* feat: add terminal shortcuts panel for mobile users
Slide-out panel providing touch-friendly shortcut buttons (Esc, Tab,
Shift+Tab, Arrow Up/Down) and scroll-to-bottom for the terminal.
Integrates into the new modular shell architecture by exposing
terminalRef and wsRef from useShellRuntime hook and reusing the
existing sendSocketMessage utility.
Includes localization keys for en, ja, ko, and zh-CN.
* fix: replace dual touch/click handlers with unified pointer events
Prevents double-fire on touch devices by removing onTouchEnd handlers
and using a single onClick for all interactions (mouse, touch, keyboard).
onPointerDown with preventDefault handles focus steal prevention.
Also clears pending close timer before scheduling a new one to avoid
stale timeout overlap.
Addresses CodeRabbit review feedback on PR #411.
---------
Co-authored-by: Haileyesus <118998054+blackmammoth@users.noreply.github.com>
* fix: preserve pending permission requests across WebSocket reconnections
- Store WebSocketWriter reference in active sessions for reconnection
- Add reconnectSessionWriter() to swap writer when client reconnects
- Add getPendingApprovalsForSession() to query pending permissions by session
- Add get-pending-permissions WebSocket message handler on server
- Add pending-permissions-response handler on frontend
- Query pending permissions on WebSocket reconnect and session change
- Reconnect SDK output writer when client resumes an active session
* fix: address CodeRabbit review feedback for websocket-permission PR
- Use consistent session matching in pending-permissions-response handler,
checking both currentSessionId and selectedSession.id (matching the
session-status handler pattern)
- Guard get-pending-permissions with isClaudeSDKSessionActive check to
prevent returning permissions for inactive sessions
* fix: prevent React 18 batching from losing messages during session sync
- Add lastLoadedSessionIdRef to skip redundant session reloads
- Add length-based guards to sessionMessages -> chatMessages sync effect
- Track whether chat is actively processing to prevent stale overwrites
- Re-enable sessionMessages sync with proper guards that only fire when
server data actually grew (new messages from server), not during re-renders
* fix: reset message sync refs on session switch
Reset prevSessionMessagesLengthRef and isInitialLoadRef when
selectedSession changes to ensure correct message sync behavior
when switching between sessions. Move ref declarations before
their first usage for clarity.
* fix: address CodeRabbit review — composite cache key and empty sync
- Use composite key (sessionId:projectName:provider) for session load
deduplication instead of sessionId alone, preventing stale cache hits
when switching projects or providers with the same session ID
- Allow zero-length sessionMessages to sync to chatMessages, so server
clearing a session's messages is correctly reflected in the UI
# Features
- File drag and drop upload: Support uploading files and folders via drag and drop
- Binary file handling: Detect binary files and display a friendly message instead of trying to edit them
- Folder download: Download folders as ZIP files (using JSZip library)
- Context menu integration: Full right-click context menu for file operations (rename, delete, copy path, download, new file/folder)
The modelUsage debug log ran on every streamed SDK message, but
modelUsage is only populated on result messages. This produced
repeated "Model was sent using: []" console output for every
non-result message during streaming.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Show dynamic browser tab title based on selected project's name, post-fixed with "CloudCLI UI" when a project is selected, improving navigation across tabs.
fix wrong regex replace of Claude project path
related #447, reopen due to forced-push
to reproduce error steps, let's try
Create a folder with @ in name like @test
Add this folder as new project in CloudCLI
Choose Claude tool in new Session
Star by typing sth 'hi'
In the dev tools, you will see errors ajax response said that session does not find for 'some-session-id'
The main problem is current encode path doesn't encode '@' to '-' as Claude did
I reversed code Claude-SDK, file 'cli.js' to find exactly regex (using in PR) that used to encode path under ~/.claude/projects/<encoded-project-name-path/<session-id>.jsonl
* feat: add copy icon for user messages
Expose a copy control on user chat bubbles so previous content can be reused quickly.
* fix: Copy control is effectively hidden on touch devices
* fix: copyTextToClipboard doesn't need timer
---------
Co-authored-by: dev <dev@host.local>
Co-authored-by: Haileyesus <118998054+blackmammoth@users.noreply.github.com>
* feat: integrate Gemini AI agent provider
- Core Backend: Ported gemini-cli.js and gemini-response-handler.js to establish the CLI bridge. Registered 'gemini' as an active provider within index.js.
