops any running turns on local sessions instead of leaving them active.
The send-now shortcut no longer accidentally enables always-approve mode while a permission card is open.
/find now keeps match counts and navigation accurate while replies or command output stream in.
Images from tools are now validated so bad files no longer break the current turn or future replays.
Screenshots and other image artifacts saved by the browser tool are now stored as real image files.
Rate-limit errors now show the expected message; very small images are rejected early.
Send-now (Ctrl+Enter) no longer accidentally answers an open question card.
Opening the queue with Up now scrolls so the selected follow-up is visible.
Command output that contains clear-screen codes is no longer discarded.
Dashboard cursor and overlay counter now correctly track the active grouping view.
Headless JSON output now reports the correct stop reason when the turn limit is reached and removes duplicate error prefixes.
Session picker footer no longer lists the same shortcut twice.
Dragging images from a file manager onto terminals like GNOME Terminal or xfce4-terminal now attaches them correctly.
Plan comments added while a revision runs are now preserved and sent when the plan is approved.
Permission prompts that fail now show a clearer message without blaming the interface.
--disallowed-tools now accepts the exact names shown to the model (run_terminal_command, spawn_subagent, etc.).
Long pastes without bracketed-paste markers are now read completely instead of stalling after ~1 KB.
A feedback send now shows "Sending feedback…" immediately; failed sends no longer repeat the lead phrase.
Typing, arrows or Enter while an approval card appears no longer accidentally approve the action or enable always-approve mode.
Error hints for legacy session import now appear in full instead of being cut off by the pager.
Skills in the folder where grok started now correctly show as (local).
Worktree creation no longer hangs when the repo contains named pipes or sockets.
--no-subagents flag now works correctly in headless / -p sessions.
Resumed sessions now correctly show aborted subagent tasks as cancelled instead of failed.
Large parallel shell batches no longer drop calls with "hook state lost" errors.
Write approval cards now show the existing file contents that will be replaced.
Slash command completion now shows the correct name for files under commands/ instead of always showing /commands.
PreToolUse hooks now validate rewrites against the tool schema and deny invalid ones before running the tool.
Ctrl+C during a turn now cancels the turn even when a modal such as /mcps or the block viewer is open.
grok -w now opens the session at the worktree root (with a note) when the original launch folder does not exist in the new worktree.
/cd and the location picker now accept quoted or escaped folder paths.
Web search hooks now fire under the correct web_search name and carry query.
Large file writes or many screenshots no longer end the session.
Glob tool hooks now match under list_dir and carry the correct directory.
/find now scrolls to the exact wrapped row containing the current match.
Downloads via FetchMcpResource are now refused in plan mode and resolve paths consistently.
Pinned dashboard sessions now survive restarts and layout changes instead of disappearing from the saved list.
Slash commands like /imagine now work regardless of capitalization and correctly substitute skills when tools are missing.
Turning off bypass mode without naming a session now correctly restores prompts in all open sessions.
/delete with extra text now shows an error instead of deleting the current session.
Failed /feedback status messages now wrap inside the dialog instead of clipping at the border.
Number-key picks in the dashboard peek no longer leak to the wrong agent after Up/Down navigation.
Worktrees created with -w are now correctly tracked and resumable even when GROK_HOME is reached through a symlink.
grok worktree rm and gc now clean up names after manually deleting a worktree folder so -w NAME works again.
Hook scripts that read stdin with read now receive a complete line.
A policy that disables non-managed hooks now applies to the next session instead of only after restart.
/resume and grok sessions list no longer show worktree sessions from a sibling submodule or separate-git-dir repo.
A hook listed under two different matchers now runs only once per tool call.
Typing a hidden command such as /cd in a session and pressing Enter now runs it instead of accepting a near-match skill.
@-file completion no longer accepts a match from an earlier query when you type quickly.
Tool hooks now run one at a time per session and stop promptly on cancel or exit.
PostToolUse hooks now receive a correct whole-millisecond duration.
PreToolUse hooks for shell commands now see the model's block_until_ms value.
MCP tool calls now show hook permission prompts and cards name the hook that asked.
Typing while a permission or ask card is visible no longer selects or answers the card.
grok worktree list now shows unfinished copies left by an interrupted create.
Auto mode now asks before deleting files that the classifier has not already allowed.
Fixed intermittent List/Read/Open failures during concurrent v2 hardening migrations.
grok -p -s with a reused session id now shows the plain refusal message instead of JSON-RPC noise.
Settings saves no longer re-insert the built-in default fork model into config.toml.
Fixed several pager rows that showed wrong paths, truncated names, or hid failure/cancel states.
--system-prompt now correctly appends rules from --append-system-prompt instead of dropping them.
/new now inherits always-approve mode from the tab it was run from instead of another tab.
Multi-select questions now let you toggle several answers with number keys before submitting.
Slash-command notices typed while a reply streams no longer hide the streaming reply.
@ file search no longer accepts keys or moves the highlight while results are still updating.
Worktree db rebuild now restores names so rm and show by name work again.
A prompt blocked by a hook no longer titles the session.
Background subagents now respect saved approvals and --always-approve after the parent turn finishes.
Per-session system prompt overrides now correctly use only the session's own rules.
Connection stalled errors now show a clearer message with network advice.
Invalid model provider tables now produce a clear startup error instead of silent fallback.
Narrow feedback dialog now shows full draft titles and readable confirm prompts.
/workflow stop|resume <name> now correctly targets the live run instead of an ended first run.
Dialogs like /feedback no longer disappear on very short terminal windows.
Delete operations now respect the same permission rules as shell rm commands.
Hooks can no longer rewrite a web-search call to restrict domains.
Workspace server now starts with per-session private Chrome by default.
Notification hooks for permission prompts and task completion now run reliably before a headless session exits.
Headless goal runs no longer print a spurious "Goal paused" message after completion.
Legacy UI density settings now combine correctly across config tiers and produce expected folding.
Approval answers sent after a turn ends no longer affect the turn.
Background shell commands now fail immediately when the working directory does not exist.
grok worktree rm without -f now safely keeps worktrees when the unsaved-work check fails or times out.
The dashboard slash menu now shows skills from the home session even when opened early.
grok worktree rm and gc now correctly remove worktrees whose only untracked content is empty folders.
Editing a queued prompt now prevents accidental removal with a stale client view.
The model is now told when memory is turned off or on for the session.
Hook results in the UI now correctly distinguish between notes and tool outcomes.
Permission cards for MCP tools can no longer be tricked into showing fake option rows.
The agent stdio mode now returns clearer errors for malformed client requests and consistent recap replies.
Scroll logging and acceleration calculations no longer produce backwards timestamps under load.
Hooks and tool commands started from the TUI can no longer read or write the interactive terminal.
Delete operations now correctly match Bash(rm …) ask/deny rules even when the path contains .. or the cwd is a symlink.
Web search and image approval cards now only offer always-approve mode to clients that support it.
Closing a session in the dashboard no longer leaves it listed under Inactive.
Running /compact on an empty session no longer resets the reported context window.
With goals disabled, a headless /goal line is now treated as a normal prompt instead of failing.
Background shell commands no longer flip back to running when late output arrives.
Only slash commands at the start of a message trigger skills.
Browser tabs restore and close more reliably with fewer races.
Fast model names in the footer no longer show double parentheses.
SubagentStart and Notification hook activity is no longer reported to the UI.
Voice error messages now display with correct punctuation.
disable_zdr_incompatible_tools changes now apply to new turns without restarting the worker.
Project skills are now labeled correctly for nested projects.

Performance

Opening the resume picker is now much faster on machines with many worktrees.
grok worktree commands start faster by skipping unnecessary model catalog work at boot.

source

Stable channel 1.0.50

2026-10-06 · stable · stable pointer is 1.0.50

1.0.50 — 2026-10-06

Breaking Changes

grok worktree rm without -f now refuses to delete worktrees that are in use or contain unsaved changes and exits with an error.

Features

Compact mode now collapses recaps and keeps inner spacing on tall terminals.
Clicking the git branch in the header now copies the full branch name.
Long branch names now abbreviate on narrow terminals while remaining copyable.
Targeted cancels now stop only the specific response when requested by clients.
Compact mode now hides the dock (or tasks pane) the same way Ctrl+G does.
New timestamps setting offers off, minimal, and on modes in Settings and /timestamps.
Shell command output now shows "Running …" while executing and "Ran …" when finished.
Live preview pins now include whether an app was observed behind the URL.
Command palette (Ctrl+P) and model picker now open at the same width as the settings modal.
Tables rendered by grok now show copy buttons underneath that export clean TSV, CSV or Markdown suitable for spreadsheets.
Dashboard now shows which worktree each session belongs to, displays the -m model in the new-agent box, and lets you toggle worktree mode in the folder picker.

Bug Fixes

Shell permission decisions now correctly prompt or approve commands matching the permission manager.
Cancelled turns now correctly mark unanswered tool calls as cancelled instead of succeeded.
Long error messages in the /usage and /session-info modals now wrap instead of being truncated.
Em-dashes no longer appear in status messages and subagent rows.
Stalled model calls during turns now retry before failing the turn.
Queue pane and Ctrl+; now appear and work when the dock is hidden with Ctrl+G.
Long sessions no longer disconnect when the conversation exceeds 4 MiB.
Permission rules now accept native tool names such as run_terminal_command and Shell.
/rewind now places the cursor at the end of the restored prompt.
Failed file writes now leave the previous contents instead of an empty file.
Fixed /loop and scheduler_create so non-ASCII intervals like 5分 no longer crash the session.
Fixed --disallowed-tools when only one scheduler tool is named; the other two no longer cause startup failure.
Fixed closing a busy tab so queued prompts are no longer sent to the model.
Fixed relative --cwd so grok --cwd a no longer ends up running inside a/a.
The terminal title now shows "Action Required" for any background session waiting on approval.
grok now accepts a prompt via piped stdin in headless mode instead of trying to open the UI.
An admin-set [features] title_refresh = false in requirements.toml now correctly disables title refresh even when GROK_TITLE_REFRESH is set.
Custom notification commands defined under [[ui.notifications.hooks]] now execute on matching events.
An admin pin of [cli] use_leader = false now correctly prevents --leader from starting a leader process.
When multiple MCP servers expose a tool with the same name, search_tool now returns all of them.
Concurrent Grok windows editing settings no longer lose or corrupt config.toml entries such as MCP servers or permission rules.
web_fetch no longer downloads huge files into memory before rejecting them.
memory_get no longer returns huge files that exceed the read limit.
Continue to run no longer kills commands running inside background subagents.
memory_search now finds Japanese, Chinese, and Korean words inside longer notes.
Drafts saved by the model now appear under their actual titles in /feedback.
Imagine replies that are not valid images are now rejected instead of being saved as .jpg files.
A workflow file that fails to parse now reports the exact file and parse error instead of 'unknown workflow'.
Re-encoded images are now oriented correctly using their EXIF tags instead of appearing sideways.
Image and log paths reported after /fork --worktree now point to real files instead of missing locations.
Tab characters in terminal command output are now preserved when shown to the model.
Transparent images are now kept as PNG when re-encoded instead of turning black.
The bundled ripgrep is now available to terminal commands even on images that do not install it.
Request bodies now stay under a small configured max_request_bytes limit instead of growing past it.
Effort level now stays the same after a bare /model switch instead of jumping to the new model's default.
Ctrl+\ and other modified digits no longer accidentally select options on permission or cancel cards.
Feedback submissions now show accurate success or failure messages without premature thanks or duplicated prefixes.
grok doctor now reports the exact config file or setting that prevents Grok from starting instead of saying zero issues.
Recap summaries now appear on screen automatically instead of requiring manual scrolling.
grok doctor fix now opens its confirmation card on Cancel to prevent accidental edits to shell or tmux config.
Slash commands now run regardless of case (e.g. /COMPACT or /Compact).
/context no longer shows an invented auto-compact threshold when the agent does not report one.
Agent text no longer keeps the pager redrawing when a tool call interrupts the sentence.
CLI tool calls no longer fail or overflow when the workspace directory is deleted or output is huge.
Logs written to files or when NO_COLOR is set are now plain text instead of containing color codes.
Prompts now fail immediately with a clear message if the session's working directory no longer exists instead of failing later or recreating it.
grok -w now works correctly with separate-git-dir repositories and shows clear errors when run outside a git repo or on an empty repo.
Shell tool descriptions now correctly state the actual output truncation limit, and grep respects globs relative to the current session directory.
Compaction messages now scroll into view instead of appearing below the bottom of the screen after a turn finishes.
Git-mode worktrees created with -w now preserve uncommitted changes instead of dropping them.
Worktrees created with --worktree-ref no longer contain wrong submodule contents or LFS pointer text.
Worktree creation now shows the same clear error messages on every code path.
Failed worktree creates now show a clear error on the home screen instead of being hidden.
Merge and rebase conflicts now appear in the changes list instead of being hidden.
Fixed a crash when starting a new session with durable rewind checkpoints enabled.
Managed permission files with one bad key no longer lose all their valid rules.
Simultaneous worktree creates with the same label now succeed instead of one failing with "already exists".
grok worktree rm without -f no longer deletes arbitrary folders that happen to share a name with a removed worktree.
Discard now affects only the exact files you select instead of also deleting or reverting similarly-named files.
Permission rules with one bad entry no longer lose the other rules in the same section.
Chat Completions streams that end without a finish reason are now retried instead of returning partial output.
grok worktree rm by label now refuses when the label exists in multiple repos.
Grep now respects Read deny rules that use a leading ./.

