Reported issues for glass Desktop Extension

Pod holds 21 of 103 GitHub reports that passed its relevance review. This can include external user reports, maintainer-confirmed bugs, and concrete feature gaps. Treat them as evidence to inspect, not a count of distinct defects.

Back to glass Desktop Extension.

Most discussed

windows: glass_stop returns Ok while Firefox keeps running

Reading (2026-08-24)

Windows 11, Firefox 154, crates/glass-windows/tests/web_probe.rs (fresh --profile, a11y: true):

glass.stop() returned Ok(()) and firefox.exe carrying the probe's profile marker was still running afterwards. Edge and Brave under the same probe left 0 survivors.
Same root as list_windows() returning empty for Firefox on this backend: the process glass discovered is not the one that owns the window, so the quit request and the group reap miss the real

Read the thread · 2026-08-24 · open · outside contributor · 3 comments

ios a11y: a WKWebView exposes no element through idb — web content is a blind spot with no tree signal

Reading (2026-08-24)

iOS 26.5 Simulator, idb_companion 1.1.8 and 1.5.0b3, examples/ios-role-fixture launched with --tab web (a stock WKWebView on examples/web-role-fixture/index.html, rendered in full — screenshot checked):

idb reports 2 nodes: AXApplication and AXGroup "Tab Bar". No AXWebArea, no page elements, and no element for the web view itself.
Unchanged from t+2 s to t+30 s and after two taps inside the page.
The Simulator's own Safari exposes browser c

Read the thread · 2026-08-24 · open · outside contributor · 3 comments

a11y-ios: a typed set_value intermittently does not land (~2 in 12), surfaced by the #359 fix

Measured on the mini against examples/ios-role-fixture — a UITextField in a static stack view — driving the real MCP surface from a granted GlassMcp.app. 12 consecutive runs of the pre-release smoke's interaction check:

tree at snapshot	runs	pass	fail
16 nodes (keyboard down)	9	8	1
18 nodes (keyboard up)	3	2	1

Both failures are the same verdict:

glass_set_value: set_value on element #N did not take — the element does not hold the r

[Read the thread](https://github.com/fixed-width/glass/issues/363) · 2026-08-07 · closed · outside contributor · 3 comments

### glass-ios: support multi-touch glass_gesture (pinch via idb HIDPinch; general N-finger needs validation)

Tracking only — not scheduled.

`glass_gesture` / `PointerEvent::Gesture` (N simultaneous straight finger paths) returns `Unsupported` on iOS (`crates/glass-ios/src/injector/mod.rs`, the `Gesture` arm). This matches every non-Android backend — x11, wayland, windows, and macos also return `Unsupported`; multi-touch currently requires Android's on-device companion.

idb_companion's HID gRPC (`crates/glass-ios/src/idb/client.rs`, `hid`) exposes:
- `HIDSwipe` — one finger, start→end (already used by

[Read the thread](https://github.com/fixed-width/glass/issues/117) · 2026-07-09 · open · outside contributor · 3 comments

### scroll_to_element: match hint-only editables when Android exposes the hint as value

## Problem

`glass_scroll_to_element` now accepts a `description` selector, but hint-only Android editables are not represented consistently across accessibility acquisition paths.

An off-screen `EditText` may expose its empty hint as `value="Search settings"` rather than `description="Search settings"`. A caller can express the expected semantic label through `description`, but the scroll sweep never matches it and eventually returns `matched:false`. Generic scrolling and repeated snapshots ar

[Read the thread](https://github.com/fixed-width/glass/issues/544) · 2026-08-27 · closed · outside contributor · 2 comments

### x11: the doctor's attach probe has no bound, so a server that never completes the handshake hangs glass doctor forever

