test: consolidate log_session_env BATS coverage (#548)

Collapse the four log_session_env cases into two, and tighten the
assertions in both:

  - Old test 1 (substring match per key) + old test 4 (block braces
    on their own lines) → one test using exact-line equality on the
    `lines` array. Locks block structure and per-key formatting in
    a single pass; substring matching could not catch a regression
    that re-ordered keys, dropped indentation, or merged lines.
  - Old test 2 (unset values are KEY=) + old test 3 (empty-string
    is KEY=) → one test covering both code paths. Exact-line match
    proves the value after `=` is truly empty; the previous
    `*'KEY='*` substring would have matched `KEY=value` and the
    old test-3 regex was fragile (depended on trailing newline
    being literal `$'\n'` vs end-of-string `$`).

Net: 77 → 42 lines, 4 → 2 cases, stronger guarantees. No change to
the helper itself or the call sites — issue #548 acceptance criteria
still hold.

.github/workflows/ci.yml

@@ -674,6 +674,7 @@ jobs:
             'gpgcheck=1' \
             'repo_gpgcheck=1' \
             'gpgkey=https://pkg.claude-desktop-debian.dev/KEY.gpg' \
+            'metadata_expire=1h' \
             > rpm/claude-desktop.repo
 
       - name: Re-upload signed RPMs to GitHub Release

README.md

@@ -259,6 +259,7 @@ Special thanks to:
 - **[zabka](https://github.com/zabka)** for identifying that `cowork-vm-service.js` was never auto-spawned on Linux and contributing a systemd-unit workaround that scoped the daemon auto-launch fix (#445)
 - **[sirfaber](https://github.com/sirfaber)** for fixing the `$`-in-minified-identifier breakage of cowork Patch 2b (vm module assignment) and Patch 6 step 2 (retry-delay auto-launch) on Claude Desktop 1.5354.0 (#555)
 - **[ProfFlow](https://github.com/ProfFlow)** for re-fixing the RPM repodata signing regression by appending `!` to the keyid passed to `gpg --default-key`, forcing `repomd.xml` to be signed by the primary key instead of the auto-selected signing subkey (#566)
+- **[jslatten](https://github.com/jslatten)** for fixing the KDE Plasma Wayland launcher-grouping bug by setting `pkg.desktopName` in the packaged `app.asar`'s `package.json`, format-conditional so deb/rpm get `claude-desktop.desktop` and AppImage gets `io.github.aaddrick.claude-desktop-debian.desktop` (#562)
 
 ## Sponsorship
 

docs/learnings/mcp-double-spawn.md

@@ -37,46 +37,67 @@ external services with single-connection contracts, etc.
 
 ## Root Cause (Upstream)
 
-Two parallel session managers live inside Electron main, each
-holding an independent Claude Agent SDK `query`:
+Multiple session managers live inside Electron main, each
+holding its own MCP coordinator state with its own registry. The
+two that spawn stdio MCPs from `claude_desktop_config.json` and
+trigger this bug:
 
 | Manager class            | IPC namespace                            | Coordinator     | Logs prefix |
 |--------------------------|------------------------------------------|-----------------|-------------|
 | `LocalSessions`          | `claude.web_$_LocalSessions_$_*`         | `n2t("ccd")`    | `[CCD]`     |
 | `LocalAgentModeSessions` | `claude.web_$_LocalAgentModeSessions_$_*`| `n2t("cowork")` | `[LAM]`     |
 
+A third coordinator class — `SshMcpServerManager` — follows the
+same per-coordinator-registry pattern but uses an SSH transport
+and doesn't contribute to the local-node double-spawn. Its
+existence does say something about the design intent: per-
+coordinator isolated state appears to be a deliberate
+architectural pattern, not a one-off oversight.
+
 The logs prefixes are what to grep `~/.config/Claude/logs/` for to
 confirm a session is hitting both coordinators (and therefore this
 bug specifically).
 
-Each `query` holds its own SDK transport. The transport's
-`spawnLocalProcess` (`Du.spawn`) launches stdio MCPs **without
-consulting the global registry** that *would* dedupe them
-(`hZ` map, accessed via `oUt(serverName)` /
-`launchMcpServer`). That registry is only used for the
-"internal" cowork in-process MessageChannelMain path.
+Each coordinator dedups **within its own scope**: CCD's launch
+function serializes per server name through a promise queue and
+shuts down any prior entry before respawn; LAM's
+`getOrCreateConnection` reuses connected entries from its own
+`connections` Map. The double-spawn is strictly **cross-
+coordinator** — one process per coordinator that has the server
+in its config.
+
+In current versions (verified against `1.5354.0`) both
+coordinators route their transport creation through a shared
+Claude Desktop-side factory, but the factory itself doesn't
+dedupe and the per-coordinator registries above it aren't
+unified.
 