Performance

Command hooks no longer cause large memory spikes when they print a lot of output.
Retry waits now vary between processes instead of being identical.

source

Enterprise channel 1.0.50

2026-10-06 · enterprise · enterprise pointer is 1.0.50

1.0.50 — 2026-10-06

Breaking Changes

grok worktree rm without -f now refuses to delete worktrees that are in use or contain unsaved changes and exits with an error.

Features

Compact mode now collapses recaps and keeps inner spacing on tall terminals.
Clicking the git branch in the header now copies the full branch name.
Long branch names now abbreviate on narrow terminals while remaining copyable.
Targeted cancels now stop only the specific response when requested by clients.
Compact mode now hides the dock (or tasks pane) the same way Ctrl+G does.
New timestamps setting offers off, minimal, and on modes in Settings and /timestamps.
Shell command output now shows "Running …" while executing and "Ran …" when finished.
Live preview pins now include whether an app was observed behind the URL.
Command palette (Ctrl+P) and model picker now open at the same width as the settings modal.
Tables rendered by grok now show copy buttons underneath that export clean TSV, CSV or Markdown suitable for spreadsheets.
Dashboard now shows which worktree each session belongs to, displays the -m model in the new-agent box, and lets you toggle worktree mode in the folder picker.

Bug Fixes

Shell permission decisions now correctly prompt or approve commands matching the permission manager.
Cancelled turns now correctly mark unanswered tool calls as cancelled instead of succeeded.
Long error messages in the /usage and /session-info modals now wrap instead of being truncated.
Em-dashes no longer appear in status messages and subagent rows.
Stalled model calls during turns now retry before failing the turn.
Queue pane and Ctrl+; now appear and work when the dock is hidden with Ctrl+G.
Long sessions no longer disconnect when the conversation exceeds 4 MiB.
Permission rules now accept native tool names such as run_terminal_command and Shell.
/rewind now places the cursor at the end of the restored prompt.
Failed file writes now leave the previous contents instead of an empty file.
Fixed /loop and scheduler_create so non-ASCII intervals like 5分 no longer crash the session.
Fixed --disallowed-tools when only one scheduler tool is named; the other two no longer cause startup failure.
Fixed closing a busy tab so queued prompts are no longer sent to the model.
Fixed relative --cwd so grok --cwd a no longer ends up running inside a/a.
The terminal title now shows "Action Required" for any background session waiting on approval.
grok now accepts a prompt via piped stdin in headless mode instead of trying to open the UI.
An admin-set [features] title_refresh = false in requirements.toml now correctly disables title refresh even when GROK_TITLE_REFRESH is set.
Custom notification commands defined under [[ui.notifications.hooks]] now execute on matching events.
An admin pin of [cli] use_leader = false now correctly prevents --leader from starting a leader process.
When multiple MCP servers expose a tool with the same name, search_tool now returns all of them.
Concurrent Grok windows editing settings no longer lose or corrupt config.toml entries such as MCP servers or permission rules.
web_fetch no longer downloads huge files into memory before rejecting them.
memory_get no longer returns huge files that exceed the read limit.
Continue to run no longer kills commands running inside background subagents.
memory_search now finds Japanese, Chinese, and Korean words inside longer notes.
Drafts saved by the model now appear under their actual titles in /feedback.
Imagine replies that are not valid images are now rejected instead of being saved as .jpg files.
A workflow file that fails to parse now reports the exact file and parse error instead of 'unknown workflow'.
Re-encoded images are now oriented correctly using their EXIF tags instead of appearing sideways.
Image and log paths reported after /fork --worktree now point to real files instead of missing locations.
Tab characters in terminal command output are now preserved when shown to the model.
Transparent images are now kept as PNG when re-encoded instead of turning black.
The bundled ripgrep is now available to terminal commands even on images that do not install it.
Request bodies now stay under a small configured max_request_bytes limit instead of growing past it.
Effort level now stays the same after a bare /model switch instead of jumping to the new model's default.
Ctrl+\ and other modified digits no longer accidentally select options on permission or cancel cards.
Feedback submissions now show accurate success or failure messages without premature thanks or duplicated prefixes.
grok doctor now reports the exact config file or setting that prevents Grok from starting instead of saying zero issues.
Recap summaries now appear on screen automatically instead of requiring manual scrolling.
grok doctor fix now opens its confirmation card on Cancel to prevent accidental edits to shell or tmux config.
Slash commands now run regardless of case (e.g. /COMPACT or /Compact).
/context no longer shows an invented auto-compact threshold when the agent does not report one.
Agent text no longer keeps the pager redrawing when a tool call interrupts the sentence.
CLI tool calls no longer fail or overflow when the workspace directory is deleted or output is huge.
Logs written to files or when NO_COLOR is set are now plain text instead of containing color codes.
Prompts now fail immediately with a clear message if the session's working directory no longer exists instead of failing later or recreating it.
grok -w now works correctly with separate-git-dir repositories and shows clear errors when run outside a git repo or on an empty repo.
Shell tool descriptions now correctly state the actual output truncation limit, and grep respects globs relative to the current session directory.
Compaction messages now scroll into view instead of appearing below the bottom of the screen after a turn finishes.
Git-mode worktrees created with -w now preserve uncommitted changes instead of dropping them.
Worktrees created with --worktree-ref no longer contain wrong submodule contents or LFS pointer text.
Worktree creation now shows the same clear error messages on every code path.
Failed worktree creates now show a clear error on the home screen instead of being hidden.
Merge and rebase conflicts now appear in the changes list instead of being hidden.
Fixed a crash when starting a new session with durable rewind checkpoints enabled.
Managed permission files with one bad key no longer lose all their valid rules.
Simultaneous worktree creates with the same label now succeed instead of one failing with "already exists".
grok worktree rm without -f no longer deletes arbitrary folders that happen to share a name with a removed worktree.
Discard now affects only the exact files you select instead of also deleting or reverting similarly-named files.
Permission rules with one bad entry no longer lose the other rules in the same section.
Chat Completions streams that end without a finish reason are now retried instead of returning partial output.
grok worktree rm by label now refuses when the label exists in multiple repos.
Grep now respects Read deny rules that use a leading ./.

Performance

Command hooks no longer cause large memory spikes when they print a lot of output.
Retry waits now vary between processes instead of being identical.

source

Alpha channel 1.0.50

2026-10-06 · alpha · alpha pointer is 1.0.50

1.0.50 — 2026-10-06

Breaking Changes

grok worktree rm without -f now refuses to delete worktrees that are in use or contain unsaved changes and exits with an error.

Features

Compact mode now collapses recaps and keeps inner spacing on tall terminals.
Clicking the git branch in the header now copies the full branch name.
Long branch names now abbreviate on narrow terminals while remaining copyable.
Targeted cancels now stop only the specific response when requested by clients.
Compact mode now hides the dock (or tasks pane) the same way Ctrl+G does.
New timestamps setting offers off, minimal, and on modes in Settings and /timestamps.
Shell command output now shows "Running …" while executing and "Ran …" when finished.
Live preview pins now include whether an app was observed behind the URL.
Command palette (Ctrl+P) and model picker now open at the same width as the settings modal.
Tables rendered by grok now show copy buttons underneath that export clean TSV, CSV or Markdown suitable for spreadsheets.
Dashboard now shows which worktree each session belongs to, displays the -m model in the new-agent box, and lets you toggle worktree mode in the folder picker.

Bug Fixes

Shell permission decisions now correctly prompt or approve commands matching the permission manager.
Cancelled turns now correctly mark unanswered tool calls as cancelled instead of succeeded.
Long error messages in the /usage and /session-info modals now wrap instead of being truncated.
Em-dashes no longer appear in status messages and subagent rows.
Stalled model calls during turns now retry before failing the turn.
Queue pane and Ctrl+; now appear and work when the dock is hidden with Ctrl+G.
Long sessions no longer disconnect when the conversation exceeds 4 MiB.
Permission rules now accept native tool names such as run_terminal_command and Shell.
/rewind now places the cursor at the end of the restored prompt.
Failed file writes now leave the previous contents instead of an empty file.
Fixed /loop and scheduler_create so non-ASCII intervals like 5分 no longer crash the session.
Fixed --disallowed-tools when only one scheduler tool is named; the other two no longer cause startup failure.
Fixed closing a busy tab so queued prompts are no longer sent to the model.
Fixed relative --cwd so grok --cwd a no longer ends up running inside a/a.
The terminal title now shows "Action Required" for any background session waiting on approval.
grok now accepts a prompt via piped stdin in headless mode instead of trying to open the UI.
An admin-set [features] title_refresh = false in requirements.toml now correctly disables title refresh even when GROK_TITLE_REFRESH is set.
Custom notification commands defined under [[ui.notifications.hooks]] now execute on matching events.
An admin pin of [cli] use_leader = false now correctly prevents --leader from starting a leader process.
When multiple MCP servers expose a tool with the same name, search_tool now returns all of them.
Concurrent Grok windows editing settings no longer lose or corrupt config.toml entries such as MCP servers or permission rules.
web_fetch no longer downloads huge files into memory before rejecting them.
memory_get no longer returns huge files that exceed the read limit.
Continue to run no longer kills commands running inside background subagents.
memory_search now finds Japanese, Chinese, and Korean words inside longer notes.
Drafts saved by the model now appear under their actual titles in /feedback.
Imagine replies that are not valid images are now rejected instead of being saved as .jpg files.
A workflow file that fails to parse now reports the exact file and parse error instead of 'unknown workflow'.
Re-encoded images are now oriented correctly using their EXIF tags instead of appearing sideways.
Image and log paths reported after /fork --worktree now point to real files instead of missing locations.
Tab characters in terminal command output are now preserved when shown to the model.
Transparent images are now kept as PNG when re-encoded instead of turning black.
The bundled ripgrep is now available to terminal commands even on images that do not install it.
Request bodies now stay under a small configured max_request_bytes limit instead of growing past it.
Effort level now stays the same after a bare /model switch instead of jumping to the new model's default.
Ctrl+\ and other modified digits no longer accidentally select options on permission or cancel cards.
Feedback submissions now show accurate success or failure messages without premature thanks or duplicated prefixes.
grok doctor now reports the exact config file or setting that prevents Grok from starting instead of saying zero issues.
Recap summaries now appear on screen automatically instead of requiring manual scrolling.
grok doctor fix now opens its confirmation card on Cancel to prevent accidental edits to shell or tmux config.
Slash commands now run regardless of case (e.g. /COMPACT or /Compact).
/context no longer shows an invented auto-compact threshold when the agent does not report one.
Agent text no longer keeps the pager redrawing when a tool call interrupts the sentence.
CLI tool calls no longer fail or overflow when the workspace directory is deleted or output is huge.
Logs written to files or when NO_COLOR is set are now plain text instead of containing color codes.
Prompts now fail immediately with a clear message if the session's working directory no longer exists instead of failing later or recreating it.
grok -w now works correctly with separate-git-dir repositories and shows clear errors when run outside a git repo or on an empty repo.
Shell tool descriptions now correctly state the actual output truncation limit, and grep respects globs relative to the current session directory.
Compaction messages now scroll into view instead of appearing below the bottom of the screen after a turn finishes.
Git-mode worktrees created with -w now preserve uncommitted changes instead of dropping them.
Worktrees created with --worktree-ref no longer contain wrong submodule contents or LFS pointer text.
Worktree creation now shows the same clear error messages on every code path.
Failed worktree creates now show a clear error on the home screen instead of being hidden.
Merge and rebase conflicts now appear in the changes list instead of being hidden.
Fixed a crash when starting a new session with durable rewind checkpoints enabled.
Managed permission files with one bad key no longer lose all their valid rules.
Simultaneous worktree creates with the same label now succeed instead of one failing with "already exists".
grok worktree rm without -f no longer deletes arbitrary folders that happen to share a name with a removed worktree.
Discard now affects only the exact files you select instead of also deleting or reverting similarly-named files.
Permission rules with one bad entry no longer lose the other rules in the same section.
Chat Completions streams that end without a finish reason are now retried instead of returning partial output.
grok worktree rm by label now refuses when the label exists in multiple repos.
Grep now respects Read deny rules that use a leading ./.

Performance

Command hooks no longer cause large memory spikes when they print a lot of output.
Retry waits now vary between processes instead of being identical.

source

Enterprise channel 1.0.49

2026-10-02 · enterprise · enterprise pointer is 1.0.49

1.0.49 — 2026-10-02

Bug Fixes

Hook permission prompts now appear regardless of approval mode.
Web searches, fetches, and image generation now correctly respect the prompt_policy setting and rule decisions before prompting.
Allow once in permission cards now applies only to the next shell command instead of the whole session.
pkill -f and pgrep -f commands are now rejected earlier when the pattern would kill their own shell wrapper.

source

Alpha channel 1.0.49

2026-10-02 · alpha · alpha pointer is 1.0.49

1.0.49 — 2026-10-02

Bug Fixes

Hook permission prompts now appear regardless of approval mode.
Web searches, fetches, and image generation now correctly respect the prompt_policy setting and rule decisions before prompting.
Allow once in permission cards now applies only to the next shell command instead of the whole session.
pkill -f and pgrep -f commands are now rejected earlier when the pattern would kill their own shell wrapper.

source

Enterprise channel 1.0.48

2026-10-02 · enterprise · enterprise pointer is 1.0.48

1.0.48 — 2026-10-02

Bug Fixes

/rewind now only lists prompts you typed yourself and skips automatic runtime turns.
Custom model tables in config.toml now preserve the original model's notice banner.
read_file, search_replace, and grep now handle large files, special files, and binary content more reliably and surface better errors.
Ask rules for WebSearch and WebFetch now correctly show a prompt instead of auto-running.
Cancelling one tool call no longer kills commands started by other parallel calls.
Authentication recovery no longer automatically retries refused prompts after login; the sign-in block now requires manual resend.

source

Alpha channel 1.0.48

2026-10-02 · alpha · alpha pointer is 1.0.48

1.0.48 — 2026-10-02

Bug Fixes

/rewind now only lists prompts you typed yourself and skips automatic runtime turns.
Custom model tables in config.toml now preserve the original model's notice banner.
read_file, search_replace, and grep now handle large files, special files, and binary content more reliably and surface better errors.
Ask rules for WebSearch and WebFetch now correctly show a prompt instead of auto-running.
Cancelling one tool call no longer kills commands started by other parallel calls.
Authentication recovery no longer automatically retries refused prompts after login; the sign-in block now requires manual resend.

source

Alpha channel 1.0.47

2026-10-01 · alpha · alpha pointer is 1.0.47

1.0.47 — 2026-10-01

Features

Model switch responses from the agent now update the displayed model list and context window.

