Commit Graph
416 Commits
Author SHA1 Message Date
bincxzandClaude Opus 4.6 df11beff8c fix: clear mainWindow reference on window destroy (#587)
The mainWindow variable was never cleared when the window was destroyed,
unlike settingsWindow which had a proper 'closed' handler. This caused
getMainWindow() to return a destroyed window object, preventing the
activate handler from correctly detecting the main window was gone and
creating a new one.

Fixes #587

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 19:16:59 +08:00
bincxzandClaude Opus 4.6 f1ce541885 fix: dock click opens main window instead of settings window (#587)
On macOS, when the main window is closed but the settings window is
still open, clicking the Dock icon would focus the settings window
instead of re-creating the main window.

- focusMainWindow() now explicitly finds the main window via
  getWindowManager() instead of using getAllWindows()[0]
- activate handler creates a new main window even when other
  windows (settings) are still open

Fixes #587

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 19:05:13 +08:00
bincxzandClaude Opus 4.6 07e003fe43 fix: distinguish settings window title from main window
Set the settings window title to "netcatty Settings" and prevent
the HTML <title> tag from overriding it, so macOS Dock menu and
Window menu can distinguish between the two windows.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 19:02:36 +08:00
bincxzandClaude Opus 4.6 b3d37d134a fix: honor keepaliveInterval=0 as disabled instead of falling back to 10s
When keepaliveInterval was set to 0 (the default, documented as "disabled"),
the code treated 0 as falsy and fell back to 10000ms. This caused ssh2 to
send keepalive@openssh.com global requests every 10s. Devices with non-OpenSSH
SSH implementations (e.g. NOKIA/ALCATEL) that don't reply to these requests
would have their connections terminated after ~40s (4 × 10s keepalive timeout).

Closes #581

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 10:27:08 +08:00
bincxzandClaude Opus 4.6 a9e561ee51 feat: show "Waiting for remote..." during ZMODEM upload finalization
After all file data is written to the buffer, the progress bar shows
100% but the remote rz is still processing. Now a "finalizing" flag
is sent with the last progress event, and the UI displays "Waiting
for remote..." instead of the misleading 100% uploading state.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:43:26 +08:00
bincxzandClaude Opus 4.6 e808b1709e fix: increase ZMODEM handshake timeout from 10s to 120s
10s was too short for large files (466MB+). After sending all data,
the remote rz still needs time to read from TCP buffer and write to
disk before it can reply with ZRINIT/ZFIN. 120s accommodates slow
links and large files while still catching genuinely dead sessions.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:38:45 +08:00
bincxzandClaude Opus 4.6 d75b58e4d8 fix: timeout on ZMODEM handshake rejects instead of resolving
withTimeout was resolving silently after 10s, which made a stalled
xfer.end()/zsession.close() look like a successful transfer. Now it
rejects with "ZMODEM handshake timeout", so the .catch handler fires
and shows an error toast instead of a false success.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:28:57 +08:00
bincxzandClaude Opus 4.6 e2430cdcab fix: cancel sentry on all session cleanup paths + upload timeout guard
- terminalBridge: cancel zmodemSentry in telnet error/close, serial
  error/close, and cleanupAllSessions before deleting sessions
- sshBridge: cancel zmodemSentry in all 4 SSH cleanup paths (stream
  close, conn error, conn timeout, conn close)
- zmodemHelper: wrap xfer.end() and zsession.close() with 10s timeout
  to prevent indefinite hang when cancel/abort leaves internal
  zmodem.js Promises unresolved (prevents fd leak)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:20:07 +08:00
bincxzandClaude Opus 4.6 8e6ac8de10 revert: remove ZACK ignore handler (caused by SOCKS5 proxy, not protocol)
The "Unhandled header: ZACK" was triggered by a SOCKS5 proxy on the
server causing abnormal protocol behavior, not a real lrzsz issue.
The handler's condition was too broad (any active send) and could
mask genuine protocol errors. Keep ZRINIT and ZRPOS handlers which
have narrow conditions and address real scenarios.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:11:03 +08:00
bincxzandClaude Opus 4.6 5495877e5a fix: ignore stray ZACK headers during ZMODEM upload
zmodem.js only handles ZACK in specific Send session states (after
ZSINIT, during file negotiation). Some receivers send extra ZACKs as
generic acknowledgements that arrive outside these states, causing
"Unhandled header: ZACK". Since ZACK is just an ack, ignoring it
is safe and keeps the transfer going.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:05:15 +08:00
bincxzandClaude Opus 4.6 5078b3776e fix: use setImmediate instead of setTimeout(50) for drain wait
setTimeout(50) per chunk would cap upload speed at ~1.28MB/s because
ssh2's 32KB highWaterMark triggers backpressure on almost every 64KB
write. setImmediate yields to the I/O phase without a fixed delay,
letting TCP flush as fast as possible.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-31 00:03:11 +08:00
bincxzandClaude Opus 4.6 f5d6b8b4d8 fix: add backpressure handling to ZMODEM upload loop
Large file uploads (466MB+) could saturate the SSH/PTY write buffer
with all data sent synchronously, causing the ZEOF/ZFIN handshake
at the end to be delayed — the UI shows 100% but the transfer hangs
while TCP flushes the backlog.

- All writeToRemote callbacks now return stream.write() result
- Sentry sender tracks _needsDrain flag when write returns false
- Upload loop calls waitForDrain() which yields 50ms when backpressure
  is detected, letting TCP flush buffered writes between chunks

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 23:58:42 +08:00
bincxzandClaude Opus 4.6 1c560dbc16 fix: reject CLI paths that fail --version probe
In both discover and resolve-cli handlers, treat --version failure
(exception or empty output) as an invalid CLI. This catches .app
bundles, broken symlinks, and other non-executable paths that pass
the filesystem check but aren't actually usable CLI tools.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 23:48:15 +08:00
bincxzandClaude Opus 4.6 4b8b0ed74c fix: reject .app directories in CLI path normalization
normalizeCliPathForPlatform used existsSync which returns true for
directories like /Applications/Codex.app. Added statSync.isFile()
check on non-Windows platforms so .app bundles are not mistaken for
CLI executables.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 23:45:58 +08:00
陈大猫andClaude Opus 4.6 308d825db7 feat: ZMODEM (lrzsz) file transfer support (#579)
* feat: add ZMODEM (lrzsz) file transfer support for terminal sessions

Adds ZMODEM protocol detection and file transfer capability to all
terminal session types (Local, SSH, Telnet, Mosh, Serial). Uses
zmodem.js library with main-process sentry pattern to intercept
binary data before string decoding, avoiding IPC pipeline changes.

- zmodemHelper.cjs: shared ZMODEM sentry with Electron dialog integration
- terminalBridge.cjs: encoding:null for PTY + sentry wrappers for all session types
- sshBridge.cjs: sentry wrapper for SSH stream data
- preload.cjs + global.d.ts: ZMODEM event IPC bridge and TypeScript types
- useZmodemTransfer.ts: React hook for ZMODEM transfer state

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: preserve charset decoding and add ZMODEM progress UI

- zmodemHelper: pass raw Buffer to onData, let callers handle decoding
- terminalBridge: use StringDecoder for telnet/serial, UTF-8 for local/mosh
- sshBridge: restore iconv decoder for SSH session charset support
- ZmodemProgressIndicator: floating progress bar with cancel button
- Terminal.tsx: wire useZmodemTransfer hook + toast notifications

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: ZMODEM listener cleanup, stream leak, and toast dedup

- preload: clean up zmodemListeners on session exit (memory leak)
- zmodemHelper: add ws.on('error') handler to close write stream on failure
- Terminal: use ref guard to prevent duplicate toast notifications

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: address code review findings for ZMODEM

- cancel/consume error now send IPC event to renderer (prevents stuck UI)
- sanitize download filename with path.basename (path traversal prevention)
- add on_detect concurrency guard (deny if transfer already active)
- formatBytes: handle negative, zero, and TB+ values safely
- closeSession: cancel active ZMODEM before destroying transport

