Add 'sftp.transfer.preparing' key to en.ts and zh-CN.ts so the
indeterminate transfer state shows localized text instead of the
raw i18n key.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* chore: remove 65 debug console.log statements from production code
Remove bracketed debug traces ([SFTP navigateTo], [SFTPBackend],
[ManagedSourceSync], [AutoSync], [CloudSync], [Settings], etc.)
across 16 files. These were development logging that shipped to
production, creating noise in the console.
Also clean up dead variables left behind after log removal
(hotkeyDebug, results, verification reads).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* chore: remove 43 unused exports and dead type definitions
Remove export keywords from symbols that are never imported outside
their defining file. Symbols still used internally keep their
definitions; symbols not used at all are removed entirely.
Removed entirely: TerminalLine, SessionLogsSettings, KDFParams,
SyncManagerConfig, GoogleTokenResponse, OneDriveTokenResponse,
getSyncStatusColor, resolveHostTerminalAppearance,
TerminalAppearanceDefaults.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: resolve SFTP tab connection key race condition in workspace mode
When rapidly switching focus between workspace panes, the single
pendingConnectionKeyRef could be overwritten before the tracking
effect mapped it to the created tab. This left tabs unmapped in
tabConnectionKeyMapRef, causing duplicate tabs on subsequent switches.
Replace the two-step async mechanism (pendingConnectionKeyRef + deferred
tracking effect) with a synchronous onTabCreated callback on connect().
The callback fires immediately after the tab ID is determined, before
any async SSH work begins, eliminating the race window entirely.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: scope SFTP transfers to active connection and prevent stale session lookups
Two fixes for workspace focus-switching issues:
1. Transfer queue now filters by the active connection's host, so
switching focus between workspace panes only shows transfers
relevant to the currently displayed SFTP tab.
2. Move sftpSessionsRef.delete() before the async closeSftp() call
to close the race window where concurrent code could look up a
stale sftpId that the backend has already removed, causing
"SFTP session not found" errors.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: allow SFTP focus switching during file transfers
Active transfers should not block workspace focus-following. Transfers
run on their own sftpId independent of the active tab, and forceNewTab
preserves old connections, so switching focus is safe.
Only interactive operations (text editor, permissions dialog, file
opener, file watches) still block host switching.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: refresh correct SFTP tab after transfer completes during focus switch
When a transfer completes while focus has switched to a different host,
refresh was targeting the currently active pane instead of the pane that
initiated the transfer.
Add optional tabId parameter to navigateTo() and refresh() so callers
can target a specific tab. Capture the tab ID at transfer start and use
it for the post-transfer refresh, ensuring the correct tab's file list
is updated regardless of which tab is currently focused.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: auto-reconnect SFTP when session is lost during navigation
When navigateTo() detected a missing or expired SFTP session, it
cleared the connection to null, showing the empty "Select a host"
state. Now it delegates to handleSessionError(), which triggers the
existing reconnection mechanism — keeping files visible while
reconnecting in the background.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* perf: eliminate redundant stat calls before file transfers
Before this change, each file transfer performed 3-4 stat calls over
the network before the progress bar started moving:
1. startTransfer: stat to get file size (~100ms)
2. processTransfer: stat again if size was 0 (~100ms)
3. Conflict check: stat source file for mtime (~100ms)
4. Backend: stat again if totalBytes missing (~100ms)
Now:
- Use the source pane's cached file list for size and mtime (zero
network cost) instead of stat calls in startTransfer
- Store sourceLastModified on TransferTask so the conflict check can
use it directly instead of a redundant source stat
- Backend already skips stat when totalBytes is provided
This saves ~200-300ms of network round-trips per file before the
progress bar starts moving.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* perf: show immediate progress feedback during transfer setup
The progress bar previously stayed at 0% for ~500ms-1s while the
backend acquired an isolated SFTP channel and waited for the first
data chunk. Users perceived this as the transfer being "sluggish".
Now start simulated progress immediately for all single-file
transfers (not just non-streaming ones). When the first real progress
update arrives from the backend, the simulation is stopped and real
progress takes over seamlessly. This gives instant visual feedback
that the transfer is in progress.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: show accurate transfer progress instead of simulated values
The progress system had fundamental issues:
1. Simulated progress ran for ALL transfers including streaming ones,
creating fake progress that could reach 95% while real progress
was at 60%. The Math.max ratchet prevented regression, so users
saw inflated numbers.
2. Speed and remaining time were based on simulated data during the
setup phase, giving misleading estimates.
Changes:
- Only use simulated progress for non-streaming transfers (no real
progress callback available). Streaming transfers get real data.
- Remove the double ratchet (Math.max) from onProgress — the backend
already enforces monotonic progress, so the frontend should trust
the reported values directly.
- Show an indeterminate "preparing..." state during the setup phase
(channel acquisition, conflict check) instead of fake progress.
This honestly communicates that the transfer is starting.
- Hide speed and remaining time during the indeterminate phase since
no real data is available yet.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* refactor: remove dead progress simulation and non-streaming transfer code
startStreamTransfer is always available in Electron, so:
- Remove the non-streaming fallback path in transferFile() that read
entire files into memory with no progress reporting
- Remove startProgressSimulation / stopProgressSimulation and all
related refs (progressIntervalsRef, useSimulatedProgress,
hasStreamingTransfer)
- Remove the cleanup effect for progress intervals
All transfers now use the streaming path with real backend-reported
progress. The indeterminate "preparing..." state covers the setup
phase until the first real progress arrives.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* perf: reduce SFTP transfer concurrency from 64 to 4
64 parallel SFTP read/write requests overwhelmed servers, causing
the first chunk response to be delayed by 46+ seconds. Reducing to
4 concurrent requests provides a responsive first progress update
(~1-2s) while still offering significant speedup over sequential
streaming.