Bug Fixes

Error messages on the home and login screens now wrap properly in narrow terminals.
Clicking the Goal chip now opens the goal panel even when workflow runs are present.

Performance

Sessions in repositories with an .envrc file now start faster.

source

Stable channel 1.0.46

2026-09-30 · stable · stable pointer is 1.0.46

1.0.46 — 2026-09-30

Bug Fixes

MCP server sources are now correctly reported by grok inspect and the MCP doctor.
Tool search activity now correctly folds as "Ran N tool searches" instead of claiming MCP tool results.
Client rules supplied at session start are now preserved when the system prompt is rebuilt later.
Permission rules with relative paths now correctly apply even when the cwd contains symlinks.
Context compacted line now shows the correct token count immediately after compaction instead of at the end of a long turn.
Enter on video cards no longer attempts playback in terminals that cannot show it or lack ffmpeg.
Video cards now open the file in the default player on Enter in terminals without inline playback support, and no longer get stuck trying to extract posters without ffmpeg.

Performance

list_dir on network or virtual filesystems now skips slow subtree counts and prints a notice instead.
Session startup is faster when loading skills from large directories and shows what it is doing instead of a generic spinner.

source

Enterprise channel 1.0.46

2026-09-30 · enterprise · enterprise pointer is 1.0.46

1.0.46 — 2026-09-30

Bug Fixes

MCP server sources are now correctly reported by grok inspect and the MCP doctor.
Tool search activity now correctly folds as "Ran N tool searches" instead of claiming MCP tool results.
Client rules supplied at session start are now preserved when the system prompt is rebuilt later.
Permission rules with relative paths now correctly apply even when the cwd contains symlinks.
Context compacted line now shows the correct token count immediately after compaction instead of at the end of a long turn.
Enter on video cards no longer attempts playback in terminals that cannot show it or lack ffmpeg.
Video cards now open the file in the default player on Enter in terminals without inline playback support, and no longer get stuck trying to extract posters without ffmpeg.

Performance

list_dir on network or virtual filesystems now skips slow subtree counts and prints a notice instead.
Session startup is faster when loading skills from large directories and shows what it is doing instead of a generic spinner.

source

Alpha channel 1.0.46

2026-09-30 · alpha · alpha pointer is 1.0.46

1.0.46 — 2026-09-30

Bug Fixes

MCP server sources are now correctly reported by grok inspect and the MCP doctor.
Tool search activity now correctly folds as "Ran N tool searches" instead of claiming MCP tool results.
Client rules supplied at session start are now preserved when the system prompt is rebuilt later.
Permission rules with relative paths now correctly apply even when the cwd contains symlinks.
Context compacted line now shows the correct token count immediately after compaction instead of at the end of a long turn.
Enter on video cards no longer attempts playback in terminals that cannot show it or lack ffmpeg.
Video cards now open the file in the default player on Enter in terminals without inline playback support, and no longer get stuck trying to extract posters without ffmpeg.

Performance

list_dir on network or virtual filesystems now skips slow subtree counts and prints a notice instead.
Session startup is faster when loading skills from large directories and shows what it is doing instead of a generic spinner.

source

Stable channel 1.0.45

2026-09-29 · stable · stable pointer is 1.0.45

1.0.45 — 2026-09-29

Features

Custom agents from plugins or your config can now be chosen directly with spawn_subagent.
MCP servers can now use a token file that is re-read on every request so rotating credentials stay fresh.
Models can now show a colored notice banner above the prompt while selected.

Bug Fixes

Subagent wait status now correctly shows "Waiting for N subagents…" when the parent turn is blocked.

source

Enterprise channel 1.0.45

2026-09-29 · enterprise · enterprise pointer is 1.0.45

1.0.45 — 2026-09-29

Features

Custom agents from plugins or your config can now be chosen directly with spawn_subagent.
MCP servers can now use a token file that is re-read on every request so rotating credentials stay fresh.
Models can now show a colored notice banner above the prompt while selected.

Bug Fixes

Subagent wait status now correctly shows "Waiting for N subagents…" when the parent turn is blocked.

source

Alpha channel 1.0.45

2026-09-29 · alpha · alpha pointer is 1.0.45

1.0.45 — 2026-09-29

Features

Custom agents from plugins or your config can now be chosen directly with spawn_subagent.
MCP servers can now use a token file that is re-read on every request so rotating credentials stay fresh.
Models can now show a colored notice banner above the prompt while selected.

Bug Fixes

Subagent wait status now correctly shows "Waiting for N subagents…" when the parent turn is blocked.

source

Stable channel 1.0.44

2026-09-28 · stable · stable pointer is 1.0.44

1.0.44 — 2026-09-28

Features

Sandbox enforcement on macOS now uses the system Seatbelt (sandbox-exec) backend.
Models can now advertise multiple context window sizes; the first listed value is the default.
New /context-window slash command lets you pick a context window size for the current session when the model offers choices.
/model selection now offers context window choices (when the model supports them) before the effort level.

Bug Fixes

Plan comments can now be deleted by clicking the [✗] button that appears on hover.
Shortcuts panel search now finds entries when you type any of the words in the label, keys, description or help text.
Fixed crashes when pasting on Windows caused by overlapping clipboard operations.

Performance

Parallel tool calls that edit the same file now run safely in sequence; unrelated tools stay concurrent.

source

Enterprise channel 1.0.44

2026-09-28 · enterprise · enterprise pointer is 1.0.44

1.0.44 — 2026-09-28

Features

Sandbox enforcement on macOS now uses the system Seatbelt (sandbox-exec) backend.
Models can now advertise multiple context window sizes; the first listed value is the default.
New /context-window slash command lets you pick a context window size for the current session when the model offers choices.
/model selection now offers context window choices (when the model supports them) before the effort level.

Bug Fixes

Plan comments can now be deleted by clicking the [✗] button that appears on hover.
Shortcuts panel search now finds entries when you type any of the words in the label, keys, description or help text.
Fixed crashes when pasting on Windows caused by overlapping clipboard operations.

Performance

Parallel tool calls that edit the same file now run safely in sequence; unrelated tools stay concurrent.

source

Alpha channel 1.0.44

2026-09-28 · alpha · alpha pointer is 1.0.44

1.0.44 — 2026-09-28

Features

Sandbox enforcement on macOS now uses the system Seatbelt (sandbox-exec) backend.
Models can now advertise multiple context window sizes; the first listed value is the default.
New /context-window slash command lets you pick a context window size for the current session when the model offers choices.
/model selection now offers context window choices (when the model supports them) before the effort level.

Bug Fixes

Plan comments can now be deleted by clicking the [✗] button that appears on hover.
Shortcuts panel search now finds entries when you type any of the words in the label, keys, description or help text.
Fixed crashes when pasting on Windows caused by overlapping clipboard operations.

Performance

Parallel tool calls that edit the same file now run safely in sequence; unrelated tools stay concurrent.

source

Stable channel 1.0.42

2026-09-26 · stable · stable pointer is 1.0.42

1.0.42 — 2026-09-26

Features

Footer and /usage now display the model actually served on smart-auto turns.
New grok worktree create command creates a managed worktree without launching an interactive session.
Auto permission mode now appears as Auto-review everywhere in the UI.

Bug Fixes

Bracketed pastes that did not come from the clipboard no longer attach leftover images (e.g. from IME commits).
grok -p now exits cleanly with the documented code when interrupted and performs session cleanup instead of being killed.
Pressing Ctrl+C twice while writing a plan comment now cancels the comment instead of getting stuck.
grok --minimal now reprints history correctly after terminal resize and keeps the prompt/status rows anchored.
Restoring a stashed draft now places the cursor at the end so continued typing appends correctly.
grok update on WinGet installs now prints the winget upgrade command and exits without writing files.
Inline images are re-uploaded after a full repaint so they no longer appear blank in terminals that drop image data on ESC[2J.
Plugin hooks now run on a fresh session without requiring an explicit reload.
Fixed plugin auto-update detection when marketplace URLs differ only by .git suffix or host case.

source

Enterprise channel 1.0.42

2026-09-26 · enterprise · enterprise pointer is 1.0.42

1.0.42 — 2026-09-26

Features

Footer and /usage now display the model actually served on smart-auto turns.
New grok worktree create command creates a managed worktree without launching an interactive session.
Auto permission mode now appears as Auto-review everywhere in the UI.

Bug Fixes

Bracketed pastes that did not come from the clipboard no longer attach leftover images (e.g. from IME commits).
grok -p now exits cleanly with the documented code when interrupted and performs session cleanup instead of being killed.
Pressing Ctrl+C twice while writing a plan comment now cancels the comment instead of getting stuck.
grok --minimal now reprints history correctly after terminal resize and keeps the prompt/status rows anchored.
Restoring a stashed draft now places the cursor at the end so continued typing appends correctly.
grok update on WinGet installs now prints the winget upgrade command and exits without writing files.
Inline images are re-uploaded after a full repaint so they no longer appear blank in terminals that drop image data on ESC[2J.
Plugin hooks now run on a fresh session without requiring an explicit reload.
Fixed plugin auto-update detection when marketplace URLs differ only by .git suffix or host case.

source

Alpha channel 1.0.42

2026-09-26 · alpha · alpha pointer is 1.0.42

1.0.42 — 2026-09-26

Features

Footer and /usage now display the model actually served on smart-auto turns.
New grok worktree create command creates a managed worktree without launching an interactive session.
Auto permission mode now appears as Auto-review everywhere in the UI.

Bug Fixes

Bracketed pastes that did not come from the clipboard no longer attach leftover images (e.g. from IME commits).
grok -p now exits cleanly with the documented code when interrupted and performs session cleanup instead of being killed.
Pressing Ctrl+C twice while writing a plan comment now cancels the comment instead of getting stuck.
grok --minimal now reprints history correctly after terminal resize and keeps the prompt/status rows anchored.
Restoring a stashed draft now places the cursor at the end so continued typing appends correctly.
grok update on WinGet installs now prints the winget upgrade command and exits without writing files.
Inline images are re-uploaded after a full repaint so they no longer appear blank in terminals that drop image data on ESC[2J.
Plugin hooks now run on a fresh session without requiring an explicit reload.
Fixed plugin auto-update detection when marketplace URLs differ only by .git suffix or host case.

source

Stable channel 1.0.41

2026-09-22 · stable · stable pointer is 1.0.41

1.0.41 — 2026-09-22

Features

Added a new sports_search tool that can look up live NFL scores, standings, schedules, player stats, and team records directly from X data.
Subagent model inheritance setting added to /settings; persists in config.toml and respects managed/overlay layers.
Per-model request size limits can now be configured to match provider HTTP body caps and control inline image eviction.
Long reasoning reminder can now be enabled via config to nudge the model after long hidden-reasoning steps.

Bug Fixes

Fixed the subagent fullscreen view so it no longer shows a stray [Dashboard] button in the header.
Fixed agent frontmatter mcpServers so that headers and URLs from the active agent's agent.md now correctly override config.toml and survive config reloads or agent switches.
Ctrl+P now opens the command palette immediately from the welcome screen.
Dashboard now focuses the "+ New Agent" row instead of leaving the previous session selected.
Sessions that were interrupted by a crash now show a clear marker instead of silently dropping the turn.
Ctrl+Z right after stashing a prompt now restores it.
Collapsed edit blocks setting now correctly collapses edits even if expanded_by_default was pinned true.
Interjections now receive a visible reply before the agent resumes prior tasks.
Canceling a turn that blocked on spawn_subagent now tells the model the child moved to the background instead of claiming it was never executed.
Saving a queued-prompt edit now returns focus to the composer so the next keys type the next message.
Subagent activation is now consistent between the TUI and grok agent stdio; tables that only set limits or models no longer disable subagents.
Effort level selection now accepts menu labels in addition to IDs.

source

Alpha channel 1.0.41

2026-09-22 · alpha · alpha pointer is 1.0.41

1.0.41 — 2026-09-22

Features

Added a new sports_search tool that can look up live NFL scores, standings, schedules, player stats, and team records directly from X data.
Subagent model inheritance setting added to /settings; persists in config.toml and respects managed/overlay layers.
Per-model request size limits can now be configured to match provider HTTP body caps and control inline image eviction.
Long reasoning reminder can now be enabled via config to nudge the model after long hidden-reasoning steps.