 Net result: 2 coordinators × N configured MCPs = 2N processes.
 
-Symbol names (`n2t`, `hZ`, `oUt`, `LocalSessions`,
-`LocalAgentModeSessions`) are minified and **will rename across
-upstream releases**.
+### Symbol drift
+
+Minified symbols rename across upstream releases. Issue
+[#546](https://github.com/aaddrick/claude-desktop-debian/issues/546)
+maintains the current symbol mappings (verified against
+`1.5354.0`) plus extraction regexes that work against both
+minified and beautified bundles.
 
 ## Status
 
-**Upstream Claude Desktop bug. Not patchable in this repo.** A
-fix would require either:
+**Upstream Claude Desktop bug. Not patchable in this repo.** The
+proximate cause is in Claude Desktop's session manager wiring. A
+real fix needs either:
 
-- Routing the SDK stdio transport through `oUt`/`hZ` (the
-  existing serialized-per-name registry), or
-- Sharing one MCP-server registry between the `ccd` and
-  `cowork` coordinators.
+- LAM proxying its MCP traffic through CCD's existing connection
+  (so only one coordinator owns the spawn), or
+- A multiplexing wrapper transport that lets one spawned stdio
+  child serve multiple SDK clients via demuxing.
 
-Both live inside the closed-source SDK transport / session
-manager wiring. Regex-matching the minified symbols from
-`scripts/patches/` would be fragile against release-to-release
-renames and exceeds this repo's "minimal Linux-compat patches
-only" charter.
+Stdio MCP is 1:1 at the protocol layer — one stdin/stdout pair,
+one transport, one SDK client. Sharing one process across
+coordinators requires real engineering, not a sed patch on
+minified code, and exceeds this repo's "minimal Linux-compat
+patches only" charter.
 
 ## What's Already Verified Clean
 
@@ -118,13 +139,15 @@ The reporter's `baro-voyager` MCP shipped both in commit
 
 - **Primary:** in-app feedback (Help → Send Feedback) or
   `support@anthropic.com`. The duplication happens in
-  closed-source Desktop main.
-- **Secondary:** an SDK-transport-flavored issue on
+  closed-source Desktop main, in the per-coordinator registry
+  wiring.
+- **Secondary:** an issue on
   [`anthropics/claude-agent-sdk-typescript`](https://github.com/anthropics/claude-agent-sdk-typescript)
-  is defensible — the spawn path goes through the **Claude Agent
-  SDK's** `query` transport (`spawnLocalProcess` / `Du.spawn`),
-  which is shared surface area. Reference the missing `hZ`
-  consultation explicitly.
+  is defensible only if it advocates for a shared-transport /
+  multiplex primitive that would make this kind of bug
+  structurally harder. The SDK's spawn implementation is doing
+  what it's told — the bug is one layer up, in Claude Desktop
+  calling spawn from two separate coordinators.
 
 The embedded Claude Code CLI subprocess inside Claude Desktop is
 **not** the cause — it receives `--mcp-config` only when the

