Recreate cursor-line markers/decorations when xterm disposes them externally, and treat empty decoration sets as intact so fully colored rows do not churn markers every frame.
Fixes#3656
Clamps the window layout minimum to the current display work area so small screens (e.g. 1366×768 at 125%) can fit and shrink. Restores terminal keyboard focus after OS resize/maximize, without stealing from active editable fields.
Fixes#3647
Merged by judgment: four Codex rounds already addressed the real bugs and multi-monitor sync; the latest P2 (protect every non-editable focusable control on resize) is edge-case thrash.
Classify SSH exit-signal as signaled so auto-reconnect does not fight admin idle kicks. Bare transport closed and error still reconnect.
Related to #3617.
* feat(terminal): disable auto-close-on-exit by default
Remote graceful channel closes (VPN half-close, server-side idle kick) can be reported as a clean exit with code 0, so with auto-close enabled the tab gets destroyed exactly when the user needs it kept - bypassing the SSH auto-reconnect loop, which intentionally skips exited events.
Keep tabs on exit by default; users who want close-on-exit can enable it in Settings > Terminal > Behavior. Attach-mode close behavior is unchanged.
* fix(terminal): keep session tabs when the auto-close setting is unavailable
The resolver's own default and the TerminalLayer call-site fallback still
treated a missing terminal setting as auto-close enabled. Remote graceful
channel closes (a jump host reaping an idle session) are reported as a
clean exit, so before the settings store had loaded, those closes silently
deleted the session tab and its scrollback. This completes the opt-in
default introduced above: only an explicit autoCloseOnExit true closes tabs.
* fix(terminal): migrate the materialized auto-close-on-exit default once
Builds before the default flip shipped autoCloseOnExit: true as the
default and materialized it into netcatty_term_settings_v1 on any
settings write, so most upgraders carry a value they never chose and
the new default-off never reaches them: bastion and idle-harvesting
servers reporting clean exits keep silently deleting their tabs.
Reset that materialized true exactly once behind a marker key before
the settings blob is first ingested, then persist the marker so every
later stored value is treated as the user's own choice. The cost is
documented: a user who deliberately enabled auto-close before this
release must re-enable it once.
* fix(terminal): keep synced legacy auto-close from coming back
A device that already wrote the one-shot migration marker was accepting
a pre-migration cloud or backup payload with autoCloseOnExit true.
Unversioned true is rewritten to false. Payloads that record the
migration still sync an explicit opt-in, including across a three-way
merge that would otherwise glue the local marker onto a legacy true.
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
---------
Co-authored-by: xu <tbqyomnnnbltzxy2t@aliyun.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
* feat(terminal): enable SSH auto-reconnect by default
Unexpected transport drops (VPN / Wi-Fi blips) are recoverable events, not
shell exits: the tab stays and the 5s in-place reconnect loop keeps the same
session id, so AI sessions and long-running work continue once the link is
back. With the previous default (off), a dropped link left the tab parked on
'Press Enter to reconnect', which reads as 'the session closed'.
Clean shell exits still close the tab; explicit opt-out in Settings is
preserved. mosh/et sessions self-heal and are unaffected by this gate.
* fix(terminal): skip auto-reconnect after idle timeouts
SSH auto-reconnect stays on by default for transport drops. A timeout
exit (TMOUT, server idle logout, or connect timeout) leaves the tab
disconnected instead of logging back in every 5 seconds.
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
---------
Co-authored-by: yuyang2230 <yuyang2230@users.noreply.github.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
* feat(ai): raise default AI session idle timeout from 30m to 24h
Sessions opened via host_open are auto-closed after 30 minutes without
terminal/file activity. In production use over flaky links this silently
destroyed in-flight agent work: the AI reopens a host, the link drops, the
session sits disconnected (counts as idle), and the reaper closes it
mid-workflow with no error surfaced anywhere. 24h is the normalize clamp
maximum; users can still choose a shorter window in Settings → AI, and the
idle manager still skips sessions with active executions.
* fix(ai): preserve 24-hour session default during renderer sync
* style(ai): keep session idle manager comment ASCII-only
Replace the unicode arrow in the Settings pointer comment with ASCII per the repo AGENTS.md convention (Maintain ASCII-only unless required by existing file content).
* fix(ai): default AI session idle timeout to 8 hours
480 minutes still covers a workday of flaky-link agent work without
sitting on the 24h clamp. Stored values are left untouched.