Bug Fixes

Fixed the subagent fullscreen view so it no longer shows a stray [Dashboard] button in the header.
Fixed agent frontmatter mcpServers so that headers and URLs from the active agent's agent.md now correctly override config.toml and survive config reloads or agent switches.
Ctrl+P now opens the command palette immediately from the welcome screen.
Dashboard now focuses the "+ New Agent" row instead of leaving the previous session selected.
Sessions that were interrupted by a crash now show a clear marker instead of silently dropping the turn.
Ctrl+Z right after stashing a prompt now restores it.
Collapsed edit blocks setting now correctly collapses edits even if expanded_by_default was pinned true.
Interjections now receive a visible reply before the agent resumes prior tasks.
Canceling a turn that blocked on spawn_subagent now tells the model the child moved to the background instead of claiming it was never executed.
Saving a queued-prompt edit now returns focus to the composer so the next keys type the next message.
Subagent activation is now consistent between the TUI and grok agent stdio; tables that only set limits or models no longer disable subagents.
Effort level selection now accepts menu labels in addition to IDs.

source

Alpha channel 1.0.39

2026-09-20 · alpha · alpha pointer is 1.0.39

1.0.39 — 2026-09-20

Features

Subagents can always use the parent model, with that choice locked in when the session starts.
Earlier image attachments now survive multiple compactions via a persisted path list.
Subagent spawning no longer requires choosing a type; omitted calls default to general-purpose.
The agent now keeps helper scripts, logs, and PR drafts in the system temp directory instead of the repository.
read_file descriptions now tell the model when offset/limit are ignored for SKILL.md and instruction files.
Effort levels for models now come from the API instead of hard-coded lists, and config aliases inherit the menu.
MCP tools that receive the wrong JSON shape are now automatically fixed when the mismatch is unambiguous.

Bug Fixes

Local worktree capture no longer refuses repositories that contain uninitialized submodules.
Memory captures no longer include opaque model reasoning blobs.
Idle timeouts now show a clear message instead of raw internal text.
Images attached to the last user prompt now survive compaction.
Custom agent profiles now persist correctly across session resume and reload.
Subagent labels now appear as "Subagent" instead of "General" in transcripts and the tasks pane.

source

Alpha channel 1.0.38

2026-09-19 · alpha · alpha pointer is 1.0.38

1.0.38 — 2026-09-19

Features

Long agent replies in prompt-suggestion transcripts are no longer cut off mid-sentence.
read_file on skill and instruction files can now be configured per deployment.
Pasted images survive yank/undo/history recall and the user is notified when any cannot be sent.

Bug Fixes

Subagent overlays no longer stay stuck showing "Cancelling" after the child turn ends.
Long quoted arguments in permission prompts now wrap instead of clipping.
Fixed tool-calling errors by sending inline-only use_tool schema.

source

Alpha channel 1.0.37

2026-09-18 · alpha · alpha pointer is 1.0.37

1.0.37 — 2026-09-18

Bug Fixes

Mermaid flowcharts with complex labels or large sizes now render and open correctly.
/context legend now shows accurate percentages that always add up to 100%.
Pressing y on a queue row now copies the prompt text to the clipboard.
/model effort picker now defaults to the model's recommended effort level.
Bash commands using quoted filename variables now respect configured permission rules instead of always prompting.

source

Alpha channel 1.0.36

2026-09-17 · alpha · alpha pointer is 1.0.36

1.0.36 — 2026-09-17

Features

New Dashboard preview setting lets you hide the selected-session preview panel on the dashboard.
New policy setting allows organizations to disable hooks that are not from managed policy.
Background shell commands now appear as live task rows with streaming output.

Bug Fixes

grok --sandbox no longer exits when accepting folder trust.
Pinned agents in the dashboard no longer jump when their activity changes.
Swift code with triple-quoted strings now highlights correctly in the pager.
Mixed-case reasoning_effort values (e.g. "xHigh") are now accepted in config.toml.
Fixed accidental empty agent sessions when pressing Enter right after sending from the dashboard.
Fixed default TUI sessions so an [agent] set in config.toml is respected instead of always using a builtin plan.
MCP plugin authentication now succeeds for servers that declare a client ID in their manifest.
Shell command approvals from background subagents are now shown instead of being silently rejected.

source

Enterprise channel 1.0.35

2026-09-16 · enterprise · enterprise pointer is 1.0.35

1.0.35 — 2026-09-16

Bug Fixes

Headless MCP status reporting and connecting reminders now match actual server handshake state.
MCP tool searches and calls now render with query, results, arguments and output on the daemon path.
Swift string interpolation with nested parentheses now highlights correctly in the TUI.
Copy-paste in --minimal mode no longer inserts extra blank lines or breaks long paths at wrap points.
/memory modal now shows copy confirmations, searches note contents, and works on narrow terminals.
Memory delete is now much faster and toggling memory on/off from a config-disabled session now correctly wires the agent.

Performance

Syntax highlighting now uses far less memory and runs faster on large TypeScript and other files.

source

Alpha channel 1.0.35

2026-09-16 · alpha · alpha pointer is 1.0.35

1.0.35 — 2026-09-16

Bug Fixes

Headless MCP status reporting and connecting reminders now match actual server handshake state.
MCP tool searches and calls now render with query, results, arguments and output on the daemon path.
Swift string interpolation with nested parentheses now highlights correctly in the TUI.
Copy-paste in --minimal mode no longer inserts extra blank lines or breaks long paths at wrap points.
/memory modal now shows copy confirmations, searches note contents, and works on narrow terminals.
Memory delete is now much faster and toggling memory on/off from a config-disabled session now correctly wires the agent.

Performance

Syntax highlighting now uses far less memory and runs faster on large TypeScript and other files.

source

Alpha channel 1.0.33

2026-09-15 · alpha · alpha pointer is 1.0.33

1.0.33 — 2026-09-15

Features

MCP tool results now include structured JSON data when the server provides it.
GROK_GROVE=1 now enables both clone and worktree features when specific knobs are unset; missing remote settings no longer force copy.
Delete memories from the /memory modal using two-press x; only actual notes (not generated indexes) can be removed.
Composer footer now shows the Alt+Enter chord for inserting a newline when Cmd+Enter is unavailable over SSH.
/memory now always opens the memory browser modal; status and toggle are available via keys inside the modal instead of command arguments.

Bug Fixes

Fixed a bug where pasting an image on macOS could attach the wrong image.
MCP tool calls listed directly now show their name, arguments, and any error message.
Background subagent tasks now correctly show cancelled status when the parent session closes.
The Subagents dock no longer shows an unclickable "Done N completed" row after all subagents finish.
Cancelled or timed-out MCP tool calls now notify the server so it can stop work instead of continuing in the background.
/rewind no longer fails when older compaction checkpoints are missing after a session sweep; only the base checkpoint for the target matters.
Large skill files are now capped at the same 25k-token limit as other reads; oversized skill bodies show a truncation note instead of flooding context.
Long-running sessions no longer lose old compaction checkpoints or prompt offloads during the 30-day cleanup sweep.
Auto "where was I" recaps no longer appear while a scheduled task, monitor, or workflow is still running.
In minimal mode, confirming /delete or /exit now immediately opens a new empty session instead of appearing stuck.
In minimal mode, resuming a session after /new now correctly reloads its history instead of showing the wrong session.
MCP tools that take no arguments no longer show a stray "{}" line in the permission sheet.
Fixed Windows clone and attach failures when .git/config values contain backslashes, quotes, or other characters that must be escaped.
/memory modal now shows clear explanations when memory is disabled and helpful empty states instead of blank panes.
/memory (and on/off toggles) no longer leaves user bubbles or host turns in scrollback when opening or changing the modal.
Fixed Windows clone staging directories that could not be cleaned up after deadline or cancel during shallow bootstrap.

source

Alpha channel 1.0.32

2026-09-14 · alpha · alpha pointer is 1.0.32

1.0.32 — 2026-09-14

Bug Fixes

Plugin and skill listings from x.ai/plugins/list and x.ai/skills/* now reflect the latest config.toml even before the first session.
Fixed crashes on the first TLS handshake for Windows on ARM64 builds.
Slash commands like /feedback typed during plan approval now execute correctly.

source

Enterprise channel 1.0.31

2026-09-13 · enterprise · enterprise pointer is 1.0.31

1.0.31 — 2026-09-13

Bug Fixes

Folded subagent groups in scrollback now correctly label how many are still running versus completed.
Dashboard search and the resume picker now clearly indicate the active text field and restore list selection after search.
Dock sections no longer keep a selection highlight after you collapse them with a click.
Worktree paths no longer show an extra suffix in the header or dashboard.
MCP tools with long server prefixes are now kept instead of being silently dropped.
Turn cancelled banners now correctly name the cause instead of always saying "by user".
Long paths in headers are now always shortened to the last two components.

source

Alpha channel 1.0.31

2026-09-13 · alpha · alpha pointer is 1.0.31

1.0.31 — 2026-09-13

Bug Fixes

Folded subagent groups in scrollback now correctly label how many are still running versus completed.
Dashboard search and the resume picker now clearly indicate the active text field and restore list selection after search.
Dock sections no longer keep a selection highlight after you collapse them with a click.
Worktree paths no longer show an extra suffix in the header or dashboard.
MCP tools with long server prefixes are now kept instead of being silently dropped.
Turn cancelled banners now correctly name the cause instead of always saying "by user".
Long paths in headers are now always shortened to the last two components.

source

Stable channel 1.0.30

2026-09-12 · stable · stable pointer is 1.0.30

1.0.30 — 2026-09-11

Features

Session header redesigned as a single row that carries the overlay context and a Dashboard button.

Bug Fixes

Elapsed-time display now rolls over into hours (for example 3h14m) instead of stopping at 60 minutes.
Dock Watchers and Loops rows now show when the next run is due.
Workflow status now shows a clickable row in the dock.

Performance

Fixed lag in tmux panes by omitting per-frame synchronized-update wrappers.

source

Alpha channel 1.0.30

2026-09-11 · alpha · alpha pointer is 1.0.30

1.0.30 — 2026-09-11

Features

Session header redesigned as a single row that carries the overlay context and a Dashboard button.

Bug Fixes

Elapsed-time display now rolls over into hours (for example 3h14m) instead of stopping at 60 minutes.
Dock Watchers and Loops rows now show when the next run is due.
Workflow status now shows a clickable row in the dock.

Performance

Fixed lag in tmux panes by omitting per-frame synchronized-update wrappers.

source

Alpha channel 1.0.29

2026-09-11 · alpha · alpha pointer is 1.0.29

1.0.29 — 2026-09-11

Bug Fixes

Tab navigation in the pager now cycles cleanly among prompt, scrollback, and the entire dock instead of stopping at every dock section.
Ctrl+G now hides or shows the dock in the pager; focus returns to scrollback when hiding.
Multiple concurrent grok agent stdio processes against the same GROK_HOME no longer crash at startup.
MCP server connections no longer fail against modern 2026-07-28 servers.
Fixed startup crash on Windows when the Client-ProjFS optional feature is disabled.

Performance

Agent dashboard no longer wastes CPU repainting identical frames when only idle spinners or hidden rows are present.

source

Alpha channel 1.0.28

2026-09-10 · alpha · alpha pointer is 1.0.28

1.0.28 — 2026-09-10

Features

/btw now works mid-message: the rest of the line becomes the side question and does not appear in the main transcript.
Per-model reasoning_summary setting is now available in config.toml for Responses API endpoints.

Bug Fixes

Auto-update no longer removes a grok binary that is still in use by another process.
Prompt editor no longer gets the cursor stuck or skips a row when pressing Up/Down across soft-wrapped lines.

source

Enterprise channel 1.0.26

2026-09-09 · enterprise · enterprise pointer is 1.0.26

1.0.26 — 2026-09-09

Features

Interject messages to subagents now interrupt long-running background waits so the message arrives immediately.
Subagent message rows in scrollback now indicate whether a message was sent as interject, queued, or steer.
Feedback form is now available directly in minimal mode without switching to fullscreen.

Bug Fixes

/btw side questions now include attached images from the composer.
Background task reminders no longer echo the original shell command.
Failed background tasks no longer produce duplicate reminders after the model reads their output.
TPM rate-limit errors during compaction are now retried instead of treated as context overflow.
Ctrl+C early in a turn now correctly rewinds the prompt without races.
Dashboard arrow keys now navigate the actions row from an empty input box.
MCP tool calls on legacy servers now correctly advertise elicitation support.
Memory v2 now recovers from init failures and avoids stalled capture after repeated errors.

source

Alpha channel 1.0.26

2026-09-09 · alpha · alpha pointer is 1.0.26

1.0.26 — 2026-09-09

Features

Interject messages to subagents now interrupt long-running background waits so the message arrives immediately.
Subagent message rows in scrollback now indicate whether a message was sent as interject, queued, or steer.
Feedback form is now available directly in minimal mode without switching to fullscreen.

Bug Fixes

/btw side questions now include attached images from the composer.
Background task reminders no longer echo the original shell command.
Failed background tasks no longer produce duplicate reminders after the model reads their output.
TPM rate-limit errors during compaction are now retried instead of treated as context overflow.
Ctrl+C early in a turn now correctly rewinds the prompt without races.
Dashboard arrow keys now navigate the actions row from an empty input box.
MCP tool calls on legacy servers now correctly advertise elicitation support.
Memory v2 now recovers from init failures and avoids stalled capture after repeated errors.

source

Stable channel 1.0.25

2026-09-09 · stable · stable pointer is 1.0.25

1.0.25 — 2026-09-09

Features

Background task list is now sent as a durable snapshot so clients can restore state after reconnect.
Hook runs that succeed are now silent; only blocking or failing hooks show status or a single line.
Vim-mode Shift+J/K now jump to the next or previous turn at the top of the viewport, matching the timeline arrows.
Double-click now selects a radio button in settings choosers.
Dashboard actions row (+ New Agent, Open Previous, Worktree) can now be navigated with arrow keys and activated with Enter.
Workflow tool now supports pause and stop sources so agents can control background runs they launched.
/theme picker now matches aliases such as transparent, dark, or system and still inserts the canonical name.