docs/upstream-reports/546-mcp-double-spawn.md

@@ -0,0 +1,164 @@
+# Upstream report draft: MCP double-spawn (issue #546)
+
+This is the draft for the upstream bug report covering [#546](https://github.com/aaddrick/claude-desktop-debian/issues/546). Filing target is `anthropics/claude-code` GitHub Issues, with an in-app `/bug` from Claude Desktop as a complement so the report ties to build telemetry.
+
+## Template mismatch note
+
+The `anthropics/claude-code` bug template is built for the Claude Code CLI, not Claude Desktop. Required fields like "Claude Code Version" and "Terminal/Shell" don't apply cleanly. Other Claude Desktop bug reports in the same repo work around this by putting `N/A — Claude Desktop <version>` in the version field and selecting `Other` for terminal (see #43705, #36319, #14807).
+
+## Title
+
+```
+[BUG] Claude Desktop 1.5354.0: stdio MCP servers double-spawn from independent CCD/LAM coordinator registries
+```
+
+## Form fields
+
+### Preflight Checklist
+
+- [x] I have searched existing issues and this hasn't been reported yet
+- [x] This is a single bug report
+- [x] I am using the latest version of Claude Code
+
+### What's Wrong?
+
+I maintain [claude-desktop-debian](https://github.com/aaddrick/claude-desktop-debian) (~2,300 package downloads/day across the last 3 releases), which repackages the Windows Electron build for Linux. I was reading the MCP spawn path in 1.5354.0 and found that stdio MCP servers configured in `claude_desktop_config.json` get spawned twice when both the chat panel and Code/Agent panel are active.
+
+The user-visible symptom is two `node` processes per MCP, both children of the Electron main PID. Killing one disconnects one panel and the other keeps working. They're independent client/server pairs with no failover between them.
+
+The original symptom report came from @communitytranslations against an earlier build (tracked in our repo as #526). I went back and read the bundle to confirm the cause. What I found was different from what we'd previously documented.
+
+CCD wraps the spawn path in a per-key promise queue keyed by server name. It shuts down any prior entry in its global registry Map before respawning. That's correct dedup within CCD. But LAM (`LocalMcpServerManager`) has its own `this.connections` Map and its own `getOrCreateConnection` path. It never consults CCD's registry.
+
+CCD and LAM each maintain independent spawn lifecycle management. They each spawn their own copy of the same MCP server. The double-spawn is structural in the current architecture. Each coordinator legitimately holds its own connection.
+
+There's also a third coordinator class, `SshMcpServerManager`, that follows the same per-coordinator-registry pattern. It uses an SSH transport, so it doesn't contribute to local-node double-spawn directly. Its existence suggests per-coordinator isolated state is a deliberate pattern, not a one-off.
+
+Secondary bug worth flagging while you're in this code. The `child_process.spawn` wrapper does proper signal escalation (end stdin, wait 2s, SIGTERM, wait 2s, SIGKILL). The `utilityProcess.fork` wrapper doesn't. It sends `process.kill()` (default SIGTERM), waits 5s, then calls `kill()` again with the same default signal. No SIGKILL escalation. A built-in-node MCP server that ignores SIGTERM could leak as an orphaned utility process.
+
+### What Should Happen?
+
+One process per stdio MCP server entry in `claude_desktop_config.json`, regardless of how many panels are open. Resource-side that means no more 2x memory and 2x stdin/stdout traffic per server. User-side that means `ps` shows one entry per declared server.
+
+The fix is architectural. CCD and LAM share a registry, or the local-spawn factory dedups at the transport layer, or LAM proxies through CCD when running in-process. Any of those would collapse the duplication.
+
+### Error Messages/Logs
+
+The user-facing log prefixes are stable across releases. Grep `~/.config/Claude/logs/` for:
+
+```
+[CCD]
+[LAM]
+[LocalMcpServerManager]
+[SshMcpServerManager]
+```
+
+For the spawn lifecycle specifically, look for:
+
+```
+"Launching MCP Server: <name>"      (CCD spawn entry)
+"Shutting down MCP Server: <name>"  (CCD shutdown entry)
+"local-mcp-server-cleanup"          (LAM cleanup path)
+```
+
+Two of these per declared MCP server is the diagnostic signal.
+
+### Steps to Reproduce
+
+1. Linux host running Claude Desktop at or near 1.5354.0
+2. Declare at least one stdio MCP server in `~/.config/Claude/claude_desktop_config.json`
+3. Open Claude Desktop, start a session, open the Code/Agent panel and let it initialize fully (the original report waited about 5 minutes)
+4. `ps -ef | grep <server-binary-name>`
+
+Expected: 1 process per MCP. Actual: 2 processes per MCP, both children of the same Electron main PID.
+
+### Claude Model
+
+Not sure / Multiple models
+
+### Is this a regression?
+
+I don't know
+
+### Last Working Version
+
+(leave blank)
+
+### Claude Code Version
+
+```
+N/A — this is a Claude Desktop issue. Bundle version: 1.5354.0
+```
+
+### Platform
+
+Anthropic API
+
+### Operating System
+
+Ubuntu/Debian Linux
+
+### Terminal/Shell
+
+Other
+
+### Additional Information
+
+Bundle reference table for 1.5354.0. Symbols rename across releases, so each row has a stable string anchor for re-finding them.
+
+| Role | Symbol in 1.5354.0 | Stable anchor |
+|---|