`attach_verdict` (`crates/glass-x11/src/doctor.rs:174`) connects with no timeout, no thread and no
budget:

```rust
x11rb::connect(Some(display)).map(|_| ()).map_err(|e| no_attach(&e))

An X server that accepts the connection and never sends the setup reply blocks it forever. Every other probe in the doctor is bounded — that is what #392 (sway) and #398 (bwrap) established, and what #455 has just done for the a11y bus — and this is the one left.

Demonstrated. A listener bound to `/tmp/.

Read the thread · 2026-08-18 · closed · outside contributor · 2 comments

ios+android: stop_app reports success when the app was not stopped, so glass_stop and the audit log record a stop that may not have happened

Both device backends discard the result of the call that actually stops the app, then return Ok(()):

glass-ios/src/platform.rs — let _ = self.target.simctl().run(&["terminate", udid, bundle])
glass-android/src/platform.rs — the same shape around am force-stop

The comment on each justifies exactly one case: the app may already be gone, because it crashed or because a prior launch's --terminate-running-process killed it. That case is real and benign. The discard is not limited to

Read the thread · 2026-08-10 · open · outside contributor · 2 comments

a11y-ios: set_value verifies by pre-order id, so a landed write into a relayouting field always errors

On the iOS Simulator, glass_set_value re-resolves its target by pre-order id to verify the write. Any app that re-lays-out in response to the input renumbers the tree, so the read-back compares against whatever now occupies that id and refuses — for a write that actually landed.

Measured, not theorized: 3 of 3 runs against stock Settings, driving the real MCP surface via a granted GlassMcp.app on the mini.

Evidence

Typing glassprobe7391 into Settings' search field:

BEFOR

[Read the thread](https://github.com/fixed-width/glass/issues/359) · 2026-08-06 · closed · outside contributor · 2 comments

## Most recent

### android a11y: distinguish displayed hints from entered editable values

## Problem

With the Android accessibility companion enabled, a hint-only empty `EditText` exposes the hint separately, so Glass correctly maps it to `description`. Android also publishes the displayed hint through the node's text, and Glass currently maps that to `value`.

The role fixture therefore renders the untouched field as:

```text
#35 TextField desc="Search settings" value="Search settings"

After entry, it correctly renders:

#35 TextField desc="Search settings" value="hea

[Read the thread](https://github.com/fixed-width/glass/issues/547) · 2026-08-27 · closed · outside contributor · 0 comments

### docs(mcp): document the Android glass_start run tuple in the tool surface

## Problem

The `glass_start` MCP schema describes `run` as a command array, but does not make the Android interpretation discoverable: the first entry is an APK path and the second entry is an exact package/activity component.

A caller that needs to install an APK and launch a specific activity cannot infer the accepted shape or whether the component should use `package/.Activity`, a fully qualified class, or another form. It must search documentation or implementation before making the first 

[Read the thread](https://github.com/fixed-width/glass/issues/543) · 2026-08-27 · closed · outside contributor · 1 comment

### android: confirm set_value after IME relayout when the editable loses its hint

## Problem

`glass_set_value` can deliver text successfully but still fail confirmation when focusing an Android `EditText` opens the IME and changes both its geometry and exposed semantic fields.

The post-focus node may retain the editable role and requested value while no longer exposing the hint that identified the original empty field. Reacquisition then rejects the moved node as potentially different, even when it is the only compatible focused editable in the active window.

## Reproducti

[Read the thread](https://github.com/fixed-width/glass/issues/545) · 2026-08-27 · closed · outside contributor · 1 comment

### a11y snapshot: include bounded editable values in compact output

## Problem

Compact accessibility snapshots can omit the current value of editable controls even when Glass has that value and can return it through `glass_wait_for_element`.

A caller sees that a node is an editable `TextField`, but cannot directly determine its current text. Exact runtime verification requires discovering and issuing a separate wait query.

## Reproduction

Using `examples/android-role-fixture`:

1. Set the hint-only `TextField` to `expected-value`.
2. Call `glass_a11y_snapsho

[Read the thread](https://github.com/fixed-width/glass/issues/534) · 2026-08-27 · closed · outside contributor · 1 comment

### mcp: make semantic, visual, and stability verification paths discoverable

## Problem

Glass exposes several runtime-observation and verification tools, but their intended division of responsibility is not obvious from the MCP surface:

- accessibility snapshot for current semantic state;
- screenshot for current visual state;
- wait-for-element for semantic conditions and values;
- wait-for-region and diff for pixel conditions;
- wait-stable for quiescence.

A caller can inspect an editable node without seeing its value, then take more snapshots or screenshots instead

[Read the thread](https://github.com/fixed-width/glass/issues/535) · 2026-08-27 · closed · outside contributor · 1 comment

### glass_do: add semantic actions, bounded execution, and per-step results

## Existing behavior

`glass_do` already provides the core batching primitive: it runs an ordered, non-empty sequence server-side, fails fast with the failed action index, and can perform a terminal `settle`, `diff`, or `screenshot` observation.

Supported actions are currently:

- `click`
- `move`
- `drag`
- `scroll`
- `type`
- `key`
- `settle`

It reuses the standalone coordinate/input implementations, records sub-actions in the audit log, and correctly does not claim rollback after partial GU

[Read the thread](https://github.com/fixed-width/glass/issues/536) · 2026-08-27 · closed · outside contributor · 1 comment

### android: reacquire editable targets across IME-driven layout changes

## Problem

On Android, focusing an editable control can open the IME, resize or scroll the activity, and replace the accessibility node between dispatch and read-back. `glass_set_value` then fails confirmation and requires the caller to take another snapshot, find the replacement node, and retry the same write.

