ssh2 emits conn.once("close") before stream.on("close") during
transport drops. The conn.close handler was sending exit + deleting
the session, then stream.close would send a second misleading exit.
Now stream.close checks sessions.has() before sending exit, while
still flushing the data buffer unconditionally. This ensures:
- Buffer flush always happens (no data loss)
- Exit event is sent exactly once
- Transport errors are correctly reported
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P1: Change authMethods.length condition from > 1 to >= 1 so the
dynamic authHandler (which includes 'none' probing) is always used,
even when only keyboard-interactive is available. Fixes the
passwordless embedded device case when no keys/agent are discovered.
P1: Add PPK encryption detection to isKeyEncrypted() — check for
"Encryption:" header in PuTTY PPK format. Without this, encrypted
.ppk files were treated as unencrypted and attempted without a
passphrase, failing silently instead of triggering the passphrase
retry flow.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P1: Add "none" to the agent-mode simple array auth path so passwordless
devices work even when agent forwarding is configured.
P1: Extend looksLikePrivateKey() to recognize PuTTY PPK format
("PuTTY-User-Key-File" prefix) so PPK keys in ~/.ssh/ are not
incorrectly filtered out.
P2: Add stat().isFile() check before readFile() in all key discovery
paths to skip FIFOs, sockets, directories, and other non-regular files
that would block readFile() indefinitely.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P1: Transport errors on established sessions now surface correctly.
The stream.on("close") handler (which fires before conn close and
after buffer flush) checks session._transportError and sends exit
with exitCode:1 and the error message instead of a misleading
exitCode:0 "closed".
P1: Add looksLikePrivateKey() content validation to all key discovery
functions. Files matching id_* that don't start with "-----BEGIN" or
"openssh-key-v1" are skipped, preventing non-key files from being
passed to ssh2 as privateKey (which would abort connect before
password/agent fallback could run).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Codex P2: when a transport error (ECONNRESET) arrives after the session
is established, the error handler was immediately sending netcatty:exit,
causing preload to remove data listeners before the stream close handler
could flush the 8ms data buffer. Users would lose the last chunk of
terminal output.
Now the error handler stores the error message on the session object
(_transportError) instead of sending exit immediately. The close handler
(which fires after stream close + buffer flush) checks for this flag
and sends the exit event with the transport error info.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Medium: Close handler now checks sessions.has(sessionId) before sending
netcatty:exit, preventing a misleading exitCode:0 "closed" event after
the error handler already reported the real transport failure.
Medium: Array-based auth path in buildAuthHandler now includes "none"
as the first method, matching the dynamic handler behavior.
Low: Set lastAttemptedLabel to "none (no credentials)" so the rejection
message is consistent with the initial onAuthAttempt callback.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the fixed DEFAULT_KEY_NAMES array ("id_ed25519", "id_ecdsa",
"id_rsa") with a directory scan using /^id_[\w-]+$/ regex, matching
Tabby's PrivateKeyLocator behavior. This discovers keys like
id_ed25519_work, id_dsa, or any custom-named key automatically.
Preferred keys (ed25519, ecdsa, rsa) are still tried first, followed
by any additional keys found in alphabetical order.
Applied to both sshBridge.cjs and sshAuthHelper.cjs (all four
key discovery functions + the get-default-keys IPC handler).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Restore unconditional 'none' auth as the first method tried. Per
RFC 4252, the 'none' request is the standard way for clients to
discover which auth methods the server supports. It also enables
passwordless login on embedded devices (#482).
This matches the behavior of OpenSSH (which always sends 'none'
first) and Tabby (which unconditionally adds { type: 'none' } as
the first element of allAuthMethods). Most SSH servers do not count
'none' toward MaxAuthTries per the RFC.
Applied to both the main SSH authHandler and the shared
buildAuthHandler used by SFTP/chain/exec connections.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Codex review identified P1 issues: automatic 'none' auth before any
other method can exhaust MaxAuthTries on hardened servers, breaking
connections that previously worked. The 'none' auth support for
embedded devices should be a user-facing option, not automatic.
This commit reverts the 'none' auth additions while keeping the
crash prevention fixes (settled guard, conn.destroy, error wrapping).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P2: Only try 'none' auth when no explicit credentials (password/key/agent)
are configured. Avoids wasting an auth attempt on servers with low
MaxAuthTries.
P2: Post-settle errors on active sessions now send netcatty:exit to the
renderer instead of being silently swallowed, so transport failures
(keepalive timeout, ECONNRESET) are correctly reported as errors.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The SSH protocol's 'none' auth method allows login without any
credentials — common on embedded devices (routers, switches) where
root has no password. ssh2 tries this by default, but Netcatty's
custom authHandler and buildAuthHandler overrode the default behavior
and never attempted 'none', making it impossible to connect to these
devices.