- Core Frontend: Extended QuickSettingsPanel.jsx, Settings.jsx, and AgentListItem.jsx to render the Gemini provider option, models (gemini-pro, gemini-flash, etc.), and handle OAuth states.
- WebSocket Pipeline: Added support for gemini-command executions in backend and payload processing of gemini-response and gemini-error streams in useChatRealtimeHandlers.ts. Resolved JSON double-stringification and sessionId stripping issues in the transmission handler.
- Platform Compatibility: Added scripts/fix-node-pty.js postinstall script and modified posix_spawnp calls with sh -c wrapper to prevent ENOEXEC and MacOS permission errors when spawning the gemini headless binary.
- UX & Design: Imported official Google Gemini branding via GeminiLogo.jsx and gemini-ai-icon.svg. Updated translations (chat.json) for en, zh-CN, and ko locales.
* fix: propagate gemini permission mode from settings to cli
- Added Gemini Permissions UI in Settings to toggle Auto Edit and YOLO modes
- Synced gemini permission mode to localStorage
- Passed permissionMode in useChatComposerState for Gemini commands
- Mapped frontend permission modes to --yolo and --approval-mode options in gemini-cli.js
* feat(gemini): Refactor Gemini CLI integration to use stream-json
- Replaced regex buffering text-system with NDJSON stream parsing
- Added fallback for restricted models like gemini-3.1-pro-preview
* feat(gemini): Render tool_use and tool_result UI bubbles
- Forwarded gemini tool NDJSON objects to the websocket
- Added React state handlers in useChatRealtimeHandlers to match Claude's tool UI behavior
* feat(gemini): Add native session resumption and UI token tracking
- Captured cliSessionId from init events to map ClaudeCodeUI's chat sessionId directly into Gemini's internal session manager.
- Updated gemini-cli.js spawn arguments to append the --resume proxy flag instead of naively dumping the accumulated chat history into the command prompt.
- Handled result stream objects by proxying total_tokens back into the frontend's claude-status tracker to natively populate the UI label.
- Eliminated gemini-3 model proxy filter entirely.
* fix(gemini): Fix static 'Claude' name rendering in chat UI header
- Added "gemini": "Gemini" translation strings to messageTypes across English, Korean, and Chinese loc dictionaries.
- Updated AssistantThinkingIndicator and MessageComponent ternary checks to identify provider === 'gemini' and render the appropriate brand label instead of statically defaulting to Claude.
* feat: Add Gemini session persistence API mapping and Sidebar UI
* fix(gemini): Watch ~/.gemini/sessions for live UI updates
Added the .gemini/sessions directory to PROVIDER_WATCH_PATHS so that Chokidar emits projects_updated websocket events when new Gemini sessions are created or modified, fixing live sidebar updates.
* fix(gemini): Fix Gemini authentication status display in Settings UI
- Injected 'checkGeminiAuthStatus' into the Settings.jsx React effect hook so that the UI can poll and render the 'geminiAuthStatus' state.
- Updated 'checkGeminiCredentials()' inside server/routes/cli-auth.js to read from '~/.gemini/oauth_creds.json' and '~/.gemini/google_accounts.json', resolving the email address correctly.
* Use logo-only icon for gemini
* feat(gemini): Add Gemini 3 preview models to UI selection list
* Fix Gemini CLI session resume bug and PR #422 review nitpicks
* Fix Gemini tool calls disappearing from UI after completion
* fix(gemini): resolve outstanding PR #422 feedback and stabilize gemini CLI timeouts
* fix(gemini): resolve resume flag and shell session initialization issues
This commit addresses the remaining PR comments for the Gemini CLI integration:
- Moves the `--resume` flag logic outside the prompt command block, ensuring Gemini sessions correctly resume even when a new prompt isn't passed.
- Updates `handleShellConnection` to correctly lookup the native `cliSessionId` from the internal `sessionId` when spawning Gemini sessions in a plain shell.
- Refactors dynamic import of `sessionManager.js` back to a native static import for code consistency.