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: prevent double-notification on cancel and stream error resilience

- Guard .then()/.catch() in promise chain: skip if cancel() already handled
- Download: add writeAborted flag to stop on_input after stream error
- Upload: pre-compute file stats to avoid O(N²) statSync calls

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use zsession.abort() instead of close() on dialog cancel

close() is only available on Send sessions. Calling it on a Receive
session throws, leaving the sentry's internal _zsession dangling and
causing subsequent terminal data to be consumed by the abandoned
ZMODEM session (terminal freeze). abort() is defined on the base
ZmodemSession class and properly fires session_end to reset the sentry.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: handle ZFIN/OO mismatch as successful transfer

When sz exits over SSH, the shell prompt often arrives before the
ZMODEM "OO" end marker, causing zmodem.js to throw a protocol error.
Since ZFIN was already exchanged (= all file data transferred), treat
this specific error as a successful completion and forward the shell
prompt data back to the terminal via sentry re-consume.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — UTF-8 decoder, ZFIN abort, session exit cleanup

- terminalBridge: use StringDecoder for local/mosh PTY to handle
  multi-byte UTF-8 split across buffer boundaries (prevents garbled
  CJK/emoji output)
- zmodemHelper: on ZFIN/OO success path, use _on_session_end() instead
  of abort() to avoid sending CAN (Ctrl-X) bytes to the remote shell
- useZmodemTransfer: listen to onSessionExit to reset state when the
  session dies mid-transfer (prevents stuck progress indicator)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — file collision handling and stream flush

- Download: auto-rename with (1), (2), etc. if file already exists
  in the target directory, preventing silent overwrite
- Download: wait for all write streams to finish flushing before
  resolving the session_end promise, ensuring data is on disk when
  the UI reports completion

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — Windows PTY string compat and Telnet binary safety

- Local/Mosh PTY: handle string data from Windows node-pty which
  ignores encoding: null; convert to Buffer before sentry.consume()
- Telnet: bypass IAC negotiation during active ZMODEM transfer to
  preserve 0xFF bytes in binary data
- Telnet writeToRemote: escape 0xFF as 0xFF 0xFF per Telnet spec
  so ZMODEM binary data is not treated as IAC commands

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — Windows PTY guard, Telnet IAC, stream cleanup

- Local/Mosh: skip ZMODEM sentry on Windows where node-pty can't
  provide raw bytes; fall back to original string pipeline
- Telnet: always run IAC negotiation (even during ZMODEM) since the
  Telnet layer still escapes 0xFF as IAC IAC; the existing handler
  already correctly collapses IAC IAC → single 0xFF
- Download: destroy un-ended write streams on session_end to prevent
  hanging promises and leaked file descriptors on abort

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — early session start, progress throttle, no dup start

- Download: call zsession.start() before showing folder picker dialog
  so lrzsz doesn't time out waiting for ZRINIT
- Download: throttle progress IPC to ~10 updates/sec (100ms interval)
  to avoid overwhelming renderer on fast links
- Download: remove duplicate zsession.start() at bottom of Promise

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: handle ZRPOS and prevent terminal flood after ZMODEM abort

- Add 500ms cooldown after ZMODEM abort: suppress residual protocol
  bytes from remote rz/sz that would otherwise flood the terminal
- Send 8x CAN (Ctrl-X) on abort/cancel/error to force remote end to
  stop transmitting even if the initial abort sequence was lost
- Handles "Unhandled header: ZRPOS" gracefully (zmodem.js doesn't
  support error recovery, so abort is the correct response)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: send Ctrl+C after abort in all cancel/error paths

Debian's rz stays attached to the TTY after receiving CAN sequences.
The cancel() path already sent Ctrl+C via scheduleRemoteInterruptAfterCancel,
but dialog-cancel and consume-error paths did not. Now all three abort
paths (dialog cancel, consume error, explicit cancel) send Ctrl+C after
150ms to ensure the remote rz/sz process exits and the shell regains control.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: add interruptRemote for SSH ZMODEM sentry

Pass SSH stream.signal("INT") as interruptRemote callback so the
ZMODEM helper can send SIGINT to the remote process when cancelling
transfers, complementing the Ctrl+C byte sent via writeToRemote.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: dialog-cancel abort uses module-level helper to avoid ReferenceError

sendExtraAbortBytes and writeToRemote are closure-scoped inside
createZmodemSentry, not accessible from handleUpload/handleDownload.
Extract abortRemoteProcess as a module-level function that takes
writeToRemote as a parameter, used in both dialog-cancel paths.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: dialog cancel throws instead of returning to avoid false complete

When user dismisses the file/folder picker, handleUpload/handleDownload
now throw "Transfer cancelled" instead of returning normally. This
ensures the .catch() handler fires (sending error event) rather than
.then() (which would incorrectly send complete event).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — preserve transferType in progress events

- useZmodemTransfer: copy transferType from progress events so the
  transfer direction is preserved if renderer re-subscribes after
  the initial detect event was missed
- zmodemHelper: clean up upload loop comments (backpressure handled
  via 64KB chunks + setImmediate yield per iteration)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: codex review — guard stale session cleanup, delete partial downloads

- Promise chain .then/.catch/.finally now compare currentZSession
  identity (=== zsession) instead of truthiness, preventing a new
  transfer from being clobbered by the old promise settling
- Aborted/incomplete downloads are deleted from disk on session_end
  so users don't end up with corrupt partial files

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: unconditional cooldown suppression after ZMODEM abort

The previous cooldown checked if data "looks like residual ZMODEM"
which fails for sz's file content (arbitrary printable bytes). Now
cooldown unconditionally drops ALL incoming data for 2 seconds after
abort, with repeated CAN bursts to ensure the remote sz stops. This
prevents the terminal flood seen when cancelling large sz downloads
on fast connections.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 23:39:35 +08:00
bincxzandClaude Opus 4.6 3844edd49f fix: clean up copilot temp dir even when provider init fails
Move COPILOT_HOME temp dir cleanup before the acpProviders entry
check so it runs even if provider creation failed before the entry
was stored in the map.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:57:00 +08:00
bincxzandClaude Opus 4.6 8f97a7e81d fix: use resolved path as copilot acpCommand and add Windows home fallback
- When building managed copilot agent config, set acpCommand to the
  resolved path instead of bare "copilot" so custom paths work for
  ACP launches
- Add USERPROFILE fallback in prepareCopilotHome for Windows where
  HOME may not be set

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:48:07 +08:00
bincxzandClaude Opus 4.6 5daf1f0d6f fix: hoist copilotConfigInfo above try block to fix ReferenceError
copilotConfigInfo was declared with let inside the try block but
referenced in the finally block for temp dir cleanup. Block scoping
caused a ReferenceError that broke list-models for Copilot agents.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:38:39 +08:00
bincxzandClaude Opus 4.6 b1a5b92ce4 fix: clean up transient copilot temp dirs and remove verbose MCP logs
- Add COPILOT_HOME cleanup in list-models finally block to prevent
  temp directory accumulation on each model fetch
- Remove verbose console.log in mcpServerBridge dispatch/connect/auth
  that fired on every MCP call for all agents