Also adds timing logs to the transfer pipeline (processTransfer,
transferFile, downloadFile, uploadFile) to aid future diagnostics.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address review findings from PR #443
Critical fixes:
- Fix refresh/navigateTo type signatures to include the tabId option
parameter — previously it was silently ignored, making tab-targeted
refresh non-functional
- Fix handleSessionError/reconnection in navigateTo for background tabs:
when called with explicit tabId, update that specific tab instead of
the active tab (which could be a different host)
- Fix uploadExternalFiles to capture and pass tabId for post-upload
refresh (was missing, only uploadExternalEntries was fixed)
Medium fixes:
- Restore Math.max monotonic ratchet on single-file onProgress to guard
against any non-monotonic backend values
- Add stat fallback in processTransfer to populate sourceLastModified
when file is not in the pane's visible file list (filtered/search)
- Adjust TRANSFER_CONCURRENCY from 4 to 8 as a better throughput/
responsiveness balance
Cleanup:
- Remove all debug timing logs (console.log with Transfer/downloadFile/
uploadFile prefixes) from both frontend and backend
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: prevent background tab navigation from rolling back active tab
Two P1 fixes from automated review:
1. navSeqRef race: navigateTo uses a per-side sequence counter, so a
background tab refresh would bump it and cause the active tab's
concurrent navigation to think it was superseded, restoring
previousPath instead of applying the fetched files. Now when
navSeqRef is superseded but tabNavSeqRef still matches, the fetched
result is applied (it's valid for this tab — only a different tab
bumped the counter).
2. Auto-follow tear down: needsNewTab only checked hostId, so same
host with different session-time overrides (port/protocol) would
reuse the tab and close the old SFTP session, aborting any
in-flight transfer. Now needsNewTab is true whenever the current
connection is alive, always preserving it with forceNewTab.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: respect global terminal appearance settings
* feat: add reset to global terminal appearance
* fix: preserve legacy host appearance overrides
* fix: show legacy appearance reset controls
* refactor: reorder terminal global reset actions
* refactor: present global theme as theme option
* refactor: present global font as font option
* feat: inline approval gate for tool execution
Replace SDK-level needsApproval with Promise-based approval gate inside
tool execute functions. The SDK stream stays alive while the UI shows
inline approve/reject buttons on ToolCall blocks.
Changes:
- Add approvalGate.ts: Promise-based approval system with event listeners
- tools.ts: requestApproval() inside execute for confirm mode
- tool-call.tsx: inline approval buttons and keyboard shortcuts
- ChatMessageList.tsx: subscribe to approval events, render approval UI
- useAIChatStreaming.ts: remove old useToolApproval hook integration
- AIChatSidePanel.tsx: remove old approval hook, clean up unused destructuring
- systemPrompt.ts: update confirm mode to not ask for text confirmation
- preload.cjs: filter pager env var prefixes from terminal display
- mcpServerBridge.cjs: add approval gate for ACP/MCP write operations
- aiBridge.cjs: wire IPC for MCP approval response and main window getter
- preload.cjs: add onMcpApprovalRequest/respondMcpApproval APIs
* fix: scope approval gate by chatSessionId and replay for late subscribers
Address Codex PR review comments:
- Add chatSessionId to ApprovalRequest for session isolation
- Scope clearAllPendingApprovals(chatSessionId?) to only clear
approvals belonging to the target session
- Add replayPendingApprovals() so late-mounting ChatMessageList
picks up approvals that fired while unmounted
- Scope MCP clearPendingApprovals in aiBridge cancel handler to
effectiveChatSessionId instead of clearing all
- Pass chatSessionId through MCP approval IPC flow
* chore: remove old approval flow code
- Delete useToolApproval.ts (unused hook)
- Delete InlineApprovalCard.tsx (replaced by ToolCall inline buttons)
- Remove stale comments referencing old hook in AIChatSidePanel
- Remove unused ai.chat.toolApprovalTitle i18n key from en/zh-CN
* fix: session-scoped approval gate and MCP replay survival
- handleStop passes activeSessionId to clearAllPendingApprovals
- setupMcpApprovalBridge stores MCP approvals in pendingApprovals map
so they survive ChatMessageList unmount/remount cycles
- ChatMessageList accepts activeSessionId prop and filters standalone
MCP approval blocks to the current session only
- AIChatSidePanel passes activeSessionId to ChatMessageList
* fix: filter PTY exec marker echoes and exit code lines from terminal
Extend filterMcpMarkers in preload.cjs to strip all shell-visible
artifacts from AI command execution:
- Echoed printf start marker: printf '%s\n' '__NCMCP_..._S'
- Echoed exit code restoration: (exit $__nc)
- PowerShell: Write-Output, $global:LASTEXITCODE, $__nc assignment
- Fish: set __nc $status
- Cmd: echo __NCMCP_...
- Widen guard to also trigger on __nc and PAGER=cat strings
* fix: scope SDK approvals, deny MCP on no renderer, fix memo comparator
- createCattyTools accepts chatSessionId and passes it to
requestApproval so SDK approvals can be matched by
clearAllPendingApprovals(activeSessionId) on stop
- useAIChatStreaming passes sessionId to createCattyTools
- mcpServerBridge: deny (resolve false) when no renderer window is
available instead of auto-approving, preserving confirm mode safety
- ChatMessageList: add activeSessionId to React.memo comparator so
switching sessions triggers re-render for correct MCP approval filter
* fix: MCP listener lifecycle, approval timeout, and UI sync on stop
- Move setupMcpApprovalBridge from ChatMessageList to AIChatSidePanel
so the IPC listener survives tab/panel switches
- Add 5-minute auto-deny timeout to requestApproval to prevent
indefinite isStreaming hangs when user walks away
- Add onApprovalCleared listener system: clearAllPendingApprovals now
notifies UI subscribers so ChatMessageList removes stale cards
- ChatMessageList subscribes to onApprovalCleared to sync local state
* fix: main-process approval timeout and full tool args in payload
- Add 5-minute auto-deny timeout to requestApprovalFromRenderer
matching the renderer-side requestApproval behavior
- Forward all tool params (excluding chatSessionId) to approval UI
instead of cherry-picking command/input/path, so sftpRename
oldPath/newPath and other tool-specific args are visible
* fix: move MCP bridge to TerminalLayer, narrow terminal filter guard
- Move setupMcpApprovalBridge from AIChatSidePanel to TerminalLayer
so the IPC listener stays alive regardless of side panel tab.