* chore: fix TypeScript errors and remove gemini CLI dependency
* fix: use cross-spawn on Windows to resolve gemini.cmd correctly
---------
Co-authored-by: Haileyesus <118998054+blackmammoth@users.noreply.github.com>
* feat: persist active tab across reloads via localStorage (closes#387)
Remember the last active tab in localStorage instead of always resetting
to 'chat'. Also stop force-switching to chat tab on session change,
so users stay on their preferred tab (shell, git, etc.).
* fix: validate localStorage tab value and add try-catch for restricted contexts
Address CodeRabbit feedback: validate stored activeTab against known
tab IDs before using it, and wrap localStorage access in try-catch
to prevent crashes in restricted environments.
* fix: intercept slash commands in handleSubmit to pass arguments correctly
When a user types a slash command with arguments (e.g. `/feature implement dark mode`)
and presses Enter, the command was not being intercepted as a slash command. Instead,
the raw text was sent as a regular message to the Claude API, which responded with
"Unknown skill: feature".
Root cause: the command autocomplete menu (useSlashCommands) detects commands via the
regex `/(^|\s)\/(\S*)$/` which only matches when the cursor is right after the command
name with no spaces. As soon as the user types a space to add arguments, the pattern
stops matching, the menu closes, and pressing Enter falls through to handleSubmit which
sends the text as a plain message — completely bypassing command execution.
This fix adds a slash command interceptor at the top of handleSubmit:
- Checks if the trimmed input starts with `/`
- Extracts the command name (text before the first space)
- Looks up the command in the loaded slashCommands list
- If found, delegates to executeCommand() which properly extracts arguments
via regex and sends them to the backend /api/commands/execute endpoint
- The backend then replaces $ARGUMENTS, $1, $2 etc. in the command template
Changes:
- Added `slashCommands` to the destructured return of useSlashCommands hook
- Added slash command interception logic in handleSubmit before message dispatch
- Added `executeCommand` and `slashCommands` to handleSubmit dependency array
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: address review — pass rawInput param and cleanup UI state
- Pass trimmedInput to executeCommand to avoid stale closure reads
- Add UI cleanup (images, command menu, textarea) before early return
- Update executeCommand signature to accept optional rawInput parameter
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat: add HOST environment variable for configurable bind address
Allow users to specify which host/IP address the servers should bind to
via the HOST environment variable. Defaults to 0.0.0.0 (all interfaces)
to maintain backward compatibility.
This enables users to restrict the server to localhost only (127.0.0.1)
for security purposes or bind to a specific network interface.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* fix: use correct proxy host when HOST is set to specific interface
When HOST is set to a specific interface (e.g., 192.168.1.100), the Vite
proxy needs to connect to that same host, not localhost. This fix:
- Adds proxyHost logic that uses localhost when HOST=0.0.0.0, otherwise
uses the specific HOST value for proxy targets
- Adds DISPLAY_HOST to show a user-friendly URL in console output
(0.0.0.0 isn't a connectable address)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
---------
Co-authored-by: Michael Fork <mjfork@users.noreply.github.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Haileyesus <118998054+blackmammoth@users.noreply.github.com>
* feat: add "Load all messages" button for long conversations
Scrolling up through long conversations requires many "load more" cycles.
This adds a "Load all messages" floating button that fetches the entire
conversation history in one shot.
- Floating overlay pill appears after each batch finishes loading, persists 2s
- Shows loading spinner while fetching all messages
- Shows green "All messages loaded" confirmation for 1s before disappearing
- Preserves scroll position when bulk-loading (no viewport jump)
- Ref-based guards prevent scroll handler from re-fetching after load-all
- Performance warning shown; "Scroll to bottom" resets visible cap
- Clean state reset on session switch
- i18n keys for en and zh-CN
Note: default page size (20) and visible cap (100) are unchanged.
These could be increased in a follow-up or made configurable via settings.
* re-implement load-all feature for new TS architecture
Subagent (Task/Explore) results were rendered as raw JSON
`[{"type":"text","text":"..."}]` with literal `\n` instead of
formatted markdown. The root cause was that the API sometimes
returns content blocks as a serialized JSON string rather than
a parsed array. The existing `Array.isArray()` check missed this
case and fell through to `String()`, displaying raw JSON.
Now the Task tool result handler tries to JSON.parse string content
before checking for array structure, matching the pattern already
used by exit_plan_mode.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