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:27:18 +08:00
bincxzandClaude Opus 4.6 c99a70831a fix: address review issues in copilot agent integration
- Fix matchesManagedAgentConfig acpCommand matching for copilot by
  using a lookup table instead of hardcoded ternary
- Remove dead nodeRuntimePath variable and unused 4th arg to
  buildMcpServerConfig
- Fix model loading useEffect double-triggering by reading
  agentModelMap via ref instead of dependency
- Add temp COPILOT_HOME cleanup in cleanupAcpProvider
- Remove dead acpForceProviderReset Set (never populated after
  stop/resume refactor)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:22:59 +08:00
bincxzandClaude Opus 4.6 4b0468b0d2 merge: resolve conflicts with main for copilot agent support
Adapt copilot agent additions to the refactored managed agent
architecture (resolveAgentPath + buildManagedAgentState pattern).
Add copilot to ManagedAgentKey type and MANAGED_AGENT_META.
Keep main's resolveMcpServerRuntimeCommand (process.execPath +
ELECTRON_RUN_AS_NODE) over PR's runtimeCommand parameter approach.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 17:14:45 +08:00
Eric Chan a525c073b9 fix: matchesAgentCommand update for windows shim 2026-03-30 16:29:14 +08:00
bincxzandClaude Opus 4.6 afceb92a55 fix: fall back to PATH search when stored CLI path is stale
When a previously stored custom path no longer exists (e.g. CLI
reinstalled to a different location), aiResolveCli now falls back
to PATH-based detection instead of returning unavailable.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-30 16:27:32 +08:00
Eric Chan d9b51c3a50 feat: add GitHub Copilot CLI agent support 2026-03-30 15:53:08 +08:00
bincxz 53cb160a6e fix codex agent path detection and MCP startup 2026-03-30 15:04:06 +08:00
bincxzandClaude Opus 4.6 0826bbb435 style: use Netcatty logo in OAuth callback page
Replace the generic terminal SVG icon with the actual Netcatty brand
logo (blue rounded-rect with terminal + cat tail motif).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-29 13:49:25 +08:00
bincxzandClaude Opus 4.6 4dd7640452 fix: allow auto encoding through same-host fast path
The encoding guard was rejecting "auto" which is the default encoding
for nearly all connections, making same-host optimization never trigger.

Frontend now allows "auto" through. Backend resolves "auto" to the
actual session encoding via resolveEncodingForRequest and only proceeds
with exec cp when the resolved encoding is UTF-8.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-29 13:25:36 +08:00
陈大猫andClaude Opus 4.6 0b08521e63 perf: optimize same-host SFTP transfer with remote cp command (#564)
* perf: optimize same-host SFTP transfer with remote cp command

When both panels are connected to the same remote host, use SSH exec
`cp -a` instead of downloading to local temp then re-uploading. This
eliminates 2x bandwidth usage and reduces latency for same-host transfers.

Closes #561

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* perf: optimize same-host directory transfer with single cp -ra command

For same-host directory transfers, use a single `cp -ra` command via SSH
exec instead of recursively walking the directory and copying files one
by one. This makes directory copies nearly instant on the remote server.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use endpoint cache key for same-host detection and guard non-UTF-8 paths

Address two code review issues:

1. Compare per-connection cache keys (hostname+port+protocol+sudo+username)
   instead of just hostId for same-host detection. This prevents false
   positives when the same hostId has different session-time overrides.

2. Restrict exec-based cp paths to UTF-8 compatible encodings only.
   Non-UTF-8 encodings (e.g. gb18030) need encodePathForSession which
   shell exec cannot use — fall back to download+upload for those cases.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: directory cp semantics, cancellation, and auto encoding guard

1. Use `cp -ra source/. target/` instead of `cp -ra source target` to
   copy directory contents into target, preserving merge semantics when
   the target directory already exists (avoids extra nesting level).

2. Check cancellation state before and after sameHostCopyDirectory call
   so cancelled transfers don't finalize as completed.

3. Exclude 'auto' from exec-safe encodings since auto can resolve to
   non-UTF-8 (e.g. gb18030) at the session level.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: wire cancellation into same-host copy paths

1. Single file cp -a: check transfer.cancelled before and after
   execSshCommand so cancelled transfers don't proceed as success.

2. Directory cp -ra: accept transferId, register in activeTransfers
   so cancelTransfer can flag it, and check cancelled state at each
   async boundary. Cleanup via finally block.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: abort remote cp process on transfer cancellation

Add execSshCommandCancellable() that wires the SSH exec stream into
transfer.abort, so cancelTransfer can close the stream and kill the
remote cp process immediately instead of waiting for it to finish.

Used in both single-file (cp -a) and directory (cp -ra) same-host paths.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: close exec stream immediately if cancelled before callback fires

Check transfer.cancelled at the start of the exec callback and close
the stream right away, preventing the remote cp from running when
cancellation happened between the exec() call and callback delivery.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fallback to download+upload when remote cp is unavailable

On non-POSIX remotes (e.g. Windows SSH servers) where cp is absent,
same-host optimization now gracefully falls back to the existing
download+upload transfer path instead of failing the transfer.

- Single file: try cp -a first, fall back to temp file on non-zero exit
- Directory: sameHostCopyDirectory returns { success: false } instead of
  throwing, frontend falls back to recursive transferDirectory

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* perf: cache cp unavailability to avoid repeated exec failures

Track sftpIds where remote cp failed in cpUnavailableSet so subsequent
file transfers in the same session skip the exec attempt and go directly
to download+upload, avoiding per-file exec round-trip overhead on
non-POSIX remotes.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: skip transferFile for directories already handled by same-host copy

Add !task.isDirectory guard to the else branch so successful
sameHostCopyDirectory doesn't also trigger a redundant transferFile
call that would duplicate data.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: dereference symlinks in same-host copy to match SFTP behavior

Use cp -aL instead of cp -a so symlinks are dereferenced (copied as
file contents), matching the existing SFTP download+upload flow which
always transfers resolved file data.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* revert: remove -L flag from same-host cp to avoid recursing symlinked dirs

Revert cp -aL back to cp -a. The -L flag dereferences all symlinks
including symlinked directories, which can unexpectedly recurse into
large unrelated directory trees. Using cp -a preserves symlinks as-is,
which is safer and consistent with how the transfer UI treats symlink
directories as non-recursive entries.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: refine cp unavailability caching and remove dead import

1. Only cache sftpId in cpUnavailableSet on exit code 127 (command not
   found). Other failures (permission denied, disk full) are transient
   or path-specific and should not disable cp for the entire session.

2. Check cpUnavailableSet at the top of sameHostCopyDirectory to skip
   exec attempt on known non-POSIX remotes. Also cache 127 exits from
   directory copies.

3. Remove unused execSshCommand import from transferBridge (replaced by
   local execSshCommandCancellable) and revert its export from sftpBridge.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-29 13:21:58 +08:00
陈大猫 6a37b8bbc6 fix: use system browser for OAuth flows (#563) (#565) 2026-03-29 12:43:21 +08:00
Eric Chan 6d19413025 Display approved AI commands in terminal sessions before their output. (#546) 2026-03-27 19:59:59 +08:00
陈大猫andClaude Opus 4.6 c3224d30c6 feat: network device mode for SSH + serial charset encoding support (#540)
* feat: add deviceType field to Host model for network device support

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: pass deviceType through session metadata pipeline

Thread deviceType from Host model through AITerminalSessionInfo, IPC
types, and mcpServerBridge so AI agents can inspect device type per session.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat: route network device SSH sessions to raw PTY execution

When deviceType === 'network', handleExec now uses execViaRawPty
instead of execViaPty so vendor CLIs (Huawei VRP, Cisco IOS, etc.)
receive commands as-is without POSIX shell wrapping or markers.
The command blocklist is also skipped for network devices, consistent
with the existing serial session bypass. AI context description updated
to document the raw-execution behaviour for network device sessions.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat: add network device mode toggle to host settings UI

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: add network device awareness to Catty Agent system prompt

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: extend network device mode to Catty Agent exec path and host context

- Add network device detection and raw execution routing to aiBridge.cjs
  (the primary Catty Agent command path), not just the MCP bridge
- Export getSessionMeta from mcpServerBridge for reuse in aiBridge
- Surface deviceType in Catty Agent system prompt host list so the AI
  can identify which sessions are network devices
- Pass deviceType through buildSystemPrompt context

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: exempt network device sessions from client-side blocklist and update ACP context