Bug Fixes

Bash command output shown in the pager is now the complete result instead of a truncated tail.
Voice dictation now inserts text at the cursor instead of always appending to the end.
Headless prompts now time out cleanly instead of hanging forever when the agent does not respond.
Scheduled tasks now guide the agent to check results and clean up or update recurring schedules.
Fixed duplicate user prompt rendering in some ACP clients by making live echo opt-in.
Fixed grok -c attaching the wrong empty session after starting without --continue then resuming.
Dock and queue hover colors now match the rest of the terminal instead of using the stronger dropdown highlight.
Fixed image upload errors when reading Adobe SVGs that contain a PNG thumbnail prefix.
Dock subagent rows now show the same [stop] label as other killable rows.
Fixed startup settings and model catalog being shared or clobbered across concurrent boots or multiple accounts on the same machine.
Dock Tasks panel now shows the spinner aligned with the header instead of nested under a rule.
/feedback <text> now sends immediately again instead of starting a model turn; failures save to Drafts with a notice.

Performance

Embedded launches no longer time out while scanning the skills directory on slow filesystems.

source

Alpha channel 1.0.25

2026-09-09 · alpha · alpha pointer is 1.0.25

1.0.25 — 2026-09-09

Features

Background task list is now sent as a durable snapshot so clients can restore state after reconnect.
Hook runs that succeed are now silent; only blocking or failing hooks show status or a single line.
Vim-mode Shift+J/K now jump to the next or previous turn at the top of the viewport, matching the timeline arrows.
Double-click now selects a radio button in settings choosers.
Dashboard actions row (+ New Agent, Open Previous, Worktree) can now be navigated with arrow keys and activated with Enter.
Workflow tool now supports pause and stop sources so agents can control background runs they launched.
/theme picker now matches aliases such as transparent, dark, or system and still inserts the canonical name.

Bug Fixes

Bash command output shown in the pager is now the complete result instead of a truncated tail.
Voice dictation now inserts text at the cursor instead of always appending to the end.
Headless prompts now time out cleanly instead of hanging forever when the agent does not respond.
Scheduled tasks now guide the agent to check results and clean up or update recurring schedules.
Fixed duplicate user prompt rendering in some ACP clients by making live echo opt-in.
Fixed grok -c attaching the wrong empty session after starting without --continue then resuming.
Dock and queue hover colors now match the rest of the terminal instead of using the stronger dropdown highlight.
Fixed image upload errors when reading Adobe SVGs that contain a PNG thumbnail prefix.
Dock subagent rows now show the same [stop] label as other killable rows.
Fixed startup settings and model catalog being shared or clobbered across concurrent boots or multiple accounts on the same machine.
Dock Tasks panel now shows the spinner aligned with the header instead of nested under a rule.
/feedback <text> now sends immediately again instead of starting a model turn; failures save to Drafts with a notice.

Performance

Embedded launches no longer time out while scanning the skills directory on slow filesystems.

source

Enterprise channel 1.0.22

2026-09-07 · enterprise · enterprise 1.0.22 notes from CLI /release-notes

1.0.22 — 2026-09-07

Features

Built-in agent tools now take precedence over user MCP servers when tool names collide.
Desktop app tools are now available through a dedicated first-party MCP server.
Remote control pickers can now group servers by device using announced device identity.
edit_file diffs now show real file line numbers instead of starting at line 1.
Permission prompts for file edits now auto-expand the diff row so you can see the change before approving.
Sending a message to a finished subagent now continues that same subagent instead of failing.
Background subagent completion messages now show the actual output instead of just a pointer to run the tool.
Dashboard now shows a unified header with location picker and an actions row for creating agents or resuming previous sessions.
New grok-workspaced daemon mode lets a long-lived process expose folders to the Computer Hub (separate from the Desktop-supervised sidecar).

Bug Fixes

MCP connectors now correctly show when they need re-authentication even if no tools are listed.
Extensions modal shortcuts now work when the tab bar is focused, and invalid actions on headers show helpful messages.
Send now button and shortcuts now work correctly when an automatic background turn is running.
Resumed sessions now show the exact text you typed for mid-turn follow-ups instead of the wrapped system message.
/skills command now updates the model with newly installed skills even when written from another NFS client.
Auto mode no longer instantly runs destructive git checkout -- commands; they now go through the model for safety.
Fixed a crash that could occur when restoring pasted content after certain skill injections or rewinds.
Subagents and workflows now respect the same bash timeout and auto-background settings configured for the parent session.
Fixed model backend selection so an explicit chat_completions setting in config.toml is no longer overwritten by same-slug siblings.
Slash command advertisements no longer spam repeatedly when your current directory is your home folder.
Foreground shell commands no longer produce spurious background-task reminders or incorrect kill results.
Queued follow-up buttons now appear in the order [Send now][edit][cancel].
/compact instructions in the pager are now passed to the compaction backend (bare /compact is unchanged).
/goal (and resume) after pressing Esc now correctly starts the goal planner instead of failing with "Planning failed".
run_terminal_cmd now correctly tells the model that foreground commands are backgrounded after ~15s instead of the 120s timeout value.
Skill announcements no longer repeat the same list of available skills multiple times in a session.
Monitor events in multi-session processes now correctly wake idle sessions instead of being buffered until the next prompt.

Performance

Opening or resuming sessions no longer pauses the interface while reading MCP config files.

source

Alpha channel 1.0.22

2026-09-07 · alpha · alpha 1.0.22 notes from CLI /release-notes

1.0.22 — 2026-09-07

Features

Built-in agent tools now take precedence over user MCP servers when tool names collide.
Desktop app tools are now available through a dedicated first-party MCP server.
Remote control pickers can now group servers by device using announced device identity.
edit_file diffs now show real file line numbers instead of starting at line 1.
Permission prompts for file edits now auto-expand the diff row so you can see the change before approving.
Sending a message to a finished subagent now continues that same subagent instead of failing.
Background subagent completion messages now show the actual output instead of just a pointer to run the tool.
Dashboard now shows a unified header with location picker and an actions row for creating agents or resuming previous sessions.
New grok-workspaced daemon mode lets a long-lived process expose folders to the Computer Hub (separate from the Desktop-supervised sidecar).

Bug Fixes

MCP connectors now correctly show when they need re-authentication even if no tools are listed.
Extensions modal shortcuts now work when the tab bar is focused, and invalid actions on headers show helpful messages.
Send now button and shortcuts now work correctly when an automatic background turn is running.
Resumed sessions now show the exact text you typed for mid-turn follow-ups instead of the wrapped system message.
/skills command now updates the model with newly installed skills even when written from another NFS client.
Auto mode no longer instantly runs destructive git checkout -- commands; they now go through the model for safety.
Fixed a crash that could occur when restoring pasted content after certain skill injections or rewinds.
Subagents and workflows now respect the same bash timeout and auto-background settings configured for the parent session.
Fixed model backend selection so an explicit chat_completions setting in config.toml is no longer overwritten by same-slug siblings.
Slash command advertisements no longer spam repeatedly when your current directory is your home folder.
Foreground shell commands no longer produce spurious background-task reminders or incorrect kill results.
Queued follow-up buttons now appear in the order [Send now][edit][cancel].
/compact instructions in the pager are now passed to the compaction backend (bare /compact is unchanged).
/goal (and resume) after pressing Esc now correctly starts the goal planner instead of failing with "Planning failed".
run_terminal_cmd now correctly tells the model that foreground commands are backgrounded after ~15s instead of the 120s timeout value.
Skill announcements no longer repeat the same list of available skills multiple times in a session.
Monitor events in multi-session processes now correctly wake idle sessions instead of being buffered until the next prompt.

Performance

Opening or resuming sessions no longer pauses the interface while reading MCP config files.

source

Alpha channel 1.0.21

2026-09-04 · alpha · alpha 1.0.21 notes from CLI /release-notes

1.0.21 — 2026-09-04

Bug Fixes

Permission mode (auto / always-approve) now stays visible when the agent enters plan mode and is restored on exit.
Typing, pasting, or pressing Shift+Tab on the welcome screen now immediately opens the session instead of creating a duplicate one.
Dock sections, queue body, and reveal rows now stay within their documented height rules even on crowded terminals.

source

Alpha channel 1.0.19

2026-09-04 · alpha · alpha 1.0.19 notes from CLI /release-notes

1.0.19 — 2026-09-04

Breaking Changes

Scheduled /loop tasks always run in the background; they no longer inject turns into your conversation.

Features

New terminal theme option makes backgrounds transparent so the terminal's own colors show through.
MCP servers blocked by organization policy now show clear messages and are refused before any config change.
Session resume now tells the model what loops, subagents, and workflows were still running.
Dashboard now shows newly dispatched sessions immediately instead of waiting for the store.
Headless sessions (grok -p) now support the --worktree flag to run in a separate git worktree.
Dashboard now says 'Open session' instead of 'Add session' for the session picker button.
Multi-line bash commands now render with proper wrapping and highlighting when opening the block viewer.
/usage now works from the dashboard and shows account allowance when no session is active.

Bug Fixes

Fixed mid-turn UI freezes when the terminal stops reading output.
Fixed crashes that occurred when the terminal pane was closed while grok was exiting.
MCP server list in minimal mode now correctly shows policy-blocked servers.
URLs that wrap across multiple lines inside quotes or lists are now fully clickable.
The welcome screen composer now grows taller when you paste multi-line text.
Enterprise policy files are no longer deleted on startup when your team login is stored at a custom GROK_AUTH_PATH.
/feedback now drops unsupported images with a notice (matching the modal) and never loses your report text on save errors.
Feedback drafts keep their paragraph breaks when updated, and the modal no longer switches tabs unexpectedly.
Turn summary lines no longer lose their spacing after background tasks finish.
Thinking blocks no longer stay expanded after switching between fullscreen and minimal modes.

Performance

First prompt after login is faster when MCP servers are configured; they connect in the background.

source

Alpha channel 1.0.18

2026-09-02 · alpha · alpha 1.0.18 notes from CLI /release-notes

1.0.18 — 2026-09-02

Features

Managed policy now blocks disallowed MCP servers and marketplace installs before any change is written.
Quoting a previous message from the block viewer now works directly from the viewer.
Welcome screen now stays until you actually send a prompt; a session is prepared in the background.
Feedback is now collected in a centered modal instead of a bottom card.
Feedback submissions now include structured type, task, and failure information.
Feedback modal now has a Drafts tab to manage and send saved drafts.
Models can now render interactive charts using Chart.js.
grok inspect and grok mcp doctor now show managed policy details.
Image generation now supports richer parameters via composer state.
Retry behavior for rate-limited requests is now configurable per model.
Models can now be configured with per-model mTLS client certificates for secure upstream connections.
/feedback <text> now saves a local draft before invoking the feedback skill.

Bug Fixes

Authentication recovery during turns no longer fails immediately on transient errors.
The session dock now aligns consistently, shows more rows inline when expanded, and reveals stop controls on hover.
Pastes from rich-text sources no longer stay inline or corrupt the composer when they contain special Unicode line separators.
Mouse wheel over a one-line prompt now scrolls the conversation history instead of being ignored.
Slash dropdowns no longer truncate long command names when a short sibling is also visible.
Subagent sessions now respect the retry and rate-limit settings of their selected model.
Session picker now opens with Ctrl+R everywhere; redo uses Ctrl+Shift+Z or Alt+Z.
Subagent views no longer duplicate the task prompt on first open after a live echo.
Strikethrough text now renders correctly inside markdown tables in the pager.
Resume picker now labels local sessions with a single disk walk instead of per-row lookups.

Performance

Startup no longer hangs on slow networks while fetching remote settings.
Responses now complete faster; uploads and session bookkeeping happen in the background.
Startup now makes fewer network requests by sharing settings fetches.
New sessions start responding faster by running start hooks in the background.

source

Enterprise channel 1.0.17

2026-09-01 · enterprise · enterprise 1.0.17 notes from CLI /release-notes

1.0.17 — 2026-09-01

Features

MCP tools now support the 2026-07-28 multi-round-trip elicitation flow where a tool call can return an input_required state that is resumed on retry.
Next-prompt ghost suggestions now appear less often and are more accurate by preferring silence when the next line is not obvious.

Bug Fixes

Table cell selection inside /btw panels now correctly highlights cells and copies as TSV instead of drawing a linear rectangle.

source

Alpha channel 1.0.17

2026-09-01 · alpha · alpha 1.0.17 notes from CLI /release-notes

1.0.17 — 2026-09-01

Features

MCP tools now support the 2026-07-28 multi-round-trip elicitation flow where a tool call can return an input_required state that is resumed on retry.
Next-prompt ghost suggestions now appear less often and are more accurate by preferring silence when the next line is not obvious.

Bug Fixes

Table cell selection inside /btw panels now correctly highlights cells and copies as TSV instead of drawing a linear rectangle.

source

Alpha channel 1.0.16

2026-09-01 · alpha · alpha 1.0.16 notes from GitHub changelog

1.0.16 — 2026-09-01

Breaking Changes

Enterprise policies can now restrict which models users may select via signed requirements.toml.

Features

MCP servers can now be supplied at session bind time for workspace integrations.

Bug Fixes

Long-running sessions interrupted by expired tokens during network issues no longer lose work.
Slash command suggestions with very long names no longer show blank labels in the dropdown.
Sending a message immediately after spawning a subagent no longer fails while the child is still starting.
Loading spinners inside the extensions modal (/mcps and other tabs) now animate correctly.
MCP server OAuth authentication triggered from /mcps no longer deadlocks the session.