Now both authHandlers try 'none' as the first method (before any
other auth) on the initial call (methodsLeft === null). If the server
accepts it, the connection succeeds immediately. If rejected, the
normal auth flow continues with publickey/password/keyboard-interactive.
This is the root cause of #482: the user's embedded device needed
'none' auth, but Netcatty never tried it, then the auth failure +
ECONNRESET combination crashed the app.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When connecting to embedded devices with legacy algorithms and no password,
the SSH connection could crash the app with an uncaught ECONNRESET exception.
Three fixes:
1. Guard against duplicate error handling in conn.on("error") — once the
promise is settled, late errors (e.g. ECONNRESET after auth failure)
are logged but no longer re-reject or re-notify the renderer.
2. Destroy the SSH connection on error/timeout to prevent the underlying
TCP socket from emitting further uncaught errors.
3. Wrap non-auth errors in startSSHSessionWrapper with clean Error objects
so Electron's ipcMain.handle can serialize them back to the renderer.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P2: Wrap keyboard-interactive handlers in SSH chain, SFTP chain, and
SFTP main connections to emit "waiting for user input..." and "user
responded" progress events, matching the SSH main connection behavior.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P2: Remove premature onAuthAttempt calls from buildAuthHandler's array
branch — methods are listed before connect(), making logs inaccurate.
P2: Handle "waiting for user input..." and "user responded" as literal
log messages, not as "Trying X..." format, in both SSH and SFTP.
P3: Clear connectionLogs after successful SFTP connect so directory
navigation doesn't replay stale auth transcript in the loading overlay.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P2: Emit onAuthAttempt notifications from buildAuthHandler's array
branch so single-method SFTP connections (e.g. password-only) show
auth method logs in the connection panel.
P3: Show connectionLogs in the cached-files loading overlay so repeat
connections still display auth progress during reconnect.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- add explicit Linux/macOS guard in server-stats hook
- return UNSUPPORTED_OS from ssh bridge when uname is not Linux/Darwin
- fail fast when stats payload cannot be parsed to avoid futile polling
- wire Terminal to pass supported-OS hint to useServerStats
P2: Guard SFTP progress callback with navSeqRef check to prevent stale
auth logs from leaking into a reused tab after retry/disconnect.
P3: Reset connectionLogs when connecting to local filesystem, avoiding
stale remote auth logs showing in the local pane.
P3: Emit 'connected' progress event when the final SFTP SSH session
is ready, so the log confirms the connection completed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- auto-detect remote OS in sshBridge using uname -s
- add macOS stats collection path (CPU, memory, swap, processes, disk, network)
- keep existing Linux stats pipeline and parsing logic
- remove Linux-only gating in useServerStats and Terminal display logic
- show server stats whenever connected (not restricted by host.os)
- add CPU hover fallback UI when per-core data is unavailable (e.g. macOS)
- update bridge type docs in global.d.ts to reflect cross-OS stats support
Add detailed authentication method logs to both SSH terminal and SFTP
connection flows, giving users visibility into which methods are tried,
rejected, or require input.
Backend (shared):
- sshAuthHelper buildAuthHandler: track lastAttemptedLabel, log method
rejections and "all methods exhausted" via onAuthAttempt callback
- sftpBridge: add sendSftpProgress helper, wire onAuthAttempt to both
chain and main buildAuthHandler calls, emit connecting/authenticating/
connected/error progress events via new IPC channel
Backend (SSH-specific):
- sshBridge: log method rejections in custom authHandler, log
keyboard-interactive prompt/response and all-methods-exhausted
IPC/Bridge:
- preload: register netcatty:sftp:connection-progress listener, expose
onSftpConnectionProgress in bridge API
- global.d.ts: add onSftpConnectionProgress type
Frontend (SFTP):
- types.ts: add connectionLogs to SftpPane
- useSftpConnections: subscribe to progress events during connect,
convert to human-readable log lines, accumulate in pane state
- SftpPaneFileList: show logs below spinner during connecting, show
expandable "Show logs" in error view with collapsible log panel
Frontend (SSH):
- createTerminalSessionStarters: format rejected methods with ✗ prefix
and "all methods exhausted" message
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
P1: Pass identityFilePaths for jump hosts in SFTP credentials so chain
connections can load IdentityFile keys for bastion hosts.
P2: When the passphrase dialog is skipped or times out (not just
cancelled), clear the encrypted key and continue to the next identity
file. Previously skip/timeout fell through and left the encrypted key
in connOpts, causing the same stall this feature is meant to fix.