- Add deviceType to ExecutorContext sessions type
- Skip renderer-side command blocklist for deviceType=network sessions
  in shared toolExecutors.ts (not just main-process side)
- Update ACP agent context hint to mention network device sessions

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: only show network device mode toggle for SSH hosts

Telnet and local hosts don't support the network device execution path,
so hiding the toggle prevents users from enabling a broken configuration.
Serial hosts already use raw mode by default.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: exclude Mosh sessions from network device raw execution path

Mosh uses a shell-backed PTY and cannot connect to vendor CLIs, so
network device mode should only apply to SSH and serial sessions.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: prefer session.protocol over metadata for Mosh detection

Mosh tabs report protocol:"ssh" in renderer metadata but "mosh" in
the main-process session object. Prioritize session.protocol (runtime
truth) to correctly exclude Mosh from network device raw execution.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: suppress deviceType metadata for Mosh sessions

Mosh requires a shell-backed PTY and cannot connect to vendor CLIs,
so omit deviceType from AI-facing metadata when session is Mosh-backed.
This prevents the AI from being told to use vendor CLI syntax when the
actual execution path uses normal shell wrapping.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use exit code 0 for network device sessions and hide toggle for Mosh

- Network device / serial sessions return exitCode: null from vendor
  CLIs. Default to 0 instead of -1 so the AI doesn't misinterpret
  successful commands as failures.
- Hide the network device mode toggle when Mosh is enabled, since
  the setting is suppressed at runtime for Mosh sessions anyway.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: preserve null exit codes and restrict raw mode to SSH/serial

- Preserve exitCode: null for network device sessions instead of
  coercing to 0, so the AI knows exit status is unavailable rather
  than seeing a misleading success code.
- Explicitly whitelist SSH/serial protocols for network device mode
  instead of just excluding mosh, preventing local/telnet sessions
  from accidentally entering raw execution.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use UTF-8 encoding for SSH network device raw execution

execViaRawPty hardcodes latin1 for serial port data decoding. Add an
encoding option (default: latin1) and pass utf8 from SSH network
device call sites so multi-byte characters aren't corrupted.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use host charset for serial port decoding instead of hardcoded latin1

- Extract charsetToNodeEncoding() to module scope in terminalBridge
- Serial sessions now read options.charset (from Host.charset) for
  both terminal display decoding and AI command output
- Store serialEncoding on session object so exec paths can use it
- Pass encoding through all execViaRawPty call sites
- Default encoding changed from latin1 to utf8 (matches most modern
  network equipment and is the safer default for CJK environments)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: move serialEncoding declaration before session object creation

serialEncoding was referenced in the session object literal before its
const declaration, causing a TDZ ReferenceError that would crash every
serial connection.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: tighten isNetworkDevice logic and clean up edge cases

- Align toolExecutors isNetworkDevice check with bridge logic: require
  explicit SSH/serial protocol match instead of trusting deviceType alone
- Remove empty-string protocol match from isSshOrSerial in both bridges
  to prevent local/unknown sessions from being treated as network devices
- Widen exitCode return type to `number | null` to match actual behavior
- Clear deviceType when enabling Mosh (incompatible combination)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: update MCP server tool descriptions for network device sessions

The get_environment and terminal_execute tool descriptions only
mentioned serial/raw sessions for network devices. Updated to also
reference deviceType: network SSH sessions so external AI agents
(Claude, Codex) know about the new execution mode.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: include deviceType in get_environment and guard execViaChannel fallback

- Add deviceType to executeWorkspaceGetInfo session mapping and return
  type so Catty Agent's get_environment tool matches MCP bridge output
- Guard both aiBridge and mcpServerBridge against falling through to
  execViaChannel for network device sessions — network devices require
  an interactive PTY and exec channels would produce broken behavior

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: add charset setting to serial host configuration UI

Serial hosts now have a charset input in the Advanced section,
defaulting to UTF-8. The value is saved to Host.charset and used
by the serial decoder in terminalBridge.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: add charset to serial quick-connect modal with full pipeline

- Add charset input to SerialConnectModal (Advanced section)
- Thread charset through onConnect callback → handleConnectSerial →
  createSerialSession → TerminalSession.charset
- Add charset field to TerminalSession interface
- Include charset in fallback host builder for quick-connect sessions
  so createTerminalSessionStarters can pass it to startSerialSession
- Saved hosts also store charset via onSaveHost

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: constrain serial connect modal height with scrollable content

Modal content could overflow the viewport when Advanced section was
expanded. Add max-h-[85vh] to DialogContent with flex layout so the
content area scrolls while header and footer buttons stay visible.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: propagate charset through all serial session creation paths

- Add charset to startSerialSession type in global.d.ts
- Copy host.charset to TerminalSession in connectToHost serial path
- Copy host.charset in createWorkspaceWithHosts serial path
- Propagate session.charset in splitSession (both workspace and standalone)
- Propagate session.charset in copySession

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: propagate charset in remaining session creation paths

Add host.charset to connectToHost (non-serial), createWorkspaceWithHosts
(non-serial), and runSnippet session creation for consistency.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-27 18:33:16 +08:00
bincxz 91fd44cccf fix terminal autocomplete path and popup behavior 2026-03-27 01:22:35 +08:00
陈大猫andClaude Opus 4.6 c924259fc0 fix: add local autocomplete specs and isolate command history per host (#536)
Add local spec files for commands missing from @withfig/autocomplete
(journalctl, yum, awk) and load them with priority over the upstream
package. Also enforce strict per-host isolation for command history —
previously cross-host matching by OS leaked host-specific commands
(e.g. cd /cq/) into unrelated sessions.

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-27 00:04:42 +08:00
bincxz f896f2a071 fix: polish autocomplete popup and bridge 2026-03-26 23:34:10 +08:00
bincxz 1851a8de71 Merge remote-tracking branch 'origin/main' 2026-03-26 23:22:15 +08:00
bincxz 5e05d25c2b fix: tighten autocomplete directory listing 2026-03-26 23:21:31 +08:00
bincxz 2d57015ac5 fix: harden path completion edge cases 2026-03-26 23:13:52 +08:00
bincxz f1fea53af6 fix: avoid preload API collision with sftp 2026-03-26 22:38:44 +08:00
bincxz aabae00970 fix: refine path completion popup behavior 2026-03-26 22:35:48 +08:00
陈大猫andClaude Opus 4.6 de874fc8c5 fix: 修复双击检查更新崩溃 & 优化更新 UX (#522) (#531)
* fix: prevent double-click update crash and improve update UX (#522)

- Add state guards to prevent checkForUpdates during active download
- Disable "Check for Updates" button during checking/downloading/ready
- Make version badge trigger in-app download instead of opening GitHub
- Change error toast action from "Open Releases" to "View in Settings"
- Add "Download Now" button in system settings as primary action
- Keep GitHub release link as secondary fallback in settings

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: reset download state when downloadUpdate() rejects

Clears _isDownloading and broadcasts error status on catch so the
update UI does not get stuck after a failed download attempt.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: only show Download Now after a completed update check

Prevents downloadUpdate() from being called with stale cached state
before electron-updater has run checkForUpdates(), avoiding a
"Please check update first" error.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use correct broadcast function and prime updater before download

- Replace undefined broadcastUpdateStatus with broadcastToAllWindows
- Call checkForUpdate before downloadUpdate to ensure electron-updater
  has populated update metadata

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use correct error payload field and guard unsupported platforms

- Use { error: ... } instead of { message: ... } in download error
  broadcast to match renderer expectations
- Bail out of startDownload when checkForUpdate returns unsupported
  or throws, instead of entering a failing download path

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: guard startDownload against in-flight and no-update check results

Bail out when checkForUpdate returns checking, not-available, or
unsupported states to prevent calling downloadUpdate prematurely.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: remove duplicate error broadcast and fallback to releases on unsupported

- Remove redundant broadcastToAllWindows in download catch (global
  error listener already handles it)
- Open release page instead of silently returning when platform
  does not support auto-update