AIChatSidePanel only mounts when activeSidePanelTab==='ai'.
- Narrow preload.cjs filter guard back to __NCMCP_ only, preventing
false-positive stripping of user scripts containing __nc or PAGER=cat
* fix: eliminate PTY wrapper echo leakage and duplicate prompts
- Posix wrapper now emits 2 lines instead of 4: start marker + command
on line 1 (joined with ;), end marker + exit on line 2. This
eliminates the duplicate prompt echo from the separate start marker.
- Rename __nc to __NCMCP_rc in all shell variants (posix/fish/powershell)
so every wrapper variable contains the __NCMCP_ prefix. The preload
guard `data.includes("__NCMCP_")` now reliably catches ALL wrapper
artifacts regardless of chunk boundaries.
- Update all filterMcpMarkers regex patterns to match the restructured
wrapper format and renamed variable.
* fix: sync main-process approval timeout with renderer UI cleanup
- When requestApprovalFromRenderer times out, send IPC event
netcatty:ai:mcp:approval-cleared to renderer so stale approval
cards are removed
- Add onMcpApprovalCleared preload bridge for the new IPC channel
- setupMcpApprovalBridge now subscribes to cleared events, removes
timed-out entries from pendingApprovals and notifies clearedListeners
so ChatMessageList drops the stale card
* fix: surface denied inline approvals as errors in UI
- Detect error or denial payloads ("error" string or "ok: false")
returned by tools when the user denies an execution
- Set isError: true on the tool-result message so the ToolCall UI
renders it as a failure (red/rejected) instead of a success (green)
* Add AI support for local terminal sessions
* Fix local AI session metadata and shell safety
* Fix local session cloning and multi-exec errors
* Refactor local shell detection helpers
* Fix local shell helper import path
* Fix CJS imports in renderer
* Use ESM local shell helpers in renderer
* Normalize local shell paths and platform metadata
Add localized aria-label to reset, zoom in, zoom out, and close
buttons. Add i18n keys for common.reset, common.zoomIn, common.zoomOut
in en and zh-CN locales.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: bundle claude-code-acp to prevent crash when binary is missing (#400)
When users select Claude Code in the AI module, the app spawns
`claude-code-acp` via ACP. Previously only the `claude` CLI was checked
during agent discovery, so if `claude-code-acp` was not on PATH the
spawn would fail with ENOENT and crash the Electron main process.
- Add `@zed-industries/claude-code-acp` as a bundled dependency
- Add `resolveClaudeAcpBinaryPath()` that checks PATH first, then
falls back to the npm-bundled binary (mirrors Codex pattern)
- Use the resolver in both the primary and fallback ACP provider paths
- Update agent discovery to detect agents via bundled ACP binary when
the standalone CLI is not installed
Closes#400
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: add claude-code-acp and its deps to asarUnpack
In packaged Electron builds, files inside app.asar cannot be executed
by child_process.spawn. Add claude-code-acp and its runtime dependencies
to asarUnpack so the binary is accessible in production.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: migrate from deprecated claude-code-acp to claude-agent-acp
The @zed-industries/claude-code-acp package has been renamed to
@zed-industries/claude-agent-acp (bin: claude-agent-acp). Update all
references across the codebase:
- package.json: replace dep with @zed-industries/claude-agent-acp@0.22.2
- electron-builder.config.cjs: update asarUnpack entries, remove stale
deps (diff, minimatch) no longer needed by the new package
- shellUtils.cjs: update binary name and require.resolve path
- aiBridge.cjs: update acpCommand, ALLOWED_AGENT_COMMANDS, isClaudeAgent
- settings types, i18n locales: update command references
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add option to auto-open sidebar on host connect
Closes#396
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: only auto-open SFTP sidebar for SSH/Mosh connections
Use allowlist (ssh, mosh) instead of blocklist so telnet and other
non-SSH protocols don't trigger SFTP sidebar which would fail.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: support auto-open SFTP for Quick Connect / temporary sessions
Build a minimal Host from session data when hostId is not in the vault,
so Quick Connect sessions also trigger auto-open SFTP sidebar.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: sync SFTP auto-open sidebar setting across windows via IPC
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: skip local terminals and preserve username for temp sessions
- Don't fallback protocol to 'ssh' so local terminals are excluded
- Include session.username in synthesized Host for Quick Connect
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: support multimodal attachments (images, PDFs, files) in AI chat
Previously uploaded images were displayed in the UI but never sent to
the AI model, and non-image files (PDF, text) were silently rejected.
- Rename useImageUpload → useFileUpload; accept image/*, PDF, and text/*
- Rename ChatMessageImage → ChatMessageAttachment with filePath support
- Build multimodal SDK messages (ImagePart/FilePart) for Catty Agent
- Fix ACP agent path: images inline, non-image files via local path hint
so ACP agents (Claude Code, etc.) read them with native file access
- Use Electron webUtils.getPathForFile() for reliable file path capture
- Compact user message bubble padding
Closes#294 (AI file upload issues)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: show real tool names in AI chat instead of ACP wrapper names
- Unwrap ACP dynamic tool calls in serializeStreamChunk to extract
real tool name, args, and toolCallId from chunk.input
- Simplify MCP tool name prefixes (mcp__server__tool → tool)
- Pass toolCallId from ACP tool-call events to match tool results
- Prevent onToolResult from overwriting correct names with wrapper name
- Build toolCallNames map in ChatMessageList for tool result display
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: backward-compatible fallback for legacy `images` field in chat messages
Persisted sessions may still have `images` instead of `attachments`.