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
---------
Co-authored-by: yuyang2230 <yuyang2230@users.noreply.github.com>
Co-authored-by: 陈奇 <16399091+binaricat@users.noreply.github.com>
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
* fix(ai): shed oversized tool output before evicting visible chat history
Under localStorage quota pressure the AI sessions serializer degraded
replay ciphertext fine-grained but never shed tool result payloads, so
AI-driven terminal chats (whose persisted bulk is terminal output)
forced the ladder to evict whole oldest sessions and finally collapse
persistence to the newest chat alone. After a restart the history list
looked empty.
Extend the per-message degradation pass to cap oversized tool result
content at a bounded head+tail with a truncation marker, mirroring the
existing ciphertext-shedding tradeoff: every visible conversation now
survives storage pressure at the cost of full tool output on affected
messages.
* fix(ai): keep tool-result truncation strictly smaller than the original
Codex review P2: for tool results only slightly over the cap, the
truncation marker (with JSON-escaped newlines) could make the degraded
copy serialize at the same size or larger than the original, breaking
the ladder's assumption that the degraded representation is the
smallest and sending quota retries farther off.
Drop the marker newlines and shrink the retained head+tail until the
whole copy fits the 8 KiB cap, so a truncated result is always
strictly shorter than its source.
---------
Co-authored-by: xu <tbqyomnnnbltzxy2t@aliyun.com>
The `Freeze trusted helper` step in the `sandbox_smoke` job used a plain
multi-line YAML scalar:
run: cp scripts/ai-automation.cjs "$RUNNER_TEMP/ai-automation.cjs"
chmod 0444 "$RUNNER_TEMP/ai-automation.cjs"
YAML folds the continuation line into the same scalar, so the step ran:
cp scripts/ai-automation.cjs "$RUNNER_TEMP/ai-automation.cjs" chmod 0444 "$RUNNER_TEMP/ai-automation.cjs"
`cp` then sees four operands and treats the last one as the destination
directory, which does not exist:
cp: target '/home/runner/work/_temp/ai-automation.cjs': No such file or directory
The job fails in 7s at step 3, and the remaining 12 steps are skipped.
Reuse the existing `&freeze_ai_helper` anchor, which is what the other eight
`Freeze trusted helper` steps in this workflow already do. It also picks up
`set -euo pipefail` and `test -f scripts/ai-automation.cjs`, so a missing
helper now fails with a clear message instead of a confusing `cp` error.
Observed on a fork mirror running commit b2ed83ae: runs 36847532363 and
36991962473 both failed at this step.
Unit tests stub execFileSync and only found ssh via /usr/bin/ssh, so the
PATH wrapper was skipped on images without that path. Prefer an absolute
findExecutable result, then command -v, then the well-known locations.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
Stats latency already pings the authenticated session. Lock that in with
a unit test that fails if getServerStats dials the session host, which
is what produced kex_exchange_identification every ~30s (#3630).
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: 陈大猫 <binaricat@users.noreply.github.com>
Match argument snippets conservatively and remove stale snippet rows on input edits before the debounced refresh.
Related to #3622; the email-entry/editor concern remains unverified and the issue stays open.
Preserve pending local notes and conflicting external content across failed writes and successful retries.
Retain explicit clear/import replacement ownership until the current generation persists.
Add a detached multi-tab editor window with acknowledged content ownership, focused Monaco shortcuts, and shared close/save confirmation. Preserve source contents until handoff acknowledgement and block edits or incoming transfers during ownership changes.
Continuation of #3439. Original implementation by pplulee.
Co-authored-by: pplulee <hi@pplulee.me>
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(terminal): bound asynchronous Windows PATH refresh
Replace blocking per-terminal registry queries with parallel async reads,
a one-second deadline, and a shared five-second snapshot. Preserve machine
before user PATH ordering and place guessed CLI directories last. Retain
per-hive successful values when registry reads fail, and honor explicit
PATH overrides without querying the registry.
* fix(terminal): cancel pending local starts through boot epochs
Register the existing pending boot controller before awaiting PATH refresh.
Honor tab close and reconnect cancellation before spawning, and clear only
the matching controller so stale completion cannot erase a newer boot.
Regression tests reproduce both the closed-tab PTY spawn and stale-close
ownership failures against the prior revision.
* fix(terminal): abort pending starts before worker close queue
Propagate an owned close to the existing pending boot controller before
waiting for its serialized start. Keep transport cleanup ordered and keep
stale boot-epoch closes from cancelling replacements.
Exercise both fire-and-forget and awaited close messages through the real
worker dispatcher and local-session bridge with a controlled PATH wait.