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: check supported before available to ensure release page fallback

Unsupported platforms return { available: false, supported: false },
so the supported check must come first to open the release page
instead of silently returning.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: skip download when update is already ready or downloading

Guard against re-downloading when checkForUpdate returns ready or
downloading sentinel, preventing overwrite of valid install state.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fallback to release page when electron-updater reports no update

When GitHub API found an update but electron-updater does not,
open the release page instead of silently doing nothing.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 20:59:11 +08:00
bincxzandClaude Opus 4.6 ae797e5fb1 feat: 远程路径补全 — cd/ls/cat 等命令自动列出文件和目录
通过 SSH exec channel 在远程机器上执行 ls 命令获取目录内容,
在补全菜单中显示文件/目录列表。

实现:
- sshBridge.cjs: 新增 netcatty:ssh:listdir IPC handler,
  使用 session.conn.exec() 在独立 channel 执行 ls -1Fap,
  不影响交互式终端
- main.cjs: 新增 netcatty:local:listdir,本地终端用 fs.readdir
- preload.cjs: 暴露 listRemoteDir/listLocalDir API
- remotePathCompleter.ts: 路径补全核心模块
  - shouldDoPathCompletion: 检测 fig spec template/generators、
    PATH_COMMANDS 白名单、或输入以 /  ./  ../  ~/ 开头
  - resolvePathComponents: 解析目录路径和过滤前缀
  - getPathSuggestions: 编排检测→解析→IPC→格式化
  - 5 秒 TTL 缓存 + in-flight 请求去重
- completionEngine.ts: SuggestionSource 新增 "path" 类型,
  CompletionSuggestion 新增 fileType 字段,
  getCompletions 接受 sessionId/protocol/cwd 参数
- AutocompletePopup.tsx: 路径建议显示 📁/📄/🔗 图标
- Terminal.tsx: 传入 protocol 和 getCwd

支持:SSH 远程目录、本地终端、cd 仅显示目录、
  空格文件名转义、head -100 限制输出

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 20:23:45 +08:00
陈大猫andClaude Opus 4.6 9a7d4decff feat: 终端命令自动补全系统 (#527)
* feat: 终端命令自动补全系统

实现类似 WindTerm/Fish 的终端命令自动补全功能,不依赖机器学习:

- 历史命令持久化存储:按主机分组,频率+时间衰减排序,跨会话共享
- 前缀匹配引擎:支持精确前缀匹配和模糊匹配(首字符+连续字符+词边界加权)
- Prompt 检测器:识别 bash/$、zsh/%、fish/> 等常见 prompt 模式,排除 vim/less 等程序
- Ghost Text 插件:xterm.js 自定义 addon,光标后灰色行内建议,→ 接受全部,Ctrl+→ 接受一词
- 弹出补全菜单:浮动列表 UI,↑↓ 导航,Tab/Enter 选中,Esc 关闭,来源标记(h/c/s/o/a)
- @withfig/autocomplete 集成:600+ 命令规范的子命令、选项、参数补全
- 上下文感知:解析命令行 token,根据当前位置提供对应类型的补全
- 用户配置:启用/禁用、Ghost Text、弹出菜单、防抖延迟、最小字符数等
- 快速打字防误触:检测打字速度,快速输入时抑制建议
- 输入防抖 100ms,异步匹配不阻塞 UI

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 补全菜单混合展示历史命令和 spec 子命令

- 输入已知命令名(如 docker)时即使没有空格也预览子命令
- 历史命令条数从 8 降为 5,留空间给 spec 建议
- 修复 wordIndex === 0 时 spec 补全被跳过的问题

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 补全菜单在终端底部时向上展开

当光标在终端下方、空间不足时,弹出菜单向上展开(底边对齐光标行),
避免溢出终端区域。列表顺序和选中逻辑不变——最可能的选项始终在顶部,
用户初始向下选择。参考 Termius 的做法。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 补全菜单跟随终端主题 + Enter 直接执行命令

1. 补全菜单颜色从终端主题动态派生(color-mix),不再硬编码色值,
   确保与任何主题视觉一致
2. 在弹出菜单中按 Enter 选择命令时,直接插入并发送 \r 执行,
   无需用户再按一次回车
3. Tab/鼠标点击仍然只插入不执行(保留选择后编辑的能力)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 修复 PR review 发现的全部 20 个问题

功能修复:
- #1 修复 selectAndExecute 导致命令双重录入历史:用 suppressNextEnterRecordRef
  标志位让 handleInput 的 Enter 分支跳过已经录入过的命令
- #2 修复 Prompt 末尾 $ 误判:重写 findPromptBoundary 为从左到右逐字符扫描,
  排除 $HOME/$PATH 等变量引用(检查 $ 前是否有空格、是否在 token 内部)
- #6 快速打字检测实际生效:快速打字时 debounce 延迟翻倍(200ms),等用户停顿
- #8 resolveSpecContext 处理带参数的 option(如 --name value):
  识别 option 的 args 字段,自动跳过下一个 token
- #9 Ghost text 位置随终端滚动/渲染更新:注册 term.onRender 回调
- #13 Escape 键不再拦截 vi-mode:仅在 popup 可见时消费 Escape,
  ghost text 显示时不拦截(ghost text 是被动的,不应阻止 shell 交互)
- #14 所有 setState 统一使用 EMPTY_STATE 常量,不再遗漏 expandUpward 字段

架构修复:
- #3 消除 CustomEvent 通信:改为 onAcceptText 回调注入,
  Terminal.tsx 直接传 writeToSession 回调给 hook,
  删除 createXTermRuntime 中约 20 行 listener 代码和 cleanupAutocompleteListener 字段
- #7 xterm 私有 API 访问集中到 xtermUtils.ts:getCellDimensions 统一入口,
  带缓存机制,仅在首次访问或 terminal 切换时触发 DOM 测量
- #16 删除 getCommandNameSuggestions 中多余的动态自导入 await import("./figSpecLoader")

性能修复:
- #5 合并 ghost text 和 popup 的查询路径:删除独立的 getInlineSuggestion,
  fetchSuggestions 只调一次 getCompletions,ghost text 取 completions[0]
- #10 preloadCommonSpecs 分批加载(每批 8 个,requestIdleCallback 间隔),
  延迟 200ms 启动,且检查 enabled 才执行
- #11 scoreEntry 改为 scoreEntryAt(entry, now),now 在查询开始时缓存一次
- #15 scrollIntoView 从 smooth 改为 instant,消除快速导航动画排队
- #19 loadSpec 添加 in-flight 去重(inFlightLoads Map),同一 spec 并发加载只触发一次 import
- #20 存储满时淘汰改为按 score 排序后保留前半,而非按插入顺序

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 修复二审发现的全部 10 个问题

功能修复:
- #1(高) insertSuggestion 改用实时 detectPrompt 而非过时的 lastPromptRef,
  修复用户继续打字后 Tab 选择建议导致字符重复插入的 bug
- #2(中) handleInput Enter 录入历史优先用实时 detectPrompt,
  修复快速打字场景下 recordCommand 记录不完整命令
- #9 suppressNextEnterRecordRef 添加 100ms 安全超时清除,防止 flag 残留
- #10 getNextWord 从 index 1 开始搜索分隔符,修复 ghost text 以 / 开头时
  一次接受全部而非逐段的问题

性能修复:
- #3(中) GhostTextAddon 注册 term.onResize 调用 invalidateCellDimensionCache,
  确保 resize/字体变化后 cell 尺寸缓存正确失效
- #4 updatePosition 缓存 lastLeft/lastTop,位置无变化时跳过 DOM 写入;
  字体属性移到 show() 中只设置一次,不再每帧写 6 个 style
- #5 统一 clearState() 函数替代所有 setState({...EMPTY_STATE}),
  带 popupVisible 守卫避免无效 re-render
- #6 hasSpec 中 specs.includes() 改为 Set.has(),O(1) 查找