Applies to all 4 identity file loading paths (SSH chain, SSH main,
SFTP chain, SFTP main).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the manual-only text input with a file picker button that opens
the system file dialog (showOpenDialog with showHiddenFiles enabled so
~/.ssh/ keys are visible). Users can still type a path manually or use
the browse button.
Changes:
- electron/main.cjs: add netcatty:selectFile IPC handler
- electron/preload.cjs: expose selectFile on bridge
- global.d.ts: add selectFile type
- HostDetailsPanel.tsx: add FolderOpen browse button next to path input
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SSH config import now parses the `IdentityFile` directive and stores
the file paths on the host as `identityFilePaths`. At connection time,
the SSH and SFTP bridges resolve these paths, read the key file content,
and use it for authentication — matching the behavior of OpenSSH and
Tabby.
If the key file is encrypted, a passphrase dialog is shown before
connecting. If the user cancels, the key is skipped and auth falls back
to other methods. If the file doesn't exist, a warning is logged and
the next key path is tried.
Changes:
- domain/models.ts: add `identityFilePaths` to Host interface
- domain/vaultImport.ts: parse `IdentityFile`, expand `~`, store paths
- global.d.ts: add `identityFilePaths` to NetcattySSHOptions and
NetcattyJumpHost types
- createTerminalSessionStarters.ts: pass identityFilePaths for both
main connection and jump hosts
- useSftpHostCredentials.ts: pass identityFilePaths for SFTP
- sshBridge.cjs: read identity files at connection time for both main
and chain connections, with encrypted key passphrase prompting
- sftpBridge.cjs: same for SFTP main and chain connections
Closes#463
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Codex P2: when the passphrase dialog is cancelled, the thrown
error now includes 'authentication' in the message and sets
level='client-authentication'. This allows the SFTP frontend's
isAuthError() check to recognize it and fall back to the password
retry path, preserving the key-first-then-password behavior.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Codex P3: the passphrase modal was showing UUIDs or generic
placeholders like "private-key" / "hop-1-key" instead of the host
label or hostname. Now pass the human-readable label/hostname as
keyName so users can identify which key needs the passphrase.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Codex P2: when using a proxy and an encrypted key, cancelling
the passphrase dialog cleaned up chain connections but leaked the
proxy socket in connectionSocket. Now explicitly destroy it.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Codex P2: when the passphrase dialog is cancelled for the
final SFTP host, any already-open proxy/jump-host connections were
leaked because the throw bypassed the cleanup path. Now explicitly
end all chain connections before throwing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address Codex P1 review: when the passphrase dialog is skipped or
times out, the encrypted key was left in connOpts.privateKey without
a passphrase. buildAuthHandler would still attempt it as publickey-user,
causing the same stall this PR fixes. Now delete connOpts.privateKey
in all non-success paths so auth falls back to password/keyboard-interactive.
Applies to SSH chain, SFTP chain, and SFTP main connection paths.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When an SSH config specifies an encrypted IdentityFile for a jump host
(e.g. `IdentityFile ~/.ssh/id_ed25519` with passphrase protection),
the chain connection passed the encrypted key to ssh2 without a
passphrase. ssh2 failed to parse it and the auth hung until timeout,
with no user-visible prompt.
The same issue existed for SFTP connections using encrypted keys.
Now detect encrypted keys via `isKeyEncrypted()` before connecting and
prompt the user for the passphrase via the existing passphrase dialog.
If the user cancels, a clear error is shown. If skipped, auth falls
back to other methods (password, keyboard-interactive, default keys).
Closes#463
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add EHOSTDOWN, ENETDOWN, EPROTO, EPERM to the isNonFatalNetworkError
check. Also refactor to switch/case for readability.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
ssh2 emits multiple error events per failed connection (e.g. ECONNRESET
followed by "Connection lost before handshake"). Several code paths used
`.once("error")` which removed the listener after the first event,
leaving the second error unhandled and crashing the process via the
uncaughtException handler's re-throw.
Root cause: `runDistroDetection` ran unconditionally after connection
attempts (including failures), creating a new SSHClient to the same
unreachable host. Its `execCommand` used `.once("error")`, so the
second ssh2 error event had no listener and became an uncaught exception.
Fixes:
- execCommand: `.once("error")` → `.on("error")` with settled guard and
explicit `conn.end()` cleanup
- runDistroDetection: move into try block so it only runs after
successful connections
- portForwardingBridge: same `.once` → `.on` fix
- sftpBridge: add catch-all error listener after cleanup() removes the
pre-ready listeners
- main.cjs: suppress non-fatal SSH/network errors in uncaughtException
and unhandledRejection handlers as defense-in-depth (log to crash
bridge, do not re-throw)
Closes#452
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: support bare IPv6 addresses in quick connect and fix IPv6 display
- Accept un-bracketed IPv6 addresses (e.g. 2607:f130::4f06) in quick
connect input. The main regex requires brackets for IPv6+port, but now
falls back to detecting bare IPv6 (2+ colons, hex-only) when the
primary pattern fails.