Performance

Long-running subagent and task output waits now default to a one-hour ceiling instead of ten minutes.

source

Alpha channel 1.0.15

2026-08-31 · alpha · alpha 1.0.15 notes from GitHub changelog

1.0.15 — 2026-08-31

Features

Tip appears after repeated scrollback drag-copies suggesting /copy and /export commands.

Bug Fixes

Session close is now faster because memory consolidation runs at the next launch instead of blocking exit.
Typed input including Enter during pager startup is now preserved and correctly interpreted as submit or newline.
Dock panel input and rollout now respect remote settings and correctly handle keyboard focus when hidden or empty.

Performance

First reply latency is reduced by opening the model connection in the background when a session starts.
Signed-in startup is faster because an expired token refresh now begins in the background at agent spawn.
Creating a new session returns faster; MCP tools and other startup work happen in the background.

source

Alpha channel 1.0.14

2026-08-31 · alpha · alpha 1.0.14 notes from GitHub changelog

1.0.14 — 2026-08-31

Features

OIDC token refresh is now proactive by default for better reliability.
PostToolUse hooks can now provide feedback and context to the model after tool execution.
SDK-registered PostToolUse hooks now provide model-facing feedback.
grok usage <session-id> now shows persisted per-turn token and cost data.
Retry status in composer and title now shows a short reason for the retry.
Models can now declare a different identifier for each reasoning-effort level instead of always sending the same id.
Prompt suggestions now respect remote configuration and default to the current session model.
Windows CLI downloads are now ~70% smaller using the same compressed sidecars as macOS and Linux.

Bug Fixes

grok inspect now correctly shows Claude bypass locks as advisory rather than enforced.
Subagent sessions no longer leak threads or file descriptors when the parent is busy.
Cold startup no longer performs duplicate remote settings fetches.
Compaction failures due to context size now degrade input instead of retrying identically.
--sandbox strict now restricts writes to ~/.grok/sessions only.
Subagent spawning now waits longer on a busy coordinator and shows clearer retry guidance instead of "unreachable".
Failed task and todo tool calls now appear in the transcript instead of disappearing without a trace.
Composer status row no longer collapses or flashes when using double-Enter to send now.
Session close is no longer delayed by a single slow hook; each SessionEnd hook now has its own timeout.
Hook removal in the extensions modal no longer offers actions that the handler will refuse.
Interjections during a turn are now delivered atomically or not at all.
Subagent tasks no longer get incorrectly cancelled when the parent session is waiting for completion.
Workflow detail view now closes the overlay on X or outside click instead of returning to the run list.
Resuming subagents now succeeds for larger transcripts that still fit the model context with headroom.

Performance

Startup now fetches remote settings only once per boot instead of potentially twice.
First message on large repositories no longer waits on repository status scan.
Large session memory no longer blocks the agent during turn completion or subagent spawning.
Signed-in CLI starts faster by serving remote settings from a local cache on warm boots.

source

Stable channel 1.0.13

2026-08-29 · stable · stable 1.0.13 notes from GitHub changelog

1.0.13 — 2026-08-28

Features

Length-truncated responses now continue automatically instead of failing the turn.
Hooks can now ask the user to confirm a tool call instead of always allowing or denying.
Hooks can now request deferral or add context shown to the model after a tool runs.
Session close now records detailed timing data for performance analysis.
Credit limit upsell now offers a Try Again button to retry the last prompt.
Pasted images now show a live pixel preview in the prompt box on iTerm2.

Bug Fixes

Transient inference failures (stalls, drops, 5xx) now retry automatically instead of ending the turn.
Windows users can now correctly open ~/.grok and worktree sessions.
Session data is now more reliably saved after prompts and on power loss.
Compaction failures now show the actual error instead of a generic message.
Truncation error messages now show the right guidance instead of suggesting an unhelpful retry.
Truncated tool calls are now executed instead of failing the turn when arguments are complete.
Images larger than 2000px no longer brick sessions on many-image requests.
Wrapped hyperlinks in the pager now remain fully clickable on Windows Terminal instead of only the first line.
Recurring scheduled tasks now include a reminder to stop the monitor when work finishes.
Scheduled task IDs are now full UUID strings, preventing collisions when tasks are created in the same millisecond.

Performance

Subagent spawning is faster when connections drop during bursts.
Session startup with MCP servers is now much faster when auth is already configured.
MCP server startup no longer stalls behind a fixed batch size.
CLI downloads are now compressed, making fresh installs and updates substantially faster.

1.0.12 — 2026-08-27

Features

Credit-limit upsell now includes a Try Again option. Max-tier users see the same question modal without Upgrade tier.

Bug Fixes

Copying wrapped table cells no longer inserts unwanted spaces at line breaks.
MCP server connections that fail transiently now retry instead of staying unavailable.
Waiting on subagents after an interjection no longer blocks on unrelated background work.
Prompt blocks now show friendly hook descriptions instead of internal IDs.
Truncation error message no longer suggests asking for a shorter answer.
Context estimates for conversations with reasoning are now more accurate.
Token usage after rewinding or switching modes now matches the actual server-reported count instead of overestimating.
Context bar now updates immediately when auto-compaction begins instead of waiting until it finishes.
File system monitoring for .NET projects no longer creates excessive watchers on Linux.
Auto recap no longer appears while a background task or monitor wake turn is streaming.

Performance

Worktree creation is faster by skipping stale reflog copies.

1.0.11 — 2026-08-26

Features

Headless sessions are now browsable in the resume picker without mixing into default history.
Default permission mode for new interactive sessions is now configurable.
Turn duration now appears in session history after resume.
Turn footers (Worked for, cancelled, failed) now appear after /resume.
mkdir and touch no longer prompt in auto mode or safe-command lists.
Subagent messages are now allowed automatically in permission Auto mode.
Headless sessions can now auto-allow permission prompts via a startup hint.

Bug Fixes

Permission prompts for common command chains and subcommands are now more reliable.
Blocked prompts no longer appear in conversation history or scrollback after restart.
Background monitors no longer have a 10-hour default timeout.
Mouse input at the right margin no longer types characters into the prompt on certain terminals.
Pasting text ending in a newline no longer accidentally submits the prompt on some terminals.
Auto mode now shows a permission card when the classifier blocks an action on interactive sessions.
Image previews no longer leave ghost artifacts when using the Kitty protocol on Warp.
Expanded Execute tool output no longer snaps closed during live progress.
/voice now falls back to parec/arecord on older PipeWire installs.

Performance

Background command waits now finish as soon as the process exits.

1.0.10 — 2026-08-24

Performance

grok clone now reuses matching local checkouts as linked worktrees for faster session creation.

1.0.9 — 2026-08-24

Features

New sessions in the interactive TUI now start in auto mode by default for fewer approval prompts.
/feedback now supports attaching images (screenshots etc.).
The / slash menu is now ordered by recency and groups skills below commands for easier navigation.
/workflow now accepts --agent-budget N or an agent_budget JSON field to set child-agent limits.
/workflow now accepts --effort LEVEL or an effort JSON field to set child reasoning effort.
MCP servers now receive a grok-cli User-Agent header so providers can attribute traffic.
Dashboard now lets you start new agents in Auto mode via Shift+Tab.
Workflows now appear alongside skills in the model prompt so the agent can launch them by name.
/edit-prompt now opens an external editor from the full TUI, not only minimal mode.
Up arrow on an empty prompt now selects queued follow-ups first instead of opening history.
/plugin now works as an alias for /plugins in the TUI.
Plugin-provided agents now appear in the /agents modal and can be toggled.
/minimal and /fullscreen now switch instantly inside the running session without restarting or losing the current turn.
The configurable status line row now appears at the bottom of minimal mode as well as fullscreen.
grok clone now fetches only the selected branch tip by default, with --full-history for legacy full clones.
grok clone can now reuse a matching local checkout as a linked worktree instead of always fetching from the network.

Bug Fixes

Subagent tasks are more resilient to temporary rate limits and will retry instead of failing immediately.
Narrow allow rules for Bash commands now correctly permit file writes without extra prompts.
Rules containing markdown headings now render correctly in the prompt.
/new and /clear now preserve the last-chosen reasoning effort instead of resetting to the model default.
Prompt dropdowns (slash, history, file search) now have consistent left alignment with the composer.
History now shows every command, slash command, and memory note in the order you typed them.
Collapsed tool-call rows no longer show a vertical accent line beside the status diamond.
Hint text above the prompt now lines up with the text you type.
Pressing Esc while peeking at a conversation from the dashboard now closes open modals instead of leaving the view.
/copy now preserves markdown formatting such as headings and code fences.
Sending a new message while a shell command is running now moves the command to the background instead of cancelling it.
grok mcp add now correctly treats bare http(s):// URLs as HTTP servers instead of stdio commands.
Plan mode no longer incorrectly blocks subagent spawns as file edits outside plan.md.
MCP servers that share a URL but have different names are now kept as separate entries instead of being collapsed.
Subagents and workflow-spawned agents no longer receive the workflow tool that only top-level sessions can use.
Memory results are now presented as historical context that should be verified against live sources before use.
Interactive grok sessions now start in ask mode by default again instead of auto.
Minimal mode no longer truncates subagent wake-turn replies after the first token.

Performance

Concurrent subagents no longer trigger rate-limit errors during bursts.
MCP server connections no longer delay session startup when many servers are configured.

1.0.8 — 2026-08-20

Features

MCP servers can now ask for form input or URL consent through the same popup used for questions.
Ctrl+S now stashes the current prompt draft so you can send something else and restore it later.
/workflow now autocompletes saved workflow names and shows only valid runs for pause/resume/stop/save.

Bug Fixes

Downloading a folder that contains only one file now produces a zip that still extracts as a folder.
Failed tool calls for tools the model invented now clearly state the tool does not exist.
Status line refresh timer now uses consistent naming and no longer shows errors on deliberately hidden rows.
Workflow agent rows now display current context usage instead of cumulative token counts.
Follow-up messages now send immediately while waiting on a subagent or task, including after using /btw.
Subagents and workflow-spawned agents no longer see the workflow tool, which can only be launched from a top-level session.

Performance

Opening many subagents at once no longer freezes the interface while loading their history.
Concurrent subagents now start much faster and no longer freeze the parent session.

1.0.7 — 2026-08-19

Features

Users hitting startup timeouts can now raise the connect budget with the GROK_CONNECT_UI_TIMEOUT_SECS environment variable.
Permission prompts now show "Always allow" and "Never allow" options by default.
Users can now delete scheduled background loops directly from the tray.
Status line command scripts can now run on a timer via refresh_interval in config.toml.
Permission prompts now offer a 'Never allow' choice for MCP tools and web-fetch domains that persists per project.
Workflows tab added to the extensions modal (Ctrl+L or /plugins) listing installed workflows with name, source, and description.
New /workflows command opens the Workflows catalog tab; use /workflow runs to view live workflow runs.
Bare /workflow (or /workflow runs) now lists active and recent workflow runs with status and progress instead of usage help.
Workflows row added to the Ctrl+P command palette, opening the Workflows catalog tab.

Bug Fixes

MCP server connections in non-interactive sessions no longer incorrectly require authentication for tokenless servers.
Fixed startup timeouts caused by concurrent auth refreshes across multiple sessions.
Tool call loops are interrupted earlier to avoid wasting time on repeated identical actions.
Subagents no longer receive the ask-user-question tool.
Bare email addresses are now turned into clickable mailto links in the pager.

1.0.6 — 2026-08-18

Breaking Changes

Subagent spawning no longer accepts capability_mode; tool access is now controlled only by agent type.

Features

Shift+arrow keys now extend text selections in the prompt like a standard text field.
Optional status line can now display live session info or script output at the bottom of the pager.
grok clone can now fetch a repo into a content store and mount a projected working tree.

Bug Fixes

Fixed session startup hangs on large or unhealthy git repositories.
Queued messages during goals no longer starve, and editing queued prompts works reliably.
Consent notice on first launch now shows clickable links and handles keyboard/mouse correctly.
Ctrl+C then edit a prompt now correctly removes the original text from the conversation.
Double-clicking a terminal command result now shows the complete output instead of a preview.
Consent notice links are now stricter and more reliable on all terminals.
Video generation now surfaces a clear ZDR error instead of raw API responses when output storage is required.
Project hooks on Windows now correctly expand $CLAUDE_PROJECT_DIR when invoking PowerShell scripts.

source

Enterprise channel 1.0.13

2026-08-28 · enterprise · enterprise 1.0.13 notes from GitHub changelog

1.0.13 — 2026-08-28

Features

Length-truncated responses now continue automatically instead of failing the turn.
Hooks can now ask the user to confirm a tool call instead of always allowing or denying.
Hooks can now request deferral or add context shown to the model after a tool runs.
Session close now records detailed timing data for performance analysis.
Credit limit upsell now offers a Try Again button to retry the last prompt.
Pasted images now show a live pixel preview in the prompt box on iTerm2.

Bug Fixes

Transient inference failures (stalls, drops, 5xx) now retry automatically instead of ending the turn.
Windows users can now correctly open ~/.grok and worktree sessions.
Session data is now more reliably saved after prompts and on power loss.
Compaction failures now show the actual error instead of a generic message.
Truncation error messages now show the right guidance instead of suggesting an unhelpful retry.
Truncated tool calls are now executed instead of failing the turn when arguments are complete.
Images larger than 2000px no longer brick sessions on many-image requests.
Wrapped hyperlinks in the pager now remain fully clickable on Windows Terminal instead of only the first line.
Recurring scheduled tasks now include a reminder to stop the monitor when work finishes.
Scheduled task IDs are now full UUID strings, preventing collisions when tasks are created in the same millisecond.

Performance

Subagent spawning is faster when connections drop during bursts.
Session startup with MCP servers is now much faster when auth is already configured.
MCP server startup no longer stalls behind a fixed batch size.
CLI downloads are now compressed, making fresh installs and updates substantially faster.