* fix(terminal): cancel superseded worker starts before queuing
Signal the existing boot-abort controller when a replacement arrives,
reject stale boot epochs, and recheck queued epochs before startup.
Only the latest reconnect may spawn after a pending PATH refresh.
Cover repeated reconnect requests and delayed old starts/closes through
the real worker dispatcher and local-session bridge.
Use xterm's native synchronized output state for viewport preservation, remove the redundant mode tracker, and exercise the production callback path in regression tests.
Fixes#3595
Co-authored-by: si1enteee <si1enteee@users.noreply.github.com>
* feat(vault): add host connection test to the host editor
Add a 'Test' button next to Save in the host editor that opens a dialog reusing the terminal's connection progress/auth/host-key UI, and runs a headless connection probe:
- Main process: netcatty:test-connection reuses startSSHSession in a testMode that authenticates then tears down without opening a shell, reports via netcatty:test:result, and supports cancellation (netcatty:test-connection:cancel).
- Renderer: useHostConnectionTest hook + HostConnectionTestDialog component; single-credential auth (no agent/default-key fallback) with short test timeouts (8s TCP / 15s auth).
- Reuses terminal chain progress, host-key verification, and auth prompts; success keeps the same dialog with retry/close.
* fix(ssh): cancel test connections during chain/proxy dial
Register the headless test-connection transport before connectThroughChain / createProxySocket runs, and update its pendingConn / chainConnections as hops and proxy sockets are established. This lets cancelTestConnection abort a slow/unreachable bastion or proxy dial instead of leaving it running until timeout.
* fix(ssh): address host connection test review feedback
- Route jump-host keyboard-interactive prompts through the external scope in test mode
- Always offer keyboard-interactive for requiresMfa hosts in test mode, including key/certificate-only auth
- Derive the renderer test deadline from the probe options and cancel the attempt on timeout
- Run the macOS Local Network preflight before connection tests
- Add regression tests for jump-host MFA, MFA auth order, test deadline, and macOS preflight
Author: yangheng
Model: claude-opus-5
---------
Co-authored-by: yangheng <492238647@qq.com>
* feat(ai): scope external MCP approvals to connection or host
* test(ai): keep MCP fixture in Netcatty temp directory
* fix(ai): refine external MCP approval UI and grant safety
* fix(ai): wrap long approval targets and support Node 22
* fix(ai): keep automatic scripts in observer mode
* fix(sftp): release preflight markers after transfer
* fix(sftp): retain preflight cap through conflict resolution
* Keep SFTP preflight cap across panel remount
* Update worker terminal metadata expectations
* Persist SFTP preflight skip state across restart
* Limit persistent host approvals to saved hosts
* Scope host-open approvals to the selected saved host
* Exclude serial hosts from persistent approval scopes
* fix(terminal): commit live IME transcription once
xterm sends a rewritten dictation hypothesis as the interim, the full
final string, and the tail. Keep the pending textarea diff from firing
again when composition starts, and send only the part that is not
already on the PTY.
* fix(terminal): retype rewritten IME tails as separate backspaces
A shared suffix has to be deleted and typed again because the cursor
stays at the end of the line. One-character hypotheses are real
prefixes, and each backspace is its own input event so Ctrl-H hosts
still erase.
* fix(terminal): rewrite a revised IME hypothesis in place
A composition that replaces the snapshot, instead of extending it,
sends the cancelled textarea edit. Deletions are counted in graphemes
so a surrogate pair is one backspace.
* fix(terminal): send a new IME word that shares a finished prefix
A composition that starts at or after the snapshot is a new word.
Sharing the first character with text already on the PTY is not
enough to treat it as the same hypothesis.
* fix(terminal): remember a rewritten IME hypothesis in full
A 229 rewrite stored only the inserted tail, so a later
compositionend with no composition marker sent the revised
line again. Remember the whole revised textarea instead.
* fix(terminal): preserve IME rewrite state across delayed commits
Observer mode stays on AI-started scripts that write to the terminal. Compose bar, connect, and panel runs are explicit user actions and no longer read that setting.
* fix(#3559): automated AI fix
* fix: address Codex review on PR #3560
* fix: address Codex review on PR #3560
* fix: address Codex review on PR #3560
* fix: address Codex review on PR #3560
* fix: keep listing size out of the dual-pane transfer plan
A stat-less source can show the pane listing in the overwrite dialog, but that size must not be written back as totalBytes or a short listing truncates the file again. Files past the preflight cap stay on an unknown plan and are not statted a second time.
* fix: stop preflight after a timeout and honor cancel
A half-open browse session no longer walks the whole selection at 10 seconds per file. Cancel during the overwrite size stat is checked before an apply-to-all replace deletes the existing target.