Add `?? m.images` fallback in SDK message builder and renderer so
historical image attachments are not silently dropped after upgrade.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: broaden file type support and handle pasted files without path
- Accept all file types except video/audio (instead of allowlist)
so .json, .yaml, .sh, etc. are not silently rejected
- For ACP agents, save pasted/virtual files (no filePath) to temp
directory so the agent can still read them
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: use managed temp dir for pasted ACP attachments
Use tempDirBridge.getTempFilePath() instead of manual os.tmpdir() path
so pasted file attachments are tracked by the app's cleanup system.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Implements automatic three-way merge for cloud sync, replacing the
binary USE_REMOTE/USE_LOCAL conflict resolution. Same principle as
Git's merge algorithm.
After every successful sync, a "base snapshot" is saved (encrypted
with AES-256-GCM using the derived master key). When a conflict is
detected, the system performs per-entity merge by ID:
- Items added on one side → included
- Items deleted on one side (unchanged on other) → removed
- Items modified on one side only → take that version
- Both sides modified same item → prefer local
- One side deleted + other modified → keep modification
Additional improvements:
- Per-provider sync base to prevent cross-provider contamination
- Deep merge for nested settings (terminalSettings, customKeyBindings)
- Entity merge for array-valued settings (customTerminalThemes)
- KnownHost deduplication by (hostname, port, keyType)
- Chunked base encoding to avoid stack overflow on large vaults
- Base cleared on provider disconnect/reconnect
- Correct version numbering after multi-provider merge
Closes#378
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: auto-close tab when user actively exits a session
When a user intentionally exits a session (e.g. typing `exit`, `logout`,
or Ctrl+D), the tab is now automatically closed instead of showing the
"Start Over" disconnected page. This matches the behavior of macOS
Terminal and other popular terminal emulators.
Network errors, timeouts, and server-initiated disconnects still show
the disconnected page with the Start Over option, so users can reconnect.
In workspace mode, only the individual terminal pane is closed, not the
entire workspace.
Implementation:
- Backend bridges now include a `reason` field in exit events to
distinguish stream-level exits ("exited") from connection errors
("error"), timeouts ("timeout"), and connection closes ("closed")
- SSH bridge captures real exit code from stream "exit" event instead
of hardcoding 0
- Frontend auto-closes session only when reason is "exited"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address review feedback for auto-close feature
1. Pass exit event to onSessionExit in local shell path (line 757)
to prevent undefined access when checking evt.reason
2. Change Telnet socket close reason from "exited" to "closed" since
a clean socket close can also be server-initiated (idle timeout,
remote shutdown), not just user exit
3. Change Serial port close reason from "exited" to "closed" since
port close can be from device disconnect, not user action
Only SSH stream close and local/mosh process exit (node-pty onExit)
now use reason "exited", which correctly represents user-initiated exits.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: only mark SSH exit as "exited" when stream exit event fired
ssh2's stream "close" event fires whenever the channel closes, not
only on normal shell exit. If the network drops and the channel closes
without a preceding "exit" event, the reason was incorrectly set to
"exited", causing the tab to auto-close instead of showing the
disconnected/Start Over page.
Now tracks whether stream "exit" actually fired via a flag, and only
uses reason "exited" in that case. Otherwise falls back to "closed".
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: classify mosh non-zero exits as errors
Mosh process exiting with a non-zero code typically indicates a
connection or auth failure. Mark these as reason "error" so the
disconnected/Start Over UI is shown instead of auto-closing the tab.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: treat SSH signal-terminated exits as disconnects
ssh2's stream "exit" event also fires for signal terminations (e.g.
SIGHUP from server idle timeout, SIGTERM from admin kill), where code
is null and signal is set. These are not user-initiated exits and
should show the disconnected/Start Over page.
Now only sets streamExited=true when there's a numeric exit code and
no signal present.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: distinguish abnormal local PTY exits from user exits
Local shell terminated by signal or crashing on startup should show
the disconnected UI, not auto-close the tab. Now only marks as
reason "exited" when exitCode is 0 and no signal, matching the same
logic used for mosh.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: use signal presence to distinguish local shell exit reason
For local shells, non-zero exit codes are common in user-initiated
exits (e.g. typing `exit` after a failed command returns that
command's exit code). Use signal presence instead: signal means the
process was killed externally (show disconnected UI), no signal
means normal process exit (auto-close tab).
Mosh keeps exitCode-based logic since non-zero there indicates
connection/auth failure, not user exit.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add "paste only" option for snippets (no auto-execute)
Add a noAutoRun flag to snippets that pastes the command into the
terminal without appending a carriage return, so users can review
and edit before manually pressing Enter.
Applies to all snippet execution paths: snippet runner (new session),
keyboard shortcut, and startup command.
Closes#371
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: use clearer wording "仅粘贴" instead of "仅上屏"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: skip onCommandExecuted for paste-only shortcut snippets
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: persist noAutoRun on save and apply to Scripts panel clicks
- Include noAutoRun in handleSubmit serialization (was being lost)
- Pass noAutoRun through ScriptsSidePanel click handler to TerminalLayer
so paste-only snippets work from the Scripts panel too
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add skip TLS verification option for AI providers
Self-hosted AI endpoints (vLLM, text-generation-webui, etc.) often use
self-signed TLS certificates which Node.js rejects by default, causing
502 Bad Gateway errors. Add a per-provider "Skip TLS certificate
verification" checkbox that sets rejectUnauthorized=false on both
streaming and non-streaming requests.