1.0.12 — 2026-08-27

Features

Credit-limit upsell now includes a Try Again option. Max-tier users see the same question modal without Upgrade tier.

Bug Fixes

Copying wrapped table cells no longer inserts unwanted spaces at line breaks.
MCP server connections that fail transiently now retry instead of staying unavailable.
Waiting on subagents after an interjection no longer blocks on unrelated background work.
Prompt blocks now show friendly hook descriptions instead of internal IDs.
Truncation error message no longer suggests asking for a shorter answer.
Context estimates for conversations with reasoning are now more accurate.
Token usage after rewinding or switching modes now matches the actual server-reported count instead of overestimating.
Context bar now updates immediately when auto-compaction begins instead of waiting until it finishes.
File system monitoring for .NET projects no longer creates excessive watchers on Linux.
Auto recap no longer appears while a background task or monitor wake turn is streaming.

Performance

Worktree creation is faster by skipping stale reflog copies.

1.0.11 — 2026-08-26

Features

Headless sessions are now browsable in the resume picker without mixing into default history.
Default permission mode for new interactive sessions is now configurable.
Turn duration now appears in session history after resume.
Turn footers (Worked for, cancelled, failed) now appear after /resume.
mkdir and touch no longer prompt in auto mode or safe-command lists.
Subagent messages are now allowed automatically in permission Auto mode.
Headless sessions can now auto-allow permission prompts via a startup hint.

Bug Fixes

Permission prompts for common command chains and subcommands are now more reliable.
Blocked prompts no longer appear in conversation history or scrollback after restart.
Background monitors no longer have a 10-hour default timeout.
Mouse input at the right margin no longer types characters into the prompt on certain terminals.
Pasting text ending in a newline no longer accidentally submits the prompt on some terminals.
Auto mode now shows a permission card when the classifier blocks an action on interactive sessions.
Image previews no longer leave ghost artifacts when using the Kitty protocol on Warp.
Expanded Execute tool output no longer snaps closed during live progress.
/voice now falls back to parec/arecord on older PipeWire installs.

Performance

Background command waits now finish as soon as the process exits.

1.0.10 — 2026-08-24

Performance

grok clone now reuses matching local checkouts as linked worktrees for faster session creation.

1.0.9 — 2026-08-24

Features

New sessions in the interactive TUI now start in auto mode by default for fewer approval prompts.
/feedback now supports attaching images (screenshots etc.).
The / slash menu is now ordered by recency and groups skills below commands for easier navigation.
/workflow now accepts --agent-budget N or an agent_budget JSON field to set child-agent limits.
/workflow now accepts --effort LEVEL or an effort JSON field to set child reasoning effort.
MCP servers now receive a grok-cli User-Agent header so providers can attribute traffic.
Dashboard now lets you start new agents in Auto mode via Shift+Tab.
Workflows now appear alongside skills in the model prompt so the agent can launch them by name.
/edit-prompt now opens an external editor from the full TUI, not only minimal mode.
Up arrow on an empty prompt now selects queued follow-ups first instead of opening history.
/plugin now works as an alias for /plugins in the TUI.
Plugin-provided agents now appear in the /agents modal and can be toggled.
/minimal and /fullscreen now switch instantly inside the running session without restarting or losing the current turn.
The configurable status line row now appears at the bottom of minimal mode as well as fullscreen.
grok clone now fetches only the selected branch tip by default, with --full-history for legacy full clones.
grok clone can now reuse a matching local checkout as a linked worktree instead of always fetching from the network.

Bug Fixes

Subagent tasks are more resilient to temporary rate limits and will retry instead of failing immediately.
Narrow allow rules for Bash commands now correctly permit file writes without extra prompts.
Rules containing markdown headings now render correctly in the prompt.
/new and /clear now preserve the last-chosen reasoning effort instead of resetting to the model default.
Prompt dropdowns (slash, history, file search) now have consistent left alignment with the composer.
History now shows every command, slash command, and memory note in the order you typed them.
Collapsed tool-call rows no longer show a vertical accent line beside the status diamond.
Hint text above the prompt now lines up with the text you type.
Pressing Esc while peeking at a conversation from the dashboard now closes open modals instead of leaving the view.
/copy now preserves markdown formatting such as headings and code fences.
Sending a new message while a shell command is running now moves the command to the background instead of cancelling it.
grok mcp add now correctly treats bare http(s):// URLs as HTTP servers instead of stdio commands.
Plan mode no longer incorrectly blocks subagent spawns as file edits outside plan.md.
MCP servers that share a URL but have different names are now kept as separate entries instead of being collapsed.
Subagents and workflow-spawned agents no longer receive the workflow tool that only top-level sessions can use.
Memory results are now presented as historical context that should be verified against live sources before use.
Interactive grok sessions now start in ask mode by default again instead of auto.
Minimal mode no longer truncates subagent wake-turn replies after the first token.

Performance

Concurrent subagents no longer trigger rate-limit errors during bursts.
MCP server connections no longer delay session startup when many servers are configured.

1.0.8 — 2026-08-20

Features

MCP servers can now ask for form input or URL consent through the same popup used for questions.
Ctrl+S now stashes the current prompt draft so you can send something else and restore it later.
/workflow now autocompletes saved workflow names and shows only valid runs for pause/resume/stop/save.

Bug Fixes

Downloading a folder that contains only one file now produces a zip that still extracts as a folder.
Failed tool calls for tools the model invented now clearly state the tool does not exist.
Status line refresh timer now uses consistent naming and no longer shows errors on deliberately hidden rows.
Workflow agent rows now display current context usage instead of cumulative token counts.
Follow-up messages now send immediately while waiting on a subagent or task, including after using /btw.
Subagents and workflow-spawned agents no longer see the workflow tool, which can only be launched from a top-level session.

Performance

Opening many subagents at once no longer freezes the interface while loading their history.
Concurrent subagents now start much faster and no longer freeze the parent session.

1.0.7 — 2026-08-19

Features

Users hitting startup timeouts can now raise the connect budget with the GROK_CONNECT_UI_TIMEOUT_SECS environment variable.
Permission prompts now show "Always allow" and "Never allow" options by default.
Users can now delete scheduled background loops directly from the tray.
Status line command scripts can now run on a timer via refresh_interval in config.toml.
Permission prompts now offer a 'Never allow' choice for MCP tools and web-fetch domains that persists per project.
Workflows tab added to the extensions modal (Ctrl+L or /plugins) listing installed workflows with name, source, and description.
New /workflows command opens the Workflows catalog tab; use /workflow runs to view live workflow runs.
Bare /workflow (or /workflow runs) now lists active and recent workflow runs with status and progress instead of usage help.
Workflows row added to the Ctrl+P command palette, opening the Workflows catalog tab.

Bug Fixes

MCP server connections in non-interactive sessions no longer incorrectly require authentication for tokenless servers.
Fixed startup timeouts caused by concurrent auth refreshes across multiple sessions.
Tool call loops are interrupted earlier to avoid wasting time on repeated identical actions.
Subagents no longer receive the ask-user-question tool.
Bare email addresses are now turned into clickable mailto links in the pager.

1.0.6 — 2026-08-18

Breaking Changes

Subagent spawning no longer accepts capability_mode; tool access is now controlled only by agent type.

Features

Shift+arrow keys now extend text selections in the prompt like a standard text field.
Optional status line can now display live session info or script output at the bottom of the pager.
grok clone can now fetch a repo into a content store and mount a projected working tree.

Bug Fixes

Fixed session startup hangs on large or unhealthy git repositories.
Queued messages during goals no longer starve, and editing queued prompts works reliably.
Consent notice on first launch now shows clickable links and handles keyboard/mouse correctly.
Ctrl+C then edit a prompt now correctly removes the original text from the conversation.
Double-clicking a terminal command result now shows the complete output instead of a preview.
Consent notice links are now stricter and more reliable on all terminals.
Video generation now surfaces a clear ZDR error instead of raw API responses when output storage is required.
Project hooks on Windows now correctly expand $CLAUDE_PROJECT_DIR when invoking PowerShell scripts.

source

Alpha channel 1.0.13

2026-08-28 · alpha · alpha 1.0.13 notes from GitHub changelog

1.0.13 — 2026-08-28

Features

Length-truncated responses now continue automatically instead of failing the turn.
Hooks can now ask the user to confirm a tool call instead of always allowing or denying.
Hooks can now request deferral or add context shown to the model after a tool runs.
Session close now records detailed timing data for performance analysis.
Credit limit upsell now offers a Try Again button to retry the last prompt.
Pasted images now show a live pixel preview in the prompt box on iTerm2.

Bug Fixes

Transient inference failures (stalls, drops, 5xx) now retry automatically instead of ending the turn.
Windows users can now correctly open ~/.grok and worktree sessions.
Session data is now more reliably saved after prompts and on power loss.
Compaction failures now show the actual error instead of a generic message.
Truncation error messages now show the right guidance instead of suggesting an unhelpful retry.
Truncated tool calls are now executed instead of failing the turn when arguments are complete.
Images larger than 2000px no longer brick sessions on many-image requests.
Wrapped hyperlinks in the pager now remain fully clickable on Windows Terminal instead of only the first line.
Recurring scheduled tasks now include a reminder to stop the monitor when work finishes.
Scheduled task IDs are now full UUID strings, preventing collisions when tasks are created in the same millisecond.

Performance

Subagent spawning is faster when connections drop during bursts.
Session startup with MCP servers is now much faster when auth is already configured.
MCP server startup no longer stalls behind a fixed batch size.
CLI downloads are now compressed, making fresh installs and updates substantially faster.

1.0.12 — 2026-08-27

Features

Credit-limit upsell now includes a Try Again option. Max-tier users see the same question modal without Upgrade tier.

Bug Fixes

Copying wrapped table cells no longer inserts unwanted spaces at line breaks.
MCP server connections that fail transiently now retry instead of staying unavailable.
Waiting on subagents after an interjection no longer blocks on unrelated background work.
Prompt blocks now show friendly hook descriptions instead of internal IDs.
Truncation error message no longer suggests asking for a shorter answer.
Context estimates for conversations with reasoning are now more accurate.
Token usage after rewinding or switching modes now matches the actual server-reported count instead of overestimating.
Context bar now updates immediately when auto-compaction begins instead of waiting until it finishes.
File system monitoring for .NET projects no longer creates excessive watchers on Linux.
Auto recap no longer appears while a background task or monitor wake turn is streaming.

Performance

Worktree creation is faster by skipping stale reflog copies.

1.0.11 — 2026-08-26

Features

Headless sessions are now browsable in the resume picker without mixing into default history.
Default permission mode for new interactive sessions is now configurable.
Turn duration now appears in session history after resume.
Turn footers (Worked for, cancelled, failed) now appear after /resume.
mkdir and touch no longer prompt in auto mode or safe-command lists.
Subagent messages are now allowed automatically in permission Auto mode.
Headless sessions can now auto-allow permission prompts via a startup hint.

Bug Fixes

Permission prompts for common command chains and subcommands are now more reliable.
Blocked prompts no longer appear in conversation history or scrollback after restart.
Background monitors no longer have a 10-hour default timeout.
Mouse input at the right margin no longer types characters into the prompt on certain terminals.
Pasting text ending in a newline no longer accidentally submits the prompt on some terminals.
Auto mode now shows a permission card when the classifier blocks an action on interactive sessions.
Image previews no longer leave ghost artifacts when using the Kitty protocol on Warp.
Expanded Execute tool output no longer snaps closed during live progress.
/voice now falls back to parec/arecord on older PipeWire installs.

Performance

Background command waits now finish as soon as the process exits.

1.0.10 — 2026-08-24

Performance

grok clone now reuses matching local checkouts as linked worktrees for faster session creation.

1.0.9 — 2026-08-24

Features

New sessions in the interactive TUI now start in auto mode by default for fewer approval prompts.
/feedback now supports attaching images (screenshots etc.).
The / slash menu is now ordered by recency and groups skills below commands for easier navigation.
/workflow now accepts --agent-budget N or an agent_budget JSON field to set child-agent limits.
/workflow now accepts --effort LEVEL or an effort JSON field to set child reasoning effort.
MCP servers now receive a grok-cli User-Agent header so providers can attribute traffic.
Dashboard now lets you start new agents in Auto mode via Shift+Tab.
Workflows now appear alongside skills in the model prompt so the agent can launch them by name.
/edit-prompt now opens an external editor from the full TUI, not only minimal mode.
Up arrow on an empty prompt now selects queued follow-ups first instead of opening history.
/plugin now works as an alias for /plugins in the TUI.
Plugin-provided agents now appear in the /agents modal and can be toggled.
/minimal and /fullscreen now switch instantly inside the running session without restarting or losing the current turn.
The configurable status line row now appears at the bottom of minimal mode as well as fullscreen.
grok clone now fetches only the selected branch tip by default, with --full-history for legacy full clones.
grok clone can now reuse a matching local checkout as a linked worktree instead of always fetching from the network.

