chore(cua): sync driver v0.1.4

docs/issues/cua-driver-v014-sync/plan.md

@@ -0,0 +1,25 @@
+# CUA Driver v0.1.4 Sync Plan
+
+## Approach
+
+Apply the upstream `libs/cua-driver` delta from `cua-driver-v0.0.15` to
+`cua-driver-v0.1.4`, then re-apply DeepChat fork patches where conflicts overlap.
+Keep the DeepChat plugin MCP server configured as the normal `cua-driver` server.
+
+## Compatibility
+
+- Remove removed upstream tools from plugin policy and manifest:
+  `get_accessibility_tree`, `type_text_chars`.
+- Keep `type_text` approved for text input and document its `delay_ms` fallback
+  path in local skills.
+- Preserve DeepChat TCC ownership and app bundle identity throughout the driver,
+  helper build scripts, diagnostics, and permission flows.
+- Preserve the DeepChat-owned update path; the bundled helper must not install
+  standalone upstream releases.
+
+## Validation
+
+- Build the Swift package from the vendored driver source.
+- Run focused plugin presenter and packaging/signing tests.
+- Run repository formatting, i18n, and lint checks after implementation.
+- Inspect protected files for accidental skill or fork-patch regressions.

docs/issues/cua-driver-v014-sync/spec.md

@@ -0,0 +1,29 @@
+# CUA Driver v0.1.4 Sync Spec
+
+## Goal
+
+Sync the vendored DeepChat CUA driver fork with upstream `trycua/cua`
+`cua-driver-v0.1.4` while preserving DeepChat-specific runtime behavior and local
+skill customizations.
+
+## Acceptance Criteria
+
+- Vendored upstream metadata points to tag `cua-driver-v0.1.4` and commit
+  `d422294b848afec99b979ac1229446c83fa44807`.
+- Removed upstream tools `get_accessibility_tree` and `type_text_chars` are no
+  longer advertised by the DeepChat plugin manifest or policy.
+- `type_text` remains the text-entry tool and supports upstream automatic CGEvent
+  fallback behavior.
+- DeepChat fork patches remain intact:
+  `DeepChatPermissionProbeCommand`, `DeepChat Computer Use.app`,
+  `com.wefonk.deepchat.computeruse`, non-blocking MCP/daemon startup,
+  DeepChat-managed update messaging, and window-scoped zoom contexts.
+- DeepChat-owned plugin skill files receive only minimal compatibility edits for
+  removed tool names.
+
+## Non-Goals
+
+- Do not expose a second Claude Code compatibility MCP server in the DeepChat plugin.
+- Do not rewrite or replace local DeepChat skill content with upstream skill text.
+- Do not change user-facing plugin UX beyond the upstream driver compatibility
+  surface required for this sync.

docs/issues/cua-driver-v014-sync/tasks.md

@@ -0,0 +1,9 @@
+# CUA Driver v0.1.4 Sync Tasks
+
+- [x] Apply upstream driver source changes through `cua-driver-v0.1.4`.
+- [x] Resolve `CuaDriverCommand.swift` conflicts while keeping DeepChat commands
+  and startup behavior.
+- [x] Update upstream metadata, plugin manifest, and tool policy.
+- [x] Patch local DeepChat skill files only for removed tool compatibility.
+- [x] Verify protected fork strings and commands remain present.
+- [x] Run Swift build, focused Vitest suites, format, i18n, and lint.

plugins/cua/plugin.json

@@ -77,7 +77,6 @@
         "list_windows": "allow",
         "get_screen_size": "allow",
         "get_window_state": "allow",
-        "get_accessibility_tree": "allow",
         "get_cursor_position": "allow",
         "get_config": "allow",
         "get_recording_state": "allow",
@@ -91,7 +90,6 @@
         "scroll": "ask",
         "move_cursor": "ask",
         "type_text": "ask",
-        "type_text_chars": "ask",
         "press_key": "ask",
         "hotkey": "ask",
         "set_value": "ask",

plugins/cua/policies/tool-policy.json

@@ -6,7 +6,6 @@
     "list_windows": "allow",
     "get_screen_size": "allow",
     "get_window_state": "allow",
-    "get_accessibility_tree": "allow",
     "get_cursor_position": "allow",
     "get_config": "allow",
     "get_recording_state": "allow",
@@ -20,7 +19,6 @@
     "scroll": "ask",
     "move_cursor": "ask",
     "type_text": "ask",
-    "type_text_chars": "ask",
     "press_key": "ask",
     "hotkey": "ask",
     "set_value": "ask",