Ref #294
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: surface real error message instead of generic 502 Bad Gateway
- Pass the actual bridge error message in statusText so Vercel AI SDK
shows the real cause (e.g. "HTTP is only allowed for localhost",
"URL host is not in the allowed list", TLS errors)
- Show real error details for 5xx provider errors instead of generic
"The AI provider returned a server error" message
Previously all connection-level errors were masked as "Bad Gateway"
making it impossible for users to diagnose configuration issues.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: pass server error body details through to the user
- Read HTTP error response body before resolving (was resolving before
body was read, losing the error detail)
- Parse OpenAI-compatible JSON error format to extract error.message
- Return error Response with body+statusText for non-2xx instead of
empty stream, so Vercel AI SDK shows the real server error
- Now users see e.g. "502 model not loaded" instead of just "Bad Gateway"
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: widen link modifier key dropdown to prevent text wrapping
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Revert "fix: widen link modifier key dropdown to prevent text wrapping"
This reverts commit 1f75686391.
* fix: apply skipTLSVerify to model listing requests
ModelSelector.aiFetch() didn't pass providerId, so the provider-level
skipTLSVerify was not applied when refreshing/listing models. Add
skipTLSVerify as a direct parameter alongside the provider lookup.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: keep error detail in Response body, not statusText
statusText only accepts single-line Latin-1 — multiline or non-ASCII
error messages from self-hosted gateways would throw TypeError before
the AI SDK could read them. Move detailed error to body instead.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: return JSON error body for AI SDK compatibility, fix FetchBridge type
- Wrap error responses in OpenAI-compatible JSON format so Vercel AI
SDK's failedResponseHandler extracts the message correctly instead
of showing a blank error
- Update FetchBridge type to match the expanded aiFetch parameter list
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: add ASCII statusText fallback for non-OpenAI SDK providers
Anthropic/Google SDKs fall back to Response.statusText when they can't
parse the error body. Add safe ASCII statusText alongside the JSON body.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add web search and URL fetch tools for AI agent
Add web_search and url_fetch tools to Catty Agent, allowing the AI to
search the internet for current information and fetch webpage content.
- Support 5 search providers: Tavily, Exa, Bocha, Zhipu, SearXNG
- Settings UI with provider selection, API key encryption, and config
- web_search is conditional on config; url_fetch is always available
- Both tools are read-only and work in all permission modes (incl. observer)
- aiFetch skipHostCheck for AI tool requests to arbitrary URLs
- System prompt guidelines for when to use search/fetch
- i18n support (en + zh-CN)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address code review findings (SSRF, key exposure, state race)
- P1: Restore SSRF protection when skipHostCheck is true — still block
localhost, RFC1918, link-local, and cloud metadata endpoints; only
skip the domain allowlist for public HTTPS hosts
- P2: Move web search API key decryption to main process via dedicated
IPC handler, matching the existing provider key security model
- P2: Use configRef to avoid stale closure in async settings callbacks
that could overwrite newer user changes
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address second review — DNS rebinding, url_fetch approval, maxResults
- P1: url_fetch now requires approval in confirm mode (outbound GET is
a side effect that could exfiltrate data via query strings)
- P1: Add DNS resolution check when skipHostCheck is set — resolve
hostname and reject if any IP is private/loopback/link-local, blocking
DNS rebinding attacks against internal services
- P2: Slice search results after provider call to enforce maxResults
consistently (Zhipu and SearXNG ignore the limit parameter)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address third review — localhost/IPv6 SSRF, API key blur race
- P1: Block localhost/loopback when skipHostCheck is enabled — restructure
isAllowedFetchUrl to check private hosts first in the skipHostCheck path,
preventing access to local services on allowlisted ports
- P1: Handle IPv6 private ranges (fc00::/7, fe80::/10, ::ffff: mapped),
strip brackets from URL.hostname, block [::1] and fd00:: addresses
- P2: Guard handleApiKeyBlur against provider change during async
encryption — skip stale write if provider switched while encrypting
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address fourth review — main-process key isolation, SearXNG compat
- P1: Replace aiWebSearchDecryptKey IPC with __WEB_SEARCH_KEY__ placeholder
pattern — renderer never sees plaintext keys; main process replaces
placeholder in headers before HTTP request, matching provider key flow
- P1: Search API requests use normal allowlist path (not skipHostCheck),
so SearXNG on localhost/HTTP/private networks works via aiSyncWebSearch;
only url_fetch uses skipHostCheck for arbitrary public HTTPS URLs
- P2: Remove needsApproval from url_fetch — treat as read-only like
sftp_read_file, consistent with observer mode allowlist
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address fifth review — private LAN providers, maxResults default
- P1: Allow private-IP hosts that are explicitly in the provider/search
allowlist (e.g. https://192.168.x.x model providers or SearXNG)
- P2: Remove .default(5) from web_search maxResults schema so the user's
configured maxResults setting is used when the model omits the param
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address sixth review — HTTPS scope, config gate, redirects
- P2: Scope HTTP exception to private/LAN IPs only — remote allowlisted
hosts still require HTTPS to protect API keys in transit
- P2: Gate web_search tool on complete config (API key for providers that
require it, apiHost for SearXNG) to avoid advertising a broken tool
- P2: Add redirect following (up to 5 hops) to aiFetch for url_fetch —
handles 301/302/307 for short links, www canonicalization, etc.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address seventh review — redirect SSRF, decrypt race, HTTPS-only
- P1: Revalidate each redirect hop against SSRF guards (allowlist check
+ DNS resolution) before following, preventing open-redirect SSRF
- P2: Add sequence counter to API key decryption effect — stale promise
results from a previous provider are discarded on provider switch
- P3: Restrict url_fetch to HTTPS-only URLs, matching the skipHostCheck
policy that already rejects HTTP in the bridge
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address eighth review — OS resolver, allowlisted HTTP hosts
- P1: Use dns.lookup (OS resolver) instead of dns.resolve4/6 for private
IP checks — matches what http.request actually connects to, respects
/etc/hosts, mDNS, and other local resolver sources
- P2: Allow HTTP for any explicitly allowlisted host (not just literal
private IPs), so self-hosted SearXNG at http://searxng.lan works
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address ninth review — HTTP scope, blur ordering, decrypt flag
- P1: Narrow HTTP exception to web search apiHost only — AI provider
endpoints remain HTTPS-only to protect credentials in transit
- P2: Add blur sequence counter to prevent out-of-order encryption
results from overwriting newer API key saves
- P2: Reset isDecrypting flag when cancelling decrypt on provider switch,
preventing permanently disabled API key input
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address tenth review — DNS pinning, prompt/tool alignment
- P1: Pin validated DNS result to the HTTP request via custom lookup
function, preventing TOCTOU/DNS-rebinding between validation and
actual connection
- P2: Extract isWebSearchReady() helper and use it consistently in
both tool registration and system prompt, so the model isn't told
web search is available when config is incomplete
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address eleventh review — single DNS lookup, redirect pinning, CGNAT
- P1: Combine DNS validation and pinning into a single lookup call,
eliminating the TOCTOU window between hasPrivateResolution and pinnedLookup
- P1: Pin DNS for redirect targets too — resolve/validate/pin in one step
before following each redirect hop
- P2: Add 100.64.0.0/10 (CGNAT) to private IP ranges for Tailscale and
similar CGNAT-addressed internal services
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address twelfth review — apiHost validation, sync on enable
- P2: Validate apiHost is a well-formed URL in isWebSearchReady(),
preventing tool exposure when user enters a malformed host
- P2: Add webSearchConfig.enabled to sync effect deps so the main
process gets updated immediately when the toggle changes
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: remove DNS-level SSRF checks that break fakedns/proxy environments
DNS resolution validation (dns.lookup + IP pinning) breaks in proxy
environments where fakedns resolves all domains to LAN addresses.