* fix: do not stat remote files again after a failed preflight
A stat that returns no usable size, times out, or never runs is not repeated for the whole selection. The transfer plan stays unknown so the bridge measures the live size.
---------
Co-authored-by: netcatty-bot <308658023+netcatty-bot@users.noreply.github.com>
Co-authored-by: bincxz <16399091+binaricat@users.noreply.github.com>
* fix(ui): apply the chosen app icon to the Windows tray
Settings already swap the taskbar icon. On Windows, the notification
area now follows that choice. Original keeps the multi-size tray ico;
other variants are scaled to the small-icon size Windows asks for.
* fix(ui): crop the app-icon margin before sizing the Windows tray
The variant PNGs keep a transparent inset, so scaling them straight into
the notification slot made every non-original style smaller than the
full-bleed tray ico. Crop to the opaque artwork first.
* test(sftp): update the side-panel reuse source lock
The single-channel change replaced the connected-only sourceSessionId
ternary with reuseTerminalTransport. The source lock still expected the
old expression, so lint-and-test fails on main.
When /proc/<pid>/cwd is unreadable, accept an lsof name ending in (verb: message) only if that path is an existing directory. This drops the AL2023 record "/proc/<pid>/cwd (readlink: Permission denied)" instead of using it as the SFTP follow or upload target. Real directories with parenthesized names stay usable.
Fixes#3493.
Add a host and group toggle for bastions that allow one session channel per TCP connection. After the handshake, BHostSSH, TERM-SSHD, and CLOUDBILITY take that path automatically. File transfer uses a separate SFTP login, and the interactive terminal is not given another exec, SFTP, copy/split, or port-forward channel.
The manual toggle remains for bastions that still advertise OpenSSH.
fontLigatures already defaulted to on, but the terminal never loaded
the ligatures addon, so Fira Code and Maple Mono kept drawing != as two
cells. Activate the addon after open and before WebGL, and rebuild the
WebGL atlas when the setting changes so glyphs are not rasterized
without the calt feature.
The addon bundles lru-cache's node:diagnostics_channel import. Alias
that builtin to a no-op channel in the Vite renderer build and in the
Electron live-test bundles.
Closes#3543
* feat(ai): add MiMo Code (mimo) agent support scaffolding
Registry/wiring only: discovery entry, managed-agent metadata, model presets, icon, i18n and Settings UI for Xiaomi's MiMo Code CLI (an OpenCode fork). The SDK driver is intentionally not wired yet - it depends on the pending @mimo-ai/sdk parity probe, so DRIVER_REGISTRY keeps a TODO and run-turn/list-models reject the mimo backend until then.
* feat(ai): add MiMo Code (mimo) agent support
Adds Xiaomi's MiMo Code CLI as a managed agent backend, mirroring the existing
OpenCode integration (discovery, managed-agent metadata, model presets, icon,
i18n, Settings UI) and adding the SDK driver that was left as a TODO in the
scaffolding commit.
Driver notes:
- New electron/bridges/aiBridge/sdk/mimoDriver.cjs reuses the pure translation
helpers exported by opencodeDriver.cjs (config/prompt building, event
translation, model mapping) because MiMo Code is an OpenCode fork and speaks
the same wire protocol.
- It does NOT use @mimo-ai/sdk's createOpencode(). On 0.1.15 that helper spawns
a binary named "opencode", waits for an "opencode server listening" banner,
and passes config via OPENCODE_CONFIG_CONTENT; the mimo CLI prints
"mimocode server listening on <url>", reads MIMOCODE_CONFIG_CONTENT, and ships
as "mimo"/"mimo.exe". So createOpencode() can only ever time out. The driver
runs "mimo serve" itself, parses the readiness URL, and attaches the SDK's
createOpencodeClient({ baseUrl }).
- Binary resolution honours binPath, MIMOCODE_BIN and MIMOCODE_BIN_PATH, then
falls back to PATH via resolveCliFromPath (the npm shim is mimo.cmd on
Windows, which Node cannot spawn directly).
- @mimo-ai/sdk ^0.1.15 added as a dependency.
Verified on Windows against mimo 0.1.15 + @mimo-ai/sdk 0.1.15:
- "mimo serve --port" boots and reports its URL; driver parses it.
- SDK symbols used all exist: createOpencodeClient, client.config.providers(),
client.global.event() (SSE stream), client.session.create/abort/promptAsync.
- Live provider catalog reads 2 providers (xiaomi, mimo); MIMO_MODEL_PRESETS
updated to the real lower-case provider-prefixed ids and its TODO removed.