Bug Fixes

Subagent tasks are more resilient to temporary rate limits and will retry instead of failing immediately.
Narrow allow rules for Bash commands now correctly permit file writes without extra prompts.
Rules containing markdown headings now render correctly in the prompt.
/new and /clear now preserve the last-chosen reasoning effort instead of resetting to the model default.
Prompt dropdowns (slash, history, file search) now have consistent left alignment with the composer.
History now shows every command, slash command, and memory note in the order you typed them.
Collapsed tool-call rows no longer show a vertical accent line beside the status diamond.
Hint text above the prompt now lines up with the text you type.
Pressing Esc while peeking at a conversation from the dashboard now closes open modals instead of leaving the view.
/copy now preserves markdown formatting such as headings and code fences.
Sending a new message while a shell command is running now moves the command to the background instead of cancelling it.
grok mcp add now correctly treats bare http(s):// URLs as HTTP servers instead of stdio commands.
Plan mode no longer incorrectly blocks subagent spawns as file edits outside plan.md.
MCP servers that share a URL but have different names are now kept as separate entries instead of being collapsed.
Subagents and workflow-spawned agents no longer receive the workflow tool that only top-level sessions can use.
Memory results are now presented as historical context that should be verified against live sources before use.
Interactive grok sessions now start in ask mode by default again instead of auto.
Minimal mode no longer truncates subagent wake-turn replies after the first token.

Performance

Concurrent subagents no longer trigger rate-limit errors during bursts.
MCP server connections no longer delay session startup when many servers are configured.

1.0.8 — 2026-08-20

Features

MCP servers can now ask for form input or URL consent through the same popup used for questions.
Ctrl+S now stashes the current prompt draft so you can send something else and restore it later.
/workflow now autocompletes saved workflow names and shows only valid runs for pause/resume/stop/save.

Bug Fixes

Downloading a folder that contains only one file now produces a zip that still extracts as a folder.
Failed tool calls for tools the model invented now clearly state the tool does not exist.
Status line refresh timer now uses consistent naming and no longer shows errors on deliberately hidden rows.
Workflow agent rows now display current context usage instead of cumulative token counts.
Follow-up messages now send immediately while waiting on a subagent or task, including after using /btw.
Subagents and workflow-spawned agents no longer see the workflow tool, which can only be launched from a top-level session.

Performance

Opening many subagents at once no longer freezes the interface while loading their history.
Concurrent subagents now start much faster and no longer freeze the parent session.

1.0.7 — 2026-08-19

Features

Users hitting startup timeouts can now raise the connect budget with the GROK_CONNECT_UI_TIMEOUT_SECS environment variable.
Permission prompts now show "Always allow" and "Never allow" options by default.
Users can now delete scheduled background loops directly from the tray.
Status line command scripts can now run on a timer via refresh_interval in config.toml.
Permission prompts now offer a 'Never allow' choice for MCP tools and web-fetch domains that persists per project.
Workflows tab added to the extensions modal (Ctrl+L or /plugins) listing installed workflows with name, source, and description.
New /workflows command opens the Workflows catalog tab; use /workflow runs to view live workflow runs.
Bare /workflow (or /workflow runs) now lists active and recent workflow runs with status and progress instead of usage help.
Workflows row added to the Ctrl+P command palette, opening the Workflows catalog tab.

Bug Fixes

MCP server connections in non-interactive sessions no longer incorrectly require authentication for tokenless servers.
Fixed startup timeouts caused by concurrent auth refreshes across multiple sessions.
Tool call loops are interrupted earlier to avoid wasting time on repeated identical actions.
Subagents no longer receive the ask-user-question tool.
Bare email addresses are now turned into clickable mailto links in the pager.

1.0.6 — 2026-08-18

Breaking Changes

Subagent spawning no longer accepts capability_mode; tool access is now controlled only by agent type.

Features

Shift+arrow keys now extend text selections in the prompt like a standard text field.
Optional status line can now display live session info or script output at the bottom of the pager.
grok clone can now fetch a repo into a content store and mount a projected working tree.

Bug Fixes

Fixed session startup hangs on large or unhealthy git repositories.
Queued messages during goals no longer starve, and editing queued prompts works reliably.
Consent notice on first launch now shows clickable links and handles keyboard/mouse correctly.
Ctrl+C then edit a prompt now correctly removes the original text from the conversation.
Double-clicking a terminal command result now shows the complete output instead of a preview.
Consent notice links are now stricter and more reliable on all terminals.
Video generation now surfaces a clear ZDR error instead of raw API responses when output storage is required.
Project hooks on Windows now correctly expand $CLAUDE_PROJECT_DIR when invoking PowerShell scripts.

source

Enterprise channel 1.0.12

2026-08-27 · enterprise · enterprise 1.0.12 notes from GitHub changelog

1.0.12 — 2026-08-27

Bug Fixes

Copying wrapped table cells no longer inserts unwanted spaces at line breaks.
MCP server connections that fail transiently now retry instead of staying unavailable.
Waiting on subagents after an interjection no longer blocks on unrelated background work.
Prompt blocks now show friendly hook descriptions instead of internal IDs.
Truncation error message no longer suggests asking for a shorter answer.
Context estimates for conversations with reasoning are now more accurate.
Token usage after rewinding or switching modes now matches the actual server-reported count instead of overestimating.
Context bar now updates immediately when auto-compaction begins instead of waiting until it finishes.
File system monitoring for .NET projects no longer creates excessive watchers on Linux.
Auto recap no longer appears while a background task or monitor wake turn is streaming.

Performance

Worktree creation is faster by skipping stale reflog copies.

source

Alpha channel 1.0.12

2026-08-27 · alpha · alpha 1.0.12 notes from GitHub changelog

1.0.12 — 2026-08-27

Bug Fixes

Copying wrapped table cells no longer inserts unwanted spaces at line breaks.
MCP server connections that fail transiently now retry instead of staying unavailable.
Waiting on subagents after an interjection no longer blocks on unrelated background work.
Prompt blocks now show friendly hook descriptions instead of internal IDs.
Truncation error message no longer suggests asking for a shorter answer.
Context estimates for conversations with reasoning are now more accurate.
Token usage after rewinding or switching modes now matches the actual server-reported count instead of overestimating.
Context bar now updates immediately when auto-compaction begins instead of waiting until it finishes.
File system monitoring for .NET projects no longer creates excessive watchers on Linux.
Auto recap no longer appears while a background task or monitor wake turn is streaming.

Performance

Worktree creation is faster by skipping stale reflog copies.

source

Enterprise channel 1.0.11

2026-08-26 · enterprise · enterprise 1.0.11 notes from GitHub changelog

1.0.11 — 2026-08-26

Features

Headless sessions are now browsable in the resume picker without mixing into default history.
Default permission mode for new interactive sessions is now configurable.
Turn duration now appears in session history after resume.
Turn footers (Worked for, cancelled, failed) now appear after /resume.
mkdir and touch no longer prompt in auto mode or safe-command lists.
Subagent messages are now allowed automatically in permission Auto mode.
Headless sessions can now auto-allow permission prompts via a startup hint.

Bug Fixes

Permission prompts for common command chains and subcommands are now more reliable.
Blocked prompts no longer appear in conversation history or scrollback after restart.
Background monitors no longer have a 10-hour default timeout.
Mouse input at the right margin no longer types characters into the prompt on certain terminals.
Pasting text ending in a newline no longer accidentally submits the prompt on some terminals.
Auto mode now shows a permission card when the classifier blocks an action on interactive sessions.
Image previews no longer leave ghost artifacts when using the Kitty protocol on Warp.
Expanded Execute tool output no longer snaps closed during live progress.
/voice now falls back to parec/arecord on older PipeWire installs.

Performance

Background command waits now finish as soon as the process exits.

source

Alpha channel 1.0.11

2026-08-26 · alpha · alpha 1.0.11 notes from GitHub changelog

1.0.11 — 2026-08-26

Features

Headless sessions are now browsable in the resume picker without mixing into default history.
Default permission mode for new interactive sessions is now configurable.
Turn duration now appears in session history after resume.
Turn footers (Worked for, cancelled, failed) now appear after /resume.
mkdir and touch no longer prompt in auto mode or safe-command lists.
Subagent messages are now allowed automatically in permission Auto mode.
Headless sessions can now auto-allow permission prompts via a startup hint.

Bug Fixes

Permission prompts for common command chains and subcommands are now more reliable.
Blocked prompts no longer appear in conversation history or scrollback after restart.
Background monitors no longer have a 10-hour default timeout.
Mouse input at the right margin no longer types characters into the prompt on certain terminals.
Pasting text ending in a newline no longer accidentally submits the prompt on some terminals.
Auto mode now shows a permission card when the classifier blocks an action on interactive sessions.
Image previews no longer leave ghost artifacts when using the Kitty protocol on Warp.
Expanded Execute tool output no longer snaps closed during live progress.
/voice now falls back to parec/arecord on older PipeWire installs.

Performance

Background command waits now finish as soon as the process exits.

source

Alpha 1.0.9 (superseded)

2026-08-24 · alpha · alpha briefly served 1.0.9; now 1.0.10

1.0.9 — 2026-08-24

Features

New sessions in the interactive TUI now start in auto mode by default for fewer approval prompts.
/feedback now supports attaching images (screenshots etc.).
The / slash menu is now ordered by recency and groups skills below commands for easier navigation.
/workflow now accepts --agent-budget N or an agent_budget JSON field to set child-agent limits.
/workflow now accepts --effort LEVEL or an effort JSON field to set child reasoning effort.
MCP servers now receive a grok-cli User-Agent header so providers can attribute traffic.
Dashboard now lets you start new agents in Auto mode via Shift+Tab.
Workflows now appear alongside skills in the model prompt so the agent can launch them by name.
/edit-prompt now opens an external editor from the full TUI, not only minimal mode.
Up arrow on an empty prompt now selects queued follow-ups first instead of opening history.
/plugin now works as an alias for /plugins in the TUI.
Plugin-provided agents now appear in the /agents modal and can be toggled.
/minimal and /fullscreen now switch instantly inside the running session without restarting or losing the current turn.
The configurable status line row now appears at the bottom of minimal mode as well as fullscreen.
grok clone now fetches only the selected branch tip by default, with --full-history for legacy full clones.
grok clone can now reuse a matching local checkout as a linked worktree instead of always fetching from the network.

Bug Fixes

Subagent tasks are more resilient to temporary rate limits and will retry instead of failing immediately.
Narrow allow rules for Bash commands now correctly permit file writes without extra prompts.
Rules containing markdown headings now render correctly in the prompt.
/new and /clear now preserve the last-chosen reasoning effort instead of resetting to the model default.
Prompt dropdowns (slash, history, file search) now have consistent left alignment with the composer.
History now shows every command, slash command, and memory note in the order you typed them.
Collapsed tool-call rows no longer show a vertical accent line beside the status diamond.
Hint text above the prompt now lines up with the text you type.
Pressing Esc while peeking at a conversation from the dashboard now closes open modals instead of leaving the view.
/copy now preserves markdown formatting such as headings and code fences.
Sending a new message while a shell command is running now moves the command to the background instead of cancelling it.
grok mcp add now correctly treats bare http(s):// URLs as HTTP servers instead of stdio commands.
Plan mode no longer incorrectly blocks subagent spawns as file edits outside plan.md.
MCP servers that share a URL but have different names are now kept as separate entries instead of being collapsed.
Subagents and workflow-spawned agents no longer receive the workflow tool that only top-level sessions can use.
Memory results are now presented as historical context that should be verified against live sources before use.
Interactive grok sessions now start in ask mode by default again instead of auto.
Minimal mode no longer truncates subagent wake-turn replies after the first token.

Performance

Concurrent subagents no longer trigger rate-limit errors during bursts.
MCP server connections no longer delay session startup when many servers are configured.

source

Alpha channel 1.0.8

2026-08-20 · alpha · alpha 1.0.8 notes from GitHub changelog

1.0.8 — 2026-08-20

Features

MCP servers can now ask for form input or URL consent through the same popup used for questions.
Ctrl+S now stashes the current prompt draft so you can send something else and restore it later.
/workflow now autocompletes saved workflow names and shows only valid runs for pause/resume/stop/save.

Bug Fixes

Downloading a folder that contains only one file now produces a zip that still extracts as a folder.
Failed tool calls for tools the model invented now clearly state the tool does not exist.
Status line refresh timer now uses consistent naming and no longer shows errors on deliberately hidden rows.
Workflow agent rows now display current context usage instead of cumulative token counts.
Follow-up messages now send immediately while waiting on a subagent or task, including after using /btw.

Performance

Opening many subagents at once no longer freezes the interface while loading their history.
Concurrent subagents now start much faster and no longer freeze the parent session.

source

X Stack changelog

Grok 4.6

Grok 4.7

Grok Bot

0.70.2 — 2026-10-11

New

Improved

Fixed

0.68.1 — 2026-10-07

New

Improved

Fixed

0.68.0 — 2026-10-05

New

Improved

Fixed

Grok Build

1.0.79 — 2026-11-08

Features

Bug Fixes

1.0.79 — 2026-11-08

Features

Bug Fixes

1.0.78 — 2026-11-06

Features

Bug Fixes

Performance

1.0.77 — 2026-11-05

Features

Bug Fixes

1.0.77 — 2026-11-05

Features

Bug Fixes

1.0.77 — 2026-11-05

Features

Bug Fixes

1.0.76 — 2026-11-04

Features

Bug Fixes

1.0.75 — 2026-11-04

Features

Bug Fixes

1.0.74 — 2026-11-02

Features

Bug Fixes

Performance

1.0.73 — 2026-10-28

Features

Bug Fixes

Performance

1.0.73 — 2026-10-28

Features

Bug Fixes

Performance

1.0.72 — 2026-10-27

Features

Bug Fixes

1.0.71 — 2026-10-26

Features

Bug Fixes

1.0.70 — 2026-10-23

Features

Bug Fixes

1.0.70 — 2026-10-23

Features

Bug Fixes

1.0.69 — 2026-10-23

Features

Bug Fixes

Performance

1.0.69 — 2026-10-23

Features

Bug Fixes

Performance

1.0.68 — 2026-10-22

Features

Bug Fixes

1.0.67 — 2026-10-21

Features

Bug Fixes