macos: add pid and tty properties to AppleScript terminal and App Intents TerminalEntity by TweedBeetle · Pull Request #11922 · ghostty-org/ghostty

TweedBeetle · 2026-03-28T11:55:35Z

Summary

Add pid and tty as read-only properties to:

The AppleScript terminal class
The App Intents TerminalEntity

This enables reliable process-to-terminal mapping for automation tools when multiple terminals share the same working directory.

Changes

macos/Ghostty.sdef: add pid and tty to the terminal class
macos/Sources/Features/AppleScript/ScriptTerminal.swift: add AppleScript getters
macos/Sources/Ghostty/Ghostty.Surface.swift: add Swift accessors for foreground PID and PTY name
src/apprt/embedded.zig: add C bridge exports for foreground PID and PTY name
include/ghostty.h: declare the new C API entry points
macos/Sources/Features/App Intents/Entities/TerminalEntity.swift: surface the same values in App Intents

Testing

zig build -Demit-macos-app=false
/opt/homebrew/bin/nu macos/build.nu --configuration Debug --action build
swiftlint lint 'macos/Sources/Features/AppleScript/ScriptTerminal.swift'
swiftlint lint 'macos/Sources/Features/App Intents/Entities/TerminalEntity.swift'
swiftlint lint 'macos/Sources/Ghostty/Ghostty.Surface.swift'
AppleScript smoke test against the built app:
- get {id, pid, tty, working directory} of terminal 1
- get pid of every terminal
- get tty of every terminal
- Verified returned pid/tty against ps -p <pid> -o pid=,tty=,comm= and lsof -a -p <pid> -d cwd -Fn

AI Disclosure

This PR was developed with AI assistance (Codex). The contributor understands the changes and can explain the implementation.

Closes #11592
Closes #10756

jcollie

Only commenting on the Zig code. The Swift code will need review by someone else.

jcollie · 2026-03-28T15:26:22Z

src/apprt/embedded.zig

        }
    };

+    // ghostty_string_s


This is partially duplicating source from main_c.zig. The String definition should be moved to a shared location (maybe src/c/string.zig?) and referenced from there.

ghostty/src/main_c.zig

Lines 62 to 102 in 0d1f77b

/// ghostty_string_s

pub const String = extern struct {

ptr: ?[*]const u8,

len: usize,

sentinel: bool,

pub const empty: String = .{

.ptr = null,

.len = 0,

.sentinel = false,

};

pub fn fromSlice(slice: anytype) String {

return .{

.ptr = slice.ptr,

.len = slice.len,

.sentinel = sentinel: {

const info = @typeInfo(@TypeOf(slice));

switch (info) {

.pointer => |p| {

if (p.size != .slice) @compileError("only slices supported");

if (p.child != u8) @compileError("only u8 slices supported");

const sentinel_ = p.sentinel();

if (sentinel_) |sentinel| if (sentinel != 0) @compileError("only 0 is supported for sentinels");

break :sentinel sentinel_ != null;

},

else => @compileError("only []const u8 and [:0]const u8"),

}

},

};

}

pub fn deinit(self: *const String) void {

const ptr = self.ptr orelse return;

if (self.sentinel) {

state.alloc.free(ptr[0..self.len :0]);

} else {

state.alloc.free(ptr[0..self.len]);

}

}

};

We have a "standard" libghostty string type in src/lib/string.zig (we use for libghostty-vt). We should probably restandardize on that. Short of that though I agree with jcollie here.

Addressed in 82e66aa17. I extracted the C API String into src/c/string.zig for this PR and updated main_c.zig, embedded.zig, and config/CApi.zig to share it. Agreed that any broader libghostty string restandardization would be follow-up work.

jcollie · 2026-03-28T15:28:12Z

src/apprt/embedded.zig

+    /// freed by the caller via ghostty_string_free.
+    export fn ghostty_surface_tty_name(surface: *Surface) String {
+        const tty_name = surface.core_surface.getProcessInfo(.tty_name) orelse return .empty;
+        const copy = global_state.alloc.dupeZ(u8, tty_name) catch |err| {


The allocator should be available from surface.app.core_app.alloc thus making the import of global.zig above unnecessary.

Addressed in 82e66aa17. ghostty_surface_tty_name now allocates via surface.app.core_app.alloc, and the extra global_state import in embedded.zig is gone.

mitchellh

Swift side looks fine, agree on the Zig review from @jcollie

mitchellh · 2026-03-28T15:51:44Z

src/apprt/embedded.zig

        }
    };

+    // ghostty_string_s


We have a "standard" libghostty string type in src/lib/string.zig (we use for libghostty-vt). We should probably restandardize on that. Short of that though I agree with jcollie here.

mitchellh · 2026-03-28T16:34:07Z

src/c/string.zig

Sorry actually we shouldn't create a whole new c directory for this. This should live somewhere else but I'm not sure where yet. Let me think about it.