Revert to hostname-level checks only (blocking localhost, 127.0.0.1,
metadata endpoints, etc.) which are sufficient without false positives.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: resolve empty catch block lint warning
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add a fourth option 'Write + Prompt on Read' that allows clipboard
writes but shows a confirmation dialog before granting read access.
This lets users benefit from remote copy (tmux/vim) while maintaining
control over clipboard reads.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add osc52Clipboard setting (off/write-only/read-write), default write-only
- Fix UTF-8 decoding: use TextDecoder instead of atob for non-ASCII content
- Support clipboard read requests when mode is read-write
- Add settings UI with Select dropdown and i18n (en + zh-CN)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add release/** to ESLint ignores (build artifacts were being linted)
- Remove unused eslint-disable directives in useAutoSync and useSettingsState
- Add missing setTerminalSettings dependency to rehydrateAllFromStorage
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add settings cloud sync support (closes#347)
Expand SyncPayload.settings to include all syncable user preferences
(theme, appearance, terminal, keyboard, editor, SFTP). Add
collectSyncableSettings/applySyncableSettings helpers in syncPayload.ts,
wire rehydrateAllFromStorage through App.tsx and SettingsPage.tsx so
in-memory React state updates after a cloud download.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: include settings in auto-sync uploads and sync empty customCSS
P1: useAutoSync.buildPayload now includes collectSyncableSettings()
so settings are uploaded alongside vault data.
P2: customCSS uses != null check instead of truthy, so clearing CSS
on one device is properly synced to others.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: include settings in auto-sync change detection hash
Settings-only changes (theme, terminal options, etc.) now trigger
auto-sync uploads. The data hash comparison includes the settings
snapshot alongside vault data.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: trigger auto-sync on settings changes and sync custom terminal themes
P1: Added settingsVersion (derived from all synced settings via useMemo)
to useAutoSync debounce effect dependencies. Settings-only changes now
trigger auto-sync uploads.
P2: Custom terminal themes (STORAGE_KEY_CUSTOM_THEMES) are now included
in the sync payload so custom themes are available on other devices.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: reload custom theme store after sync, include in change detection
P1: customThemeStore.loadFromStorage() is now called in
rehydrateAllFromStorage so synced custom themes are immediately
reflected in the live theme store.
P2a: customThemes added to settingsVersion dependencies so custom
theme edits trigger auto-sync.
P2b: Empty custom themes array is now preserved in sync payload
to properly propagate theme deletion.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: notify subscribers after custom theme store reload
loadFromStorage now calls notify() to trigger useSyncExternalStore
subscribers, so synced custom terminal themes are immediately
visible in all windows after apply.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add auto-update toggle setting (closes#346)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: re-check auto-update toggle when startup timer fires
Address review feedback: the startup check effect now re-reads the
toggle from localStorage when the delayed timer fires, so toggling
off after launch cancels the pending check. Also avoids setting
hasCheckedOnStartupRef when disabled, allowing re-enable to trigger
a check without restart.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address review feedback on auto-update toggle
P1: When autoDownload=false, onUpdateAvailable no longer transitions
to 'downloading' status. Instead keeps autoDownloadStatus idle so
the manual download link surfaces correctly.
P2: Added reactive autoUpdateEnabled state (synced via storage event)
as a dependency to the startup check effect. Re-enabling the toggle
mid-session now re-triggers the deferred startup check.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: address P1/P2 review feedback on auto-update toggle
P1: Main process update-available handler now checks updater.autoDownload
before setting _lastStatus to 'downloading'. When autoDownload=false,
status stays 'idle' so late-opened windows don't hydrate to a stuck
0% download state.
P2: useUpdateCheck now accepts autoUpdateEnabled as a prop from the
caller instead of relying solely on storage events (which don't fire
in the same window). SettingsPage passes settings.autoUpdateEnabled
directly, so toggling in the current window takes effect immediately.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: preserve update-available info for late-opening windows
When autoDownload is off, use status 'available' (instead of 'idle')
in the main process snapshot so late-opening windows can hydrate
version info. The renderer maps 'available' to hasUpdate=true while
keeping autoDownloadStatus='idle' for the manual download path.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: re-schedule auto-check on re-enable and guard startup timer
- IPC handler now calls startAutoCheck(2000) when re-enabling so the
user gets automatic checks without restarting the app.