- Server process is killed on close (taskkill tree kill), no orphan processes.
Not yet verified at runtime:
- A full end-to-end turn (no prompt was sent, so no tokens were spent): text /
reasoning / tool-part streaming and abort handling are untested against a
live model.
- MIMOCODE_CONFIG_CONTENT accepting the OpenCode config schema (permission,
mcp, model, share, autoupdate) end-to-end; only that the env var exists in
the binary has been confirmed.
- Whether the loopback server needs MIMOCODE_SERVER_PASSWORD when the API
client attaches.
- The placeholder public/ai/agents/mimo.svg still needs an official asset.
- Multi-turn resume (resumeSessionId) and MCP injection against the live CLI.
* feat(ai): inject MIMOCODE_BIN for discovered MiMo Code agents
Mirror the OpenCode OPENCODE_BIN handling so a discovered or re-detected
mimo binary path reaches the driver, and cover it with the same assertion
shape as the opencode managed-agent env test.
* fix(ai): use MiMo branding in the mimo driver
* test(ai): cover the mimo driver
Add a unit test file for the MiMo Code driver, matching the existing
opencode driver test in shape, runner (node:test via node --test) and
`.test.cjs` extension. It is the last driver in the registry without
one.
Coverage:
- spawn/launch: `mimo serve --hostname/--port` argv, binary resolution
(explicit binPath / MIMOCODE_BIN / MIMOCODE_BIN_PATH), pass-through
env and MIMOCODE_CONFIG_CONTENT; `child_process.spawn` is stubbed, so
no real `mimo` process is ever started.
- readiness: parsing of `mimocode server listening on <url>` from
stdout (including split chunks) and stderr, the OpenCode spelling,
and the MIMO_SERVE_TIMEOUT_MS timeout path (asserts the rejection and
that the child is killed).
- abort/cleanup: stopMimoProcess kill/idempotency/swallow, the Windows
taskkill tree kill, server.close() teardown, and abort mid-turn with
a throwing close().
- branding: OpenCode -> MiMo Code for the busy/idle/status, failed and
tool-failure fallback strings, with tool output and non-brand
technical errors left verbatim.
- happy path: session create -> streamed event translation -> done,
plus listMimoModels catalog mapping and the empty/aborted paths.
* fix(ai): filter MiMo events by session and stop the server behind the npm shim
Address review on #3500:
- The global SSE stream carries every session on the server, and
runMimoTurn translated all of them, so another session's text could leak
into the reply and its session.idle could end the turn early. Skip events
whose session id is not ours, reusing the OpenCode driver's
getOpenCodeSessionIdFromEvent (now exported). Covered by a test with
interleaved sessions.
- On macOS/Linux the official @mimo-ai/cli `mimo` is a Node shim that runs
the native binary through spawnSync, so killing the shim left the server
listening. `mimo serve` is now spawned detached on POSIX and
stopMimoProcess signals its whole process group (SIGTERM, then SIGKILL
after a grace period). Windows keeps taskkill /T.
Verified against the real @mimo-ai/cli@0.1.15 npm shim: on Linux (WSL
Ubuntu 22.04) the listener is the native .mimocode child; before this change
it was still listening 3s after close(), now the port is released within 1s
and no mimo process remains. On Windows the .cmd shim path also releases the
port within 1s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: harden MiMo CLI integration before merge
* fix: wait for MiMo events and handle setup cancellation
* fix: finish MiMo stream handling and scope skill access
* fix: preflight MiMo skill paths before one server startup
* fix: scope MiMo skill reads to trusted roots
* fix: allow MiMo skills from ancestor project roots
* fix: honor MiMo observer and confirm command permissions
* fix: clear MiMo approval when its service exits
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bincxz <16399091+binaricat@users.noreply.github.com>
* feat(vault): let Skills + CLI operate notes with the same approval as MCP
Expose vault note list, read, write, and markdown import on the tool CLI so agents can save generated documents only after the user approves.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(vault): keep explicit empty note fields in the CLI
A supplied empty string was treated as an omitted flag, so note clears and empty markdown imports never reached the note service.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(vault): allow an explicit empty note body on CLI create
notes create --content "" was rejected as a missing flag, so it did not match vault_notes_create.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(vault): stop echoing imported note bodies and recognize leading CLI flags
A batch import resent every document over IPC, and a leading --json hid the notes command from artifact inference.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(vault): make note CLI import safe and fail writes cleanly
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: bincxz <16399091+binaricat@users.noreply.github.com>