架构修复:
- #7 Terminal.tsx 中 autocompleteAcceptTextRef 去掉多余的 useCallback 包装
- #8 删除 AutocompletePopup 的 onClose 死代码 prop

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: popup 默认不选中任何项,用户按 ↑/↓ 后才选中

修复输入 ls 等简单命令时回车误执行联想结果的问题:
- selectedIndex 初始为 -1(无选中),Enter 直接执行用户输入的命令
- 用户按 ↑/↓ 导航后 selectedIndex >= 0,此时 Enter 才执行选中的建议
- Tab 仍然可以直接接受第一条建议(主动接受行为)
- Enter 无选中时关闭 popup 并让按键透传到终端

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fig spec 改为从静态资源 fetch 加载,修复生产构建中补全不工作

根因:@vite-ignore 动态 import 在 Electron 生产构建中无法解析
node_modules 路径(app:// 协议只能访问 dist/ 目录)。

修复方案(与 Monaco 编辑器相同的模式):
- 新增 scripts/copy-fig-specs.cjs,prebuild 时将全部 739 个 fig spec
  从 node_modules/@withfig/autocomplete/build/ 复制到 public/fig-specs/
- Vite 自动将 public/ 内容复制到 dist/,app:// 协议可以正常访问
- figSpecLoader.ts 改用 fetch + Blob URL + dynamic import 加载 spec,
  同时保留 @vite-ignore import 作为 fallback(兼容 dev 模式)
- public/fig-specs 加入 .gitignore(构建时生成,不进版本控制)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: ESLint 忽略 public/fig-specs 目录(第三方生成代码)

与 public/monaco 相同的处理方式——这些是从 node_modules 复制的
第三方构建产物,不应被项目 ESLint 规则检查。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 输入完整子命令名时展示其选项(如 git commit 显示 --message 等)

当 currentToken 完全匹配一个子命令时(如 "git commit" 中的 "commit"),
导航进入该子命令并展示其 options 和 sub-subcommands 作为预览。

之前的逻辑因为 name !== currentToken 过滤掉了完全匹配的项,
且 resolveSpecContext 的 consumedTokens 不包含当前 token,
导致停留在父级而看不到子级的选项。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 修复 fig spec index.js 解析失败导致补全不工作

根因:index.js 格式为 var e=[...],diffVersionedCompletions=[...];
正则 /var\s+\w+\s*=\s*(\[[\s\S]*?\]);/ 要求 ] 后紧跟 ;,
但第一个数组后面是 , 不是 ;,导致非贪婪匹配跳到第二个 ];,
捕获了两个数组拼在一起,JSON.parse 失败,spec 列表为空。

修复:改用 indexOf 找第一个 [ 和对应的 ],直接截取子串解析。
fig spec 的 index 是简单的字符串平坦数组,无嵌套括号。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fig spec 改用 URL 直接 dynamic import,移除 fetch+Blob 方案

fetch + Blob URL + import() 方案可能被 Electron CSP 策略阻止。
改为直接用完整 URL 做 dynamic import:
- dev: import("http://localhost:5173/fig-specs/git.js")
- prod: import("app://./fig-specs/git.js")

两种环境下动态 import 都能正常解析模块,无需 fetch 中间步骤。
同时简化 getAvailableSpecs 也用同样方式,移除 fetch+正则解析。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fig spec 改为通过 Electron IPC 加载,彻底解决 dev/prod 加载问题

之前的方案(静态文件 + dynamic import / fetch + Blob URL)都因为
Vite dev server 对 .js 文件的模块转换和 Electron CSP 限制而失败。

新方案:通过 main process 的 Node.js require() 加载 fig spec,
通过 IPC 传给 renderer:
- main.cjs: 添加 netcatty:figspec:list 和 netcatty:figspec:load handler
- preload.cjs: 暴露 listFigSpecs() 和 loadFigSpec() API
- figSpecLoader.ts: 通过 window.netcatty bridge 调用 IPC

优势:
- main process 直接访问 node_modules,dev 和 production 都可靠
- 无需复制文件到 public/、无需 @vite-ignore hack
- spec 数据通过 IPC 序列化传输,无 CSP 限制
- 删除了 scripts/copy-fig-specs.cjs 和 public/fig-specs/ 相关代码

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: main process fig spec 加载改用 import() 替代 require()

@withfig/autocomplete 是 ESM 包("type": "module"),
CommonJS 的 require() 无法加载 ESM 模块会抛 ERR_REQUIRE_ESM。
改用 dynamic import() 在 async handler 中加载。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: fig spec 加载用 pathToFileURL 绕过 package.json exports 限制

@withfig/autocomplete 的 exports 字段只允许 import "." 和 "./dynamic",
Node.js 严格遵守 exports map 拒绝解析 build/git.js 等子路径。

改为手动拼接文件绝对路径 + pathToFileURL 转换为 file:// URL 后 import,
完全绕过 Node.js 的 package exports 限制。

同时修复 promptDetector 不再 trim 尾部空格(用 cursorX 确定实际输入长度),
确保 "git commit " 的尾部空格被保留,触发空 token 显示选项列表。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: 补全菜单添加详情面板 + 清理调试日志

- 选中或悬停补全项时,右侧显示详情面板(类似 VS Code IntelliSense)
  - 显示完整命令名、来源类型标签(Option/Subcommand/History 等)
  - 显示完整的描述文本(不再截断)
- source 标记移到左侧,与描述分离,更易读
- 悬停和键盘选中都能触发详情面板
- 向上展开时详情面板也正确对齐
- 清理所有临时调试 console.log

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: 清理全部调试日志

移除 autocomplete 模块中所有临时 console.log 调试语句,
仅保留 figSpecLoader 中的 console.warn 用于真实错误报告。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 三审问题修复 — 移除多余 prop、过滤子路径 spec、防路径遍历

1. 移除 Terminal.tsx 传给 AutocompletePopup 的多余 onClose prop
2. getCommandNameSuggestions 过滤含 / 的 spec 名(aws/s3 等不是直接命令)
3. figspec:load IPC handler 添加 .. 路径遍历检查

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 5 个问题全部修复

1. [P1] fuzzy 匹配建议不以 userInput 开头时,用 Ctrl+U 清行再写入完整命令,
   避免 substring 截断产生损坏的命令行
2. [P2] Ghost addon 初始化改用 polling 等待 termRef,解决首次挂载时
   termRef.current 为 null 导致 ghost text 永远不激活的问题
3. [P2] popup overlay 改为 pointer-events-none 透传,仅 popup 自身设
   pointer-events: auto,不再阻止终端区域的鼠标交互
4. [P2] getCompletions 异步返回后重新 detectPrompt 校验输入是否已变,
   丢弃过时的补全结果避免覆盖新状态
5. [P2] prompt 检测支持折行:当 line.isWrapped 时向上回溯查找 prompt 行,
   拼接多行内容作为完整 userInput

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第二轮 3 个问题修复

1. [P2] broadcast 模式下 autocomplete 插入也触发广播 —
   onAcceptText 回调中调用 onBroadcastInputRef 通知其他 session
2. [P2] 支持无尾随空格的 prompt(如 cmd.exe C:\path>)—
   prompt 字符后允许直接是行尾,boundary 为 i+1
3. [P2] 光标移动 escape 序列(Left/Home/End)清除过时建议 —
   不再静默忽略,改为 clearState() 清除 popup 和 ghost text

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第三轮 3 个问题修复

1. [P2] commandBufferRef 处理 Ctrl+U 清行 — fuzzy 匹配发送 \x15 时
   重置 buffer,避免 onCommandExecuted 记录错误的拼接命令
2. [P2] fetchVersionRef 递增计数器废弃过时异步结果 — clearState/Escape
   关闭 popup 时 bump version,getCompletions 返回后检查 version 匹配,
   防止已关闭的 popup 被旧请求重新打开
3. [P2] prompt scanLimit 从 80 提高到 200 — 支持包含 git branch、
   kube context、长路径的 prompt,超过 80 列不再失效