This is a normal consequence of focusing a field, not necessarily application interference. The operation should retain enough semantic identity to reacquire the control and finish conf

[Read the thread](https://github.com/fixed-width/glass/issues/533) · 2026-08-27 · closed · outside contributor · 1 comment

### scroll_to_element: support description selectors for unnamed controls

## Problem

`glass_scroll_to_element` matches by `name` and `role`, but some controls are unnamed and identifiable only through another accessibility property such as `description`.

On Android, an `EditText` whose only label is a hint can arrive as an unnamed `TextField` with `desc="Search settings"`. If it begins off-screen, the caller cannot ask `glass_scroll_to_element` to find it by that description. It must scroll generically and repeatedly inspect the tree.

## Reproduction

Using `exampl

[Read the thread](https://github.com/fixed-width/glass/issues/537) · 2026-08-27 · closed · outside contributor · 1 comment

### core: a write whose read-back was unavailable renders as "did not take"

## Read from the code

`crates/glass-core/src/error.rs` (the `AxValueNotApplied` Display, ~:170-177): with `observed: None` the message opens "set_value on element #N did not take" and goes on to explain a value it never read. All three desktop readers can produce that case — a widget exposing a writable interface but no readable one (on Linux, `EditableText` without `Text`, now that `set_value` confirms by read-back in #517).

`AxWriteUnconfirmed` plus `read_back_failed` (`crates/glass-core/src

[Read the thread](https://github.com/fixed-width/glass/issues/522) · 2026-08-24 · closed · outside contributor · 1 comment

### linux a11y: an <iframe>'s AT-SPI internal frame maps to Window inside the page

## Reading (2026-08-24)

Firefox 153 on X11 and Wayland, `examples/web-role-fixture/index.html` through `crates/glass-a11y-linux/tests/web_probe.rs`:

- The page's `<iframe>` container arrives as AT-SPI `internal frame` (4 per page), which the pre-existing map (`crates/glass-a11y-linux/src/mapping.rs`: `Role::Frame | Role::Window | Role::InternalFrame => AxRole::Window`) renders as a **`Window` node inside the `Document`**.
- The nested `document web` → `Document` sits inside it, so the iframe's

[Read the thread](https://github.com/fixed-width/glass/issues/520) · 2026-08-24 · closed · outside contributor · 1 comment

### deps: rmcp 3.x / MCP 2026-07-28 — the HTTP session gate is built on what becomes the legacy path

glass is on `rmcp 2.2.0`. Both rmcp 2.2 and 3.x still define `ProtocolVersion::LATEST` as `2025-11-25`; rmcp 3.x adds full support for the newer MCP `2026-07-28` protocol rather than making it the default. That revision rewrites much of the lifecycle and transport plumbing, and one change lands directly on code glass owns.

## What changes underneath

For clients that negotiate `2026-07-28`, the modern lifecycle removes protocol-level HTTP sessions: there is no `Mcp-Session-Id` header and no `in

[Read the thread](https://github.com/fixed-width/glass/issues/511) · 2026-08-22 · open · outside contributor · 1 comment

### mcp: the initialize handshake omits the description, websiteUrl and icons that server.json already carries

`get_info` sets only `name`, `title` and `version` on the `initialize` handshake's
`Implementation` (`crates/glass-mcp/src/server.rs:662-672`). The struct carries
three more optional fields that rmcp 2.2 already models — `description`,
`website_url` and `icons` (the last two added to the spec in 2025-11-25, SEP-973,
to align the handshake with the registry's `server.json` shape).

glass already publishes two of those values in its own `server.json`:

"description": "MCP server giving a codin

Read the thread · 2026-08-22 · closed · outside contributor · 0 comments

mcp: tools carry no annotations, so a client cannot tell an observation from an input action

Every tool glass registers advertises annotations: None. get_info declares ServerCapabilities::builder().enable_tools() and nothing else (crates/glass-mcp/src/server.rs:662), and no #[tool(...)] attribute in server.rs carries an annotations(...) clause.

MCP tool annotations — readOnlyHint, destructiveHint, idempotentHint, openWorldHint (in the spec since 2025-03-26) — are what a client reads to decide whether a call can be auto-approved or has to be put in front of the use

Read the thread · 2026-08-22 · closed · outside contributor · 0 comments

The remaining reports are on the project's issue tracker.