plugins/cua/skills/cua-driver/SKILL.md

@@ -25,7 +25,7 @@ Use DeepChat's CUA plugin MCP tools for macOS app automation. Treat the tools ex
 2. Start or reuse the target with `launch_app({ bundle_id })`. Use the returned `pid` when available.
 3. Inspect windows with `list_windows({ pid })` when the launch result lacks a usable window.
 4. Snapshot before every UI action with `get_window_state({ pid, window_id })`.
-5. Act with the matching MCP tool: `click`, `right_click`, `double_click`, `drag`, `scroll`, `type_text`, `type_text_chars`, `press_key`, `hotkey`, `set_value`, `page`, or `launch_app` with `urls`.
+5. Act with the matching MCP tool: `click`, `right_click`, `double_click`, `drag`, `scroll`, `type_text`, `press_key`, `hotkey`, `set_value`, `page`, or `launch_app` with `urls`. For web inputs that reject AX text insertion, call `type_text({ pid, text, delay_ms })` and let the driver use its CGEvent fallback.
 6. Snapshot again after each action and verify visible evidence: selected state, changed text, playback progress, new panels, highlighted rows, or updated window content.
 
 Element indices come from the latest `get_window_state` result for the same `pid` and `window_id`. Re-snapshot when an index is missing, stale, or from another window.

plugins/cua/vendor/cua-driver/source/.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.0.15
+current_version = 0.1.4
 commit = True
 tag = True
 tag_name = cua-driver-v{new_version}

plugins/cua/vendor/cua-driver/source/README.md

@@ -3,3 +3,21 @@
 Background computer-use driver for any agents. Speaks MCP over stdio; drives native macOS apps without stealing focus.
 
 **[Documentation](https://cua.ai/docs/cua-driver)** - Installation, guides, and API reference.
+
+## Claude Code computer-use compatibility
+
+Standard Claude Code MCP registration:
+
+```bash
+claude mcp add --transport stdio cua-driver -- cua-driver mcp
+```
+
+If you want Claude Code's vision/computer-use-style flow to ground on CuaDriver window screenshots, register the compatibility mode:
+
+```bash
+claude mcp add --transport stdio cua-computer-use -- cua-driver mcp --claude-code-computer-use-compat
+```
+
+This keeps CuaDriver's normal MCP tools and changes only `screenshot`, which requires `pid` and `window_id` and captures that window only.
+
+Use MCP for this Claude Code vision/computer-use-style path. CLI screenshots still work as CuaDriver calls, but they do not expose the `mcp__cua-computer-use__screenshot` tool name that Claude Code appears to use as the image-grounding cue.

plugins/cua/vendor/cua-driver/source/Skills/cua-driver/RECORDING.md

@@ -6,7 +6,7 @@ user explicitly asks to record — the skill does not auto-enable this.
 
 `set_recording` turns on a session-scoped trajectory recorder. While
 enabled, every action-tool call (`click`, `right_click`, `scroll`,
-`type_text`, `type_text_chars`, `press_key`, `hotkey`, `set_value`)
+`type_text`, `press_key`, `hotkey`, `set_value`)
 writes a numbered turn folder under a caller-chosen output
 directory. Read-only tools (`get_window_state`, `list_windows`,
 `screenshot`, `list_apps`, permission probes, agent-cursor getters /
@@ -101,7 +101,7 @@ keyed on `(pid, window_id)`, so a recorded
 resolve today — the pid is usually different, the window_id always
 is. The call returns `Invalid element_index` or `No cached AX
 state`. Pixel clicks (`click({pid, x, y})`) and keyboard tools
-(`press_key`, `type_text_chars`, `hotkey`, `type_text` without
+(`press_key`, `hotkey`, `type_text` without
 element_index) replay cleanly; element-indexed actions require a
 live snapshot that replay doesn't currently re-emit (read-only tools
 like `get_window_state` aren't recorded). For a reliable replay, either

plugins/cua/vendor/cua-driver/source/Skills/cua-driver/SKILL.md

@@ -477,8 +477,8 @@ you're interacting with a long-lived process). In the default
 get_window_state → reason over PNG → pixel click`. When you need
 `element_index` dispatch (AX-addressable elements, backgrounded
 clicks), flip to `som` first: `cua-driver set_config '{"key":
-"capture_mode", "value": "som"}'`, or call `get_accessibility_tree`
-directly. The rest of this section walks through `som` mode, which
+"capture_mode", "value": "som"}'`, then call `get_window_state`
+again. The rest of this section walks through `som` mode, which
 is what you want once you've decided element-indexed addressing is
 required.
 