- startAutoCheck timer checks updater.autoDownload at fire time, so
if the renderer disables auto-update via IPC before the 5s startup
timer fires, the check is skipped.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: deduplicate auto-check scheduling and clear error on fallback success
P1: startAutoCheck now cancels any existing timer before scheduling
a new one, preventing duplicate concurrent checks from multiple
windows or re-enable toggles.
P2: checkNow fallback now clears manualCheckStatus='error' when
electron-updater successfully finds an update (res.available=true),
so the UI shows 'available' instead of a stale error state.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: only reschedule on actual re-enable and hydrate cache before toggle check
P2: Track previous autoDownload state in IPC handler so startAutoCheck
is only called on actual false→true transitions, not on every window
mount that syncs the current value.
P3: Move cache hydration (STORAGE_KEY_UPDATE_LATEST_RELEASE) before
the auto-update toggle check so cached update info is always visible
even when automatic updates are disabled.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: persist auto-update preference in main process across restarts
Read/write auto-update preference to a JSON file in userData so the
main process honors it on next launch without waiting for renderer IPC.
getAutoUpdater() now initializes autoDownload from the persisted value.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: suppress cached update toast when disabled and update IPC types
P2: Cache hydration now gates hasUpdate on autoUpdateEnabled so the
App.tsx toast doesn't fire when automatic updates are disabled.
P3: Updated global.d.ts to include 'available' in getUpdateStatus
status union and 'checking' in checkForUpdate return type.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: preserve dismissed releases, show cached updates in Settings, guard concurrent checks
P2a: Updater fallback now checks STORAGE_KEY_UPDATE_DISMISSED_VERSION
before re-surfacing a release found by electron-updater.
P2b: Cache hydration always sets hasUpdate truthfully so Settings
shows the available update. Toast suppression for disabled auto-update
moved to App.tsx (reads localStorage directly).
P3: Re-enable IPC handler checks _isChecking before scheduling
startAutoCheck to prevent concurrent electron-updater calls.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: use localStorageAdapter for lint compliance, skip IPC on initial mount
P1: Replace direct localStorage access with localStorageAdapter in
App.tsx toast guard to fix no-restricted-globals lint error.
P2: Skip setAutoUpdate IPC on initial mount to prevent overwriting
the main-process preference file when renderer localStorage has been
cleared (where the default would be true).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: hydrate auto-update state from main-process preference on mount
Add getAutoUpdate IPC handler so the renderer can query the persisted
preference from auto-update-pref.json. On mount, useSettingsState
reconciles localStorage with the main-process truth, preventing the
toggle from showing 'enabled' when the user had previously disabled
it and localStorage was cleared.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previously, Netcatty checked if the OpenSSH Authentication Agent Windows
service was running via `sc query ssh-agent`. This broke compatibility
with third-party SSH agents (Bitwarden, 1Password, gpg-agent) that
provide the same named pipe without running the system service.
Now we probe `\\.\pipe\openssh-ssh-agent` directly with fs.statSync,
which works regardless of which agent provides the pipe.
Fixes#343
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Security: API keys no longer transit IPC as plaintext; renderer sends
providerId and main process decrypts via safeStorage. Add ReDoS
protection for user-supplied blocklist regex. Sanitize error messages
to strip file paths and sensitive URLs before displaying in chat.
- Bug fixes: approval timeout now notifies user in chat instead of
silently aborting. statusText cleared consistently across all 7 code
paths. useAIState persistence race condition fixed with mountedRef
guard, storage sync validated with type checks.
- Accessibility: InlineApprovalCard uses role="alertdialog" with focus
on approve button. ChatInput menus have proper ARIA roles, labels,
and aria-expanded. ThinkingBlock toggle has aria-expanded/controls.
Model selector submenu supports keyboard navigation.
- UX: switching agents preserves old session and creates new one.
Approval card buttons disabled immediately after click.
- Performance: text-delta streaming batched via requestAnimationFrame.
- Refactor: extract useAIChatStreaming, useToolApproval, and
useConversationExport hooks from AIChatSidePanel (1514 → 751 lines).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
All external agents now use ACP protocol exclusively. The Claude Agent
SDK flow was fully implemented but never wired into the UI.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Security:
- Apply checkCommandSafety() in aiExec IPC handler to enforce command blocklist
- Add critical path validation in handleSftpRemove (normalize + blocklist)
- Change rm -rf to rm -r so permission errors surface
Stream lifecycle:
- Abort all active streams on component unmount (abortControllersRef cleanup)
- Wrap stream reader in try/finally with releaseLock() to prevent leaks
- Use refs for inputValue/images in handleSend to stabilize callback identity
State persistence:
- Clear debounce timer before synchronous persist in destructive operations
(clearSessionMessages, deleteSession, deleteSessionsByTarget)
- Add 10MB max buffer guard in ACP client and MCP server NDJSON parsing
i18n:
- Replace hardcoded English strings with t() calls in InlineApprovalCard,
PermissionDialog, ConversationExport, ThinkingBlock, AgentSelector
- Add 23 new i18n keys to en.ts and zh-CN.ts
Misc:
- Remove debug console.log statements in mcpServerBridge
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Show a clickable chip next to the model selector in ChatInput footer
that lets users quickly toggle between Observer/Confirm/Auto permission
modes. Only visible when Catty Agent is selected (ACP agents handle
their own tool approval flow).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the full-screen PermissionDialog modal with inline InlineApprovalCard
rendered within chat messages. Implement the multi-turn approval flow so that
clicking Approve actually resumes the agent loop via a new streamText call with
proper SDK tool-approval-response messages.
Fix toolCallId extraction (toolCall.toolCallId, not chunk.toolCallId) and use
correct SDK field name (input, not args) for ToolCallPart content parts.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Use Vercel AI SDK's native `needsApproval` on write tools (terminal_execute,
terminal_send_input, sftp_write_file, multi_host_execute) when permission
mode is "confirm". When the SDK emits a `tool-approval-request` stream event,
show the existing PermissionDialog component for user to approve/reject.