Ok, no problem 👍

Ghostty's AppleScript API exposes working directory but not TTY. We cross-reference window CWDs against child shell process CWDs to resolve TTYs indirectly. Windows with duplicate CWDs get tiling but no status colors. Will switch to direct tty property once Ghostty ships it (ghostty-org/ghostty#11922). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* feat: add Ghostty to TERMINAL_APPS set Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: add Ghostty TTY resolution via CWD matching Ghostty's AppleScript API exposes working directory but not TTY. We cross-reference window CWDs against child shell process CWDs to resolve TTYs indirectly. Windows with duplicate CWDs get tiling but no status colors. Will switch to direct tty property once Ghostty ships it (ghostty-org/ghostty#11922). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: skip windows with unresolved TTY in tty map building Ghostty windows can have tty=None when CWD matching fails (e.g. duplicate CWDs). Guard against None TTYs flowing into _resolve_tty_cwd by checking info.get("tty") is truthy before storing in tty_map. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: use case-insensitive pgrep for Ghostty process name On macOS, Ghostty's process name is "Ghostty" (capital G) when launched from the .app bundle. Adding -i flag to pgrep ensures we find it regardless of case. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: add Ghostty to supported terminals in README Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: fix Ghostty AppleScript and process tree walking Two bugs found during live testing: 1. Ghostty windows don't have 'bounds' property — use System Events for position/size and Ghostty scripting API for working directory 2. ps -g doesn't find Ghostty child shells (different process groups) — use recursive pgrep -P to walk ghostty → login → shell tree Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: coryszatkowski <213997628+coryszatkowski@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

TweedBeetle · 2026-03-31T11:07:11Z

Happy to revert src/c/ and just import String from main_c.zig in embedded.zig if that's simpler for now. Or I can move it wherever you'd prefer. Just let me know 🫡

bo2themax

AppIntents/Shortcut.app builds their cache really weirdly, it took me some time to make it update to new entity, but it worked.

TweedBeetle · 2026-04-04T09:23:56Z

Hey @mitchellh, just wanted to check in on this. Happy to go whichever direction on the Zig file placement. As far as I understand, there are three options:

Revert src/c/ and import String from main_c.zig in embedded.zig
Move the file somewhere else you'd prefer
Keep current approach

Let me know if you have thoughts. Everything else is approved 🫡

TweedBeetle · 2026-04-08T08:56:08Z

Went ahead with option 1 to unblock: reverted src/c/ and inlined the String type back into main_c.zig (force-pushed). Zig diff is now a no-op aside from the two new CAPI exports in embedded.zig. Happy to move the extraction elsewhere if you'd still prefer that, just let me know and I'll reshape it.

Expose the foreground process PID and TTY device path as read-only properties on the AppleScript terminal class and App Intents TerminalEntity. This enables reliable process-to-terminal mapping for automation tools when multiple terminals share the same CWD. Closes ghostty-org#11592 Closes ghostty-org#10756 Session: 019d341c-a165-7843-a2f7-2f426114cf17

TweedBeetle requested a review from a team as a code owner March 28, 2026 11:55

ghostty-bot bot added the os/macos label Mar 28, 2026

jcollie requested changes Mar 28, 2026

View reviewed changes

mitchellh requested changes Mar 28, 2026

View reviewed changes

TweedBeetle force-pushed the pid-tty-applescript branch from 0fde537 to 82e66aa Compare March 28, 2026 16:28

mitchellh requested changes Mar 28, 2026

View reviewed changes

coryszatkowski mentioned this pull request Mar 29, 2026

Add Ghostty terminal support with TTY mapping coryszatkowski/ClawDeck#8

Merged

3 tasks

bo2themax approved these changes Mar 31, 2026

View reviewed changes

grimmerk mentioned this pull request Apr 3, 2026

Ghostty: no per-tab TTY/PID — missing cross-reference and TTY switch fallback grimmerk/codev#63

Open

TweedBeetle force-pushed the pid-tty-applescript branch from 82e66aa to d5e7965 Compare April 8, 2026 08:55

TweedBeetle force-pushed the pid-tty-applescript branch from d5e7965 to 58573a2 Compare April 9, 2026 17:53

	/// ghostty_string_s
	pub const String = extern struct {
	ptr: ?[*]const u8,
	len: usize,
	sentinel: bool,

	pub const empty: String = .{
	.ptr = null,
	.len = 0,
	.sentinel = false,
	};

	pub fn fromSlice(slice: anytype) String {
	return .{
	.ptr = slice.ptr,
	.len = slice.len,
	.sentinel = sentinel: {
	const info = @typeInfo(@TypeOf(slice));
	switch (info) {
	.pointer => \|p\| {
	if (p.size != .slice) @compileError("only slices supported");
	if (p.child != u8) @compileError("only u8 slices supported");
	const sentinel_ = p.sentinel();
	if (sentinel_) \|sentinel\| if (sentinel != 0) @compileError("only 0 is supported for sentinels");
	break :sentinel sentinel_ != null;
	},
	else => @compileError("only []const u8 and [:0]const u8"),
	}
	},
	};
	}

	pub fn deinit(self: *const String) void {
	const ptr = self.ptr orelse return;
	if (self.sentinel) {
	state.alloc.free(ptr[0..self.len :0]);
	} else {
	state.alloc.free(ptr[0..self.len]);
	}
	}
	};

Conversation

TweedBeetle commented Mar 28, 2026

Summary

Changes

Testing

AI Disclosure

Uh oh!

jcollie left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mitchellh left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

TweedBeetle commented Mar 31, 2026

Uh oh!

bo2themax left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

TweedBeetle commented Apr 4, 2026

Uh oh!

TweedBeetle commented Apr 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

bo2themax left a comment •

edited

Loading