@@ -573,7 +573,7 @@ anchor the conversion against a specific window):
 | Focus + send key | `press_key({pid, key, window_id, element_index, modifiers})` | element_index sets AXFocused, then posts key |
 | Send key to pid | `press_key({pid, key, modifiers})` | no focus change; key goes to pid's current focus |
 | Modifier combo | `hotkey({pid, keys})` | e.g. `["cmd","c"]`; posted per-pid, not HID tap |
-| Unicode keystrokes | `type_text_chars({pid, text, delay_ms})` | CGEvent-to-pid; reaches Chromium/Electron inputs |
+| Unicode keystrokes | `type_text({pid, text, delay_ms})` | AX insertion with CGEvent-to-pid fallback; reaches Chromium/Electron inputs |
 
 **All keyboard/text primitives require `pid`.** There is no
 frontmost-routed variant — every key goes to the named target via

plugins/cua/vendor/cua-driver/source/Skills/cua-driver/TESTS.md

@@ -116,11 +116,11 @@ These are where CuaDriver's AX-activation trio matters:
 ### 8. Slack — sparse Electron, hotkey fallback
 **Prompt:** `In Slack, jump to the #pr-reviews channel using the quick switcher (⌘K).`
 
-**Exercises:** "retry snapshot once, then `hotkey` + `type_text_chars`" fallback path.
+**Exercises:** "retry snapshot once, then `hotkey` + `type_text` with `delay_ms`" fallback path.
 
 **Success:**
 - Re-snapshot shows Slack's channel title area (AX role `AXStaticText` or similar) contains `pr-reviews`.
-- Claude uses `hotkey` and `type_text_chars` — NOT `simulate_click` / pixel coords.
+- Claude uses `hotkey` and `type_text` with `delay_ms` — NOT `simulate_click` / pixel coords.
 
 **Fail signals:** Claude drops to `simulate_click` (guardrail violation — pixel fallback is not allowed on sparse AX trees), wrong channel joined, typed text echoes into the message composer instead of the switcher.
 
@@ -168,7 +168,7 @@ These are where CuaDriver's AX-activation trio matters:
 ### 12. Chrome proper — omnibox
 **Prompt:** `In Google Chrome, open a new tab and navigate to https://trycua.com.`
 
-**Exercises:** `hotkey(["cmd","t"])` + `type_text_chars` + Return. Chrome's omnibox typically isn't AX-exposed even after activation.
+**Exercises:** `hotkey(["cmd","t"])` + `type_text({pid, text, delay_ms})` + Return. Chrome's omnibox typically isn't AX-exposed even after activation.
 
 **Success:**
 - A new Chrome tab whose AX title contains `Cua` or `trycua` is present.

plugins/cua/vendor/cua-driver/source/Skills/cua-driver/WEB_APPS.md

@@ -46,10 +46,9 @@ pixels:
    `hotkey({pid, keys: ["cmd", "enter"]})`, `hotkey({pid, keys:
    ["cmd", "k"]})`, etc. Posted via `CGEvent.postToPid`, reaches the
    target regardless of AX state, no activation required.
-3. For typing into web inputs where `type_text` silently drops
-   (input doesn't implement `AXSelectedText`), use `type_text_chars`
-   — pure CGEvent keystrokes reach any focused keyboard receiver,
-   including Unicode / emoji.
+3. For typing into web inputs where AX insertion is rejected or silently
+   dropped, use `type_text` with a small `delay_ms` so the driver can use
+   its CGEvent fallback, including Unicode / emoji.
 4. If none of the above reaches the target, tell the user this
    interaction isn't reachable from the driver today and ask for
    guidance.
@@ -86,7 +85,7 @@ documented only as historical context:
 ```
 # DON'T DO THIS — ⌘L steals focus. Use launch_app above.
 hotkey({pid, keys: ["cmd", "l"]})
-type_text_chars({pid, text: "https://cua.ai", delay_ms: 30})
+type_text({pid, text: "https://cua.ai", delay_ms: 30})
 get_window_state({pid, window_id})
 click({pid, window_id, element_index: <suggestion>})
 ```
@@ -157,7 +156,7 @@ When the target window is **minimized** (genie'd into the Dock):
   `AXFocused=true` on a minimized window's descendants doesn't
   propagate to real keyboard focus). Symptom: macOS system-alert
   beep, or silent no-op. Example: `hotkey cmd+L` +
-  `type_text_chars URL` + `press_key return` on minimized Chrome —
+  `type_text({pid, text: URL, delay_ms: 30})` + `press_key return` on minimized Chrome —
   the URL lands in the omnibox AX value but Return doesn't commit
   the navigation.
 - **Primary workaround — use `set_value` to commit directly**: For
@@ -464,7 +463,7 @@ Playwright / a WebExtension with Native Messaging for Firefox.
 type_text({pid, window_id, element_index: <input_field>, text: "…"})
 ```
 
-If it silently drops (some web inputs don't implement
+If AX insertion silently drops (some web inputs don't implement
 `AXSelectedText`), click the field first, then use
-`type_text_chars({pid, text})` — pure CGEvent keystrokes delivered
-to the pid, reaching any focused keyboard receiver.
+`type_text({pid, text, delay_ms})` so the driver falls back to
+pid-targeted CGEvent keystrokes.