- Add formatHostPort() helper that wraps IPv6 addresses in brackets
when appending a port, preventing ambiguous displays like
"2607:f130::4f06:22"
- Apply formatHostPort in QuickConnectWizard, TerminalConnectionDialog,
and SftpSidePanel
- Fix hop label formatting in sshBridge and sftpBridge for IPv6 jump
hosts
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: truncate long hostnames in connection dialog
Add truncate to the host label and protocol subtitle in the connection
dialog so long IPv6 addresses don't overflow into the action buttons.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: constrain connection dialog header so truncate works correctly
Add min-w-0/flex-1 to the left side of the header flex container and
shrink-0 to the avatar so long hostnames truncate instead of pushing
into the Show logs / close buttons.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: prevent action buttons from being squeezed by long hostname
Add shrink-0 and left margin to the right-side button group so truncated
text doesn't crowd into Show logs / close buttons.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: tighten bare IPv6 detection to avoid MAC address false positives
Only accept bare (un-bracketed) hex:colon strings as IPv6 if they
contain '::' (unambiguously IPv6) or have exactly 7 colons (full
8-group notation). This rejects MAC addresses like aa:bb:cc:dd:ee:ff
(5 colons) which would otherwise trigger quick-connect mode.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: avoid double-wrapping already-bracketed IPv6 hop labels
Add !startsWith('[') guard so hostnames that are already bracketed
(e.g. from URL-imported hosts) don't produce malformed labels like
[[2607:f130::4f06]]:22.
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 SSH chain connection hang and improve connection progress
- Fix Promise never settling when conn 'close' fires before 'ready'
during chain connections, which caused "reply was never sent" error
- Replace fake timed progress animation with real backend events
- Send granular connection progress for all SSH connections (not just
chain), including: connecting, key exchange, auth attempts, forwarding,
shell opening
- Surface auth method attempts (SSH agent, key names, password) in
progress logs so users can diagnose authentication failures
- Include error details in progress events for better error visibility
Closes#463
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: scope progress events by sessionId, prevent duplicate errors, hide chain UI for direct SSH
- Add sessionId to chain progress payload so events are scoped per session (P1)
- Set settled=true in error/timeout handlers to prevent close handler from
emitting a second misleading 'closed unexpectedly' error (P2)
- Only show chain progress UI when total > 1 so direct SSH connections
don't render as 'Chain 1/1' (P3)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: mark shell-open failure as settled before closing connection
The conn.shell() error branch calls conn.end() which triggers the close
handler, but settled was not set yet, causing a duplicate 'closed
unexpectedly' error to overwrite the real shell-open failure message.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Close/destroy the SSH exec stream when the 5s timeout fires to
avoid leaking session slots (MaxSessions).
- Treat SFTP realpath('.') returning '/' as non-authoritative so
non-root users fall through to the candidate probe chain instead
of incorrectly opening at root.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Prevent indefinite blocking when the remote shell init hangs or a
forced command never exits. Falls through to SFTP realpath after
timeout.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Query the remote server for the real home directory using two methods:
1. SSH exec `echo ~` — works for any user regardless of home path
2. SFTP realpath('.') — fallback, SFTP cwd is typically home dir
Falls back to the previous hardcoded /home/{username} candidates if
both methods fail. This fixes SFTP auto-open sidebar not navigating
to the correct directory for users with non-standard home paths
(e.g. /usr/home, /export/home, custom paths).
Fixes#458
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Extract error properties (code, errno, syscall, hostname, port,
signal, level) into errorMeta field for system-level diagnostics.
- Add extra field for structured context (e.g. render-process-gone
reason and exitCode as separate fields, not just a string).
- Add process PID for correlating with OS-level logs.
- Accept optional extra parameter in captureError() for callers to
attach structured context data.
- Display errorMeta and extra as tagged badges in the crash log viewer.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add electronVersion, osVersion, memoryUsage (RSS/heap in MB),
activeSessionCount, and process uptime to each crash log entry.
Display these fields inline in the Settings crash log viewer.
These extra fields help diagnose issues like #452 where knowing
the session count and memory state at crash time is critical.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- listLogs: stream-count newlines instead of reading entire file content
just to compute entryCount.
- readLog: read only the last 256KB of large files and parse the tail,
avoiding O(file_size) memory/CPU for crash-loop scenarios.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>