- tools.ts: replace manual checkToolPermission() calls with `needsApproval`
property on write tools; keep blocklist checks in execute()
- AIChatSidePanel: handle `tool-approval-request` chunk type, show
PermissionDialog via Promise-based pause, resolve on user action
- Add i18n key for tool denied message
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The StorageEvent handler was missing cases for STORAGE_KEY_AI_HOST_PERMISSIONS
and STORAGE_KEY_AI_AGENT_MODEL_MAP, so changes made in the settings window
were not picked up by the main window.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Safety enforcement:
- Command Timeout: use user setting in MCP Server (execViaPty/execViaChannel)
and Catty Agent path (aiBridge execViaPtyForCatty/execViaChannel fallback)
- Max Iterations: read user setting for Catty (stepCountIs) and ACP (via IPC)
- Permission Mode: observer mode hard-blocks write ops in MCP Server dispatch;
Catty SDK tools wired to checkToolPermission(); all synced via IPC on change
Command Blocklist UI:
- Add editable regex pattern list in Settings AI Safety section
- Reset to defaults button, add/remove patterns
- IPC sync to MCP Server on change and on mount
Provider UX:
- ModelSelector rewritten as combobox: type-to-filter with suggestions dropdown
- All providers use unified ModelSelector (modelsEndpoint optional)
- API key passed as auth header for model fetching (Bearer / x-api-key)
- Skip model fetch when no API key (except Ollama)
- Provider toggle is now mutually exclusive (activating one disables others)
- New providers default to disabled (switch off)
- Custom provider supports editable display name
- No-provider error: friendly message shown in chat
i18n:
- Full i18n coverage for Settings AI tab (~55 keys)
- Full i18n for AI chat panel (placeholder, empty state, time, sessions)
- en.ts and zh-CN.ts locale files updated
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Replace single selectedAgentModel state with per-agent agentModelMap
persisted to localStorage, so each agent remembers its last selected model
- Default to first preset when no prior selection exists
- Fix Claude model presets: use 'default' instead of 'opus' to match
what claude-code-acp actually exposes (default=Opus 4.6)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Restore currentAgentId from active session when switching scopes
- Add i18n for "Agent Settings" label in agent selector
- Add agent icons to settings page default agent dropdown
- Replace catty.svg with new icon, use Settings icon for manage button
- Fix lint: missing useCallback deps, no-restricted-globals, useMemo deps
- Guard webContents.send against disposed render frames in windowManager
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix workspace host discovery: use `activeWorkspace.root` instead of
non-existent `.tree` property, which caused `collectSessionIds` to
always return empty for workspaces
- Isolate AI chat state per-scope (tab/workspace): activeSessionId,
isStreaming, abortController, and inputValue are now keyed by
`${scopeType}:${scopeTargetId}` so different tabs don't interfere
- Add per-chatSession MCP scope metadata to prevent host list mixing
across workspaces, with chatSessionId passed through the full IPC chain
- Store hostname/username in SSH session objects as fallback for MCP
host info when renderer metadata is unavailable
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AI Settings:
- Remove External Agents section, add dedicated Codex/Claude Code sections
with auto-detection and manual path override
- Add OpenRouter model list auto-fetch with searchable dropdown
- Remove standalone Default Model section, integrate into provider cards
- Provider toggle now acts as mutual-exclusive active selector
- Add cross-window state sync via storage events
- Add ErrorBoundary around AI tab for graceful error handling
Catty Agent fixes:
- Fix streaming: use .chat() instead of default Responses API, use
getReader() pattern instead of for-await, handle text-delta/text chunk
types correctly per SDK v6
- Fix API key: decrypt encrypted key before passing to SDK
- Fix terminal_execute: use PTY stream (visible in terminal) with MCP
markers instead of invisible SSH exec channel
- Fix multi-turn: only pass user/assistant text history, skip tool
messages to avoid SDK schema validation errors
- Fix tool result display: create new assistant message after tool
results so follow-up text renders correctly
Other:
- Add netcatty:ai:resolve-cli IPC handler for CLI path validation
- Remove Gemini from agent discovery (only Codex + Claude Code)
- Fix lint errors across ChatInput, TerminalLayer, SettingsAITab
- Strip MCP markers from tool execution stdout
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add image paste/drop support with base64 encoding and chat display
- Add + button popover (Files, Image, Mention Host) with @ auto-complete
- Add model selector with Codex thinking level sub-menus (GPT 5.4, Codex 5.x, o3/o4-mini)
- Switch Claude Code to ACP protocol; remove CLAUDE_CONFIG_DIR isolation
- Filter PTY exec markers in preload data pipe (precise regex, preserves command echo)
- Increase PTY exec timeout to 5min with Ctrl+C cancellation on abort
- Fix tool call loading spinner (only animate during active streaming)
- Reset active session on terminal/workspace switch
- Add AI sparkle button in top bar to toggle AI panel
- Display user-attached images in chat message list
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Create netcatty-remote-hosts MCP server (TCP bridge + stdio child process)
exposing terminal_execute, sftp_*, get_environment tools to ACP agents
- Execute commands via PTY stream with self-erasing markers for terminal
visibility; disable pagers automatically; 60s timeout fallback
- Inject MCP server into both Codex (ACP) and Claude Code (ACP) sessions
- Add model selector popover with hardcoded presets for Claude (Opus/Sonnet/
Haiku) and Codex (GPT 5.4, Codex 5.3/5.2/5.1, o3/o4-mini) with thinking
level sub-menus
- Fix multi-step tool call message flow (mutable flag instead of stale closure)
- Remove CLAUDE_CONFIG_DIR isolation to fix Claude auth; switch Claude to ACP
- Add startup cleanup for orphaned AI sessions
- Guard collectSessionIds against undefined workspace tree
- Remove permission mode chip from chat input
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Switching agent deactivates current session, next message creates a new one
- Filter history sessions by both scopeType and targetId
- Restore agent selector when resuming a historical session
- Auto-cleanup AI sessions when terminal/workspace instances are destroyed
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>