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第四轮 3 个问题修复

1. [P1] 拒绝绝对路径 — figspec:load IPC handler 检查 commandName
   不以 / 或 \ 开头,防止 path.join 丢弃前缀导致任意 JS 执行
2. [P1] cmd.exe prompt > 后不要求空格 — 对 > ❯ ➜ › 等 prompt 字符
   不强制要求后跟空格,支持 C:\src>dir 格式
3. [P2] serial line mode 下 autocomplete 走 serialLineBufferRef —
   在串口 lineMode 时不直接 writeToSession,而是缓冲到 line buffer
   并处理 local echo,与正常按键输入行为一致

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第五轮 — translateToString(false) 保留尾部空格

translateToString(true) 会 trim 行尾空格,导致 cursorX 截取的
userInput 与实际行内容不一致。改为 translateToString(false) 保留
原始空格,确保 "git commit " 的尾部空格被正确保留用于触发选项补全。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: 设置页添加自动补全开关(启用/Ghost Text/弹出菜单)

在终端设置页末尾新增「自动补全」区域,包含三个开关:
- 启用自动补全:总开关
- 行内建议(Ghost Text):光标后灰色建议文本
- 弹出菜单:浮动补全列表

子开关在总开关关闭时 disabled。中英文 i18n 翻译齐全。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第六轮 3 个问题修复

1. [P1] 光标不在行尾时禁止补全 — 检测 cursorX 后方是否有字符,
   有则 clearState 不显示建议,避免 mid-line 插入导致文本重复
2. [P2] Enter 录入历史改为先尝试实时 detectPrompt,失败则 fallback
   到 lastPromptRef 缓存,应对高延迟 SSH 下 buffer 未回显的情况
3. [P2] fuzzy 替换在 Windows host 上用退格清行而非 Ctrl+U —
   cmd.exe/PowerShell 不支持 Ctrl+U,改为发送 \b 退格序列

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第七轮 — commandBuffer 退格处理 + 接受后历史记录

1. [P2] commandBufferRef 处理 \b 退格 — Windows fuzzy 替换用退格
   清行时正确移除 buffer 末尾字符,避免记录拼接错误的命令
2. [P3] lastAcceptedCommandRef 追踪接受的补全文本 — Tab/→ 接受后
   立即 Enter 时用追踪值录入历史,不依赖可能未回显的 buffer

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第八轮 — 历史记录准确性 + 设置同步

1. [P2] 用户继续编辑后清除 lastAcceptedCommandRef — Tab 接受
   "git status" 后追加 " --short" 再 Enter 时记录完整编辑后的命令
2. [P2] Ghost text →/Tab 接受路径也设置 lastAcceptedCommandRef —
   确保所有接受路径在快速 Enter 时都能准确记录命令
3. [P2] autocomplete 设置加入 SYNCABLE_TERMINAL_KEYS —
   跨设备同步时保留自动补全偏好

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第九轮 — REPL 误识别 + 本地终端 OS 检测

1. [P1] local terminal 的 hostOs 改用 navigator.platform 检测实际 OS,
   避免 Windows 上 fallback 到 "linux" 导致 Ctrl+U 清行失败
2. [P2] 回退 > 无条件接受改动,恢复要求 > 后跟空格或行尾 —
   避免 python >>>、mysql>、sqlite> 等 REPL 被误识别为 shell prompt
3. 新增 REPL NON_PROMPT_PATTERNS:>>>(python)和 word>(mysql/redis)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第十轮 4 个问题修复

1. [P1] cmd.exe prompt C:\path> — 对 > 特判:前面是 \ 或 / 时允许无空格,
   避免误匹配 REPL(python>>>、mysql>)的同时支持 Windows cmd prompt
2. [P2] serial lineMode autocomplete 不再 early return — fall through 到
   共享的 commandBuffer/broadcast 更新逻辑
3. [P2] serial 字符模式 + localEcho 时 autocomplete 插入文本也本地回显
4. [P3] 运行时关闭 autocomplete 时调用 clearState() 清除已显示的 popup

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第十一轮 — option args、PS2 误识别、bridge 缓存

1. [P2] resolveSpecContext 返回 option 的 args — 当光标在 option 参数
   位置时(如 git archive --format |),返回该 option 的 args 而非
   subcommand 的 args,使 tar/zip 等枚举值能正确补全
2. [P2] 排除 bare > 作为 shell prompt — bash PS2 续行提示 > 加入
   NON_PROMPT_PATTERNS,避免在多行命令续行和 REPL 中误触发补全
3. [P3] bridge 不存在时不缓存 null — preload 时 bridge 可能未就绪,
   缓存 null 会永久禁用该命令的 spec 补全

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第十二轮 — prompt 检测取最后一个分隔符

Starship/Powerlevel10k 等 prompt 包含多个 prompt 字符
(如 ➜  repo git:(main) $),之前在第一个 ➜ 就停了,
把后续 prompt 文本当成用户输入。

改为收集所有候选 prompt 边界,返回最后一个。确保
"➜  repo git:(main) $ ls" 中 userInput 正确为 "ls"。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 第十三轮 — prompt 搜索范围限制 + cmd.exe 路径

1. [P2] prompt 扫描限制在行前 60% — 避免 "echo foo > bar" 中的
   重定向符 > 被当作 prompt 结束(prompt 不会出现在行尾部分)
2. [P3] cmd.exe 路径检测扩展 — 除了 \ / 前缀,也检测行首是否有
   驱动器号 (X:) 模式,支持 C:\Users\me> 等标准 Windows prompt

P1 (高延迟 SSH buffer 滞后) 和 P2 (Enter 时 stale prompt) 属于
prompt 检测方案的固有局限,根本解决需要 OSC 133 Shell Integration,
不在本 PR 范围内。已有 lastAcceptedCommandRef fallback 缓解。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 19:45:34 +08:00
陈大猫andClaude Opus 4.6 34f9d2a663 chore: 死代码清理与架构分层修复 (#524)
* chore: 移除死代码并修复架构分层违规

- 删除未使用的 ACP 模块 (infrastructure/ai/acp/)
- 删除未使用的 AI 组件 (ExecutionPlan, PermissionDialog)
- 将 syncPayload.ts 从 domain 移至 application 层,修复分层违规
- 移除未使用的导出 (useSecurityState, useProviderStatus, GitHubAuthState,
  getAgentCommandLabel, ImageAttachment, HotkeyActions)
- 收窄 Electron bridge module.exports,移除未使用的导出函数
- 将仅内部使用的函数/类型取消导出 (isSupportedLocale, SyncDashboard)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: 二次审查清理 — 移除更多死代码和架构违规

- 移除未使用的 ConversationEmptyState 组件和类型
- 移除未使用的 PromptInputSelect 系列组件 (5 个导出)
- 移除 global.d.ts 中残留的 SMBConfig 类型和 cloudSyncSmb* 方法声明
- 移除 useAutoSync.ts 中未使用的 toast 导入 (同时修复 application→components 反向依赖)
- 清理因删除而产生的多余 import

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: 消除直接 localStorage 访问,提取 safeSend 共享工具

localStorage 集中化:
- 新增 storageKeys 常量: SIDE_PANEL_WIDTH, PF_RECONNECT_CANCEL, DEBUG_HOTKEYS, DEBUG_UPDATE_DEMO
- TerminalLayer/SettingsApplicationTab/App.tsx/useUpdateCheck 改用 localStorageAdapter
- CloudSyncManager 内部方法改用 localStorageAdapter
- portForwardingService 改用 localStorageAdapter + 集中 key

safeSend 去重:
- 新增 electron/bridges/ipcUtils.cjs 共享模块
- sshBridge/sftpBridge/portForwardingBridge/sshAuthHelper/aiBridge 统一引用

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: 终审清理 — 移除未使用的 require 和废弃类型别名

- 移除 sftpBridge.cjs 中未使用的 require("node:net")
- 移除 aiBridge.cjs 中未使用的 require("node:path")
- 移除 types.ts 中已废弃的 ChatMessageImage 类型别名

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 修复 ESLint 错误 — 组件不再直接导入 infrastructure

- 新增 useStoredNumber hook,TerminalLayer 通过 hook 访问侧边栏宽度
- SettingsApplicationTab 的 isUpdateDemoMode 改为从 useUpdateCheck hook 传入
- 移除 useCloudSync.ts 中未使用的 CloudSyncManager 导入和 GitHubAuthState 接口

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: 提取 notification port,消除 application 层对 components 的依赖

将 toast 通知抽象为 application/notification.ts 端口,
UI 层通过 setNotify 注入实现,useAutoSync 改用 notify 接口。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 14:14:37 +08:00
陈大猫andClaude Opus 4.6 90d161c1b5 refactor: 精简 MCP server 工具集,移除 SFTP/multiExec/terminalWrite
精简 ACP agent 工具集,与 Catty Agent 保持一致,只保留核心工具:
- get_environment
- terminal_execute
- terminal_send_input

移除内容:
- 7 个 sftp_* 工具 (sftp_list_directory, sftp_read_file, sftp_write_file,
  sftp_mkdir, sftp_remove, sftp_rename, sftp_stat)
- multi_host_execute 工具
- ENABLE_SFTP_TOOLS 环境变量和 sftpAvailable 字段
- WRITE_METHODS 中的 sftp/multiExec 条目
- dispatch 中的 sftp/multiExec 路由和 multiExec scope 验证
- mcpServerBridge 中的 sessionSupportsSftp/scopeHasSftpSessions 函数
- getContext description 中的 SFTP 说明

bridge 层的 SFTP/multiExec handler 函数保留(UI SFTP 面板仍在使用)。

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 12:51:43 +08:00
陈大猫andClaude Opus 4.6 7a5b6f506e feat: Catty Agent 支持串口会话命令执行 (#520)
* feat: Catty Agent 支持串口会话命令执行 (#520)

串口连接的网络设备(华为交换机、Cisco 路由器等)使用厂商自有 CLI,
无法识别 Agent 原有的 shell 包裹语法(__NCMCP_ markers、eval、trap)。

新增 execViaRawPty 函数,直接发送原始命令到串口,通过 idle timeout
检测命令完成,无 shell 语法包裹。

- 新增 execViaRawPty:原始命令执行,2s idle timeout 检测完成
- terminalBridge: 串口 session 添加 protocol/shellKind 字段
- mcpServerBridge: handleGetContext 发现串口会话,handleExec/handleTerminalWrite 支持串口
- aiBridge: ai:exec 和 ai:terminal:write 增加 serialPort 分支
- systemPrompt: Agent 提示词增加串口会话使用指南

Closes #520

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: review 问题全量修复

P1:
- handleExec 移除死代码(内层 if 条件永远为 true)
- 串口会话跳过 shell 安全黑名单(shutdown 在 Cisco 是正常接口命令)
- MCP tool 描述更新:terminal_execute/get_environment/multi_host_execute 不再只说 "shell command"
- 串口检查增加 protocol === "serial" guard,不再纯靠 duck typing

P2:
- execViaRawPty 编码改为 latin1,与 terminalBridge 终端解码一致
- exitCode 改为 null(而非 -1),MCP 响应中 null 时不输出 exit code 行
- idle timer 改为收到第一个数据后才启动,避免慢设备超时返回空输出
- idle timeout 默认从 2s 调为 3s,适配低速串口
- serialPort.write 统一用 safeWrite 包裹 try-catch
- echo 剥离仅在 lines.length > 1 时执行,避免误删唯一输出行

P3:
- cancelKey 用简单自增序列替代 crypto.randomBytes
- serialPort.on 前增加 typeof 检查
- finish 函数签名差异增加注释说明

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: 第二轮 review 问题修复

P2:
- MCP server terminal 类工具 (terminal_execute/terminal_send_input/multi_host_execute)
  跳过 blocklist,由 bridge 层做 session-aware 检查,解决串口 shutdown 等命令
  在 MCP 层就被拦截的问题
- handleTerminalWrite (mcpServerBridge + aiBridge) 串口会话跳过 blocklist,
  与 handleExec 保持一致
- handleMultiExec 移除外层 blocklist,每个 session 由 handleExec 独立检查
- 移除 execViaRawPty 中的死代码 receivedFirstChunk 变量
- handleGetContext 返回的 description 补充 serial 会话说明

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex review 问题修复

- [P2] toolExecutors.ts executeTerminalExecute 也需要跳过串口 blocklist,
  否则 Catty Agent renderer 侧的 checkCommandSafety 会在命令到达 bridge
  之前拦截 shutdown 等合法设备命令
- [P2] execViaRawPty 增加 noResponseTimer,无输出命令(enable、
  configure terminal 等)不再等满 60s 整体超时,而是 2×idleMs 后正常返回
- [P1] 串口 blocklist skip 设计决策加注释:serial 协议由用户主动选择,
  如果串口连的是 Linux shell 应使用 local 协议

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex 第二轮 review 修复

- [P2] noResponseTimer 从 2×idleMs 调整为 min(idleMs*4, timeoutMs/4),
  默认 12s,避免截断慢速网络设备操作
- [P1] 串口 blocklist skip 设计说明扩充:serial 协议由用户主动选择,
  且 execViaRawPty 不做 shell 解释,blocklist 中的 shell 元字符
  即使发到串口连接的 Linux shell 也不会被解释执行

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: execViaRawPty echo 阶段使用更长的 idle timeout

ping/traceroute/copy 等命令在回显后可能沉默数秒才产出真正输出。
引入 chunkCount 区分 echo 阶段(前 2 个 chunk)和正式输出阶段:
echo 阶段使用 2×idleMs(默认 6s),正式输出阶段使用 idleMs(3s)。
避免在回显后就误判命令已完成导致输出截断。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: noResponseTimer 增加无输出提示

设备无响应时返回提示信息 "(no output received — command may have
completed silently or may still be running)",让 AI 知道命令可能
仍在执行,避免误认为命令已成功完成后立即发送下一条命令。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: Codex 第 5 轮 review 修复

- [P2] terminal_send_input 串口写入时将 \n 转换为 \r,
  网络设备期望 CR 作为回车而非 LF
- [P2] execViaRawPty 增加 512KB 输出上限,达到上限后停止
  重置 idle timer,避免 noisy session(持续发日志的设备)
  导致命令永远无法完成

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-26 12:04:17 +08:00
panwk 316e46de4b Mod:Removed waitForRootPaint polling helper from electron/bridges/windowManager.cjs.
Removed did-finish-load polling trigger that called markRendererReady via DOM child count checks.
Kept deferred show behavior based on:
ready-to-show
renderer-ready IPC from renderer
timeout fallback (dev and prod values unchanged)
2026-03-25 23:48:56 +08:00
陈大猫 35194036cb Merge pull request #502 from crawt/perf/settings-window-prewarm-hide-on-close
perf(settings): prewarm settings window and hide on close
2026-03-25 01:24:48 +08:00
panwk a886d509f8 perf(settings): prewarm settings window and hide on close
Instead of creating a new BrowserWindow on each user click, the settings window is now:
1. Pre-warmed silently 3 s after app startup (showOnLoad: false)
2. Hidden instead of destroyed when the user closes it
3. Instantly shown/focused on subsequent opens
2026-03-25 00:54:32 +08:00
陈大猫 9deb39dec2 Merge pull request #499 from binaricat/codex/jump-host-proxy-support
Support proxy config on jump hosts
2026-03-25 00:32:09 +08:00
bincxz bb45279d4e Track jump-host proxy socket during chain setup 2026-03-25 00:23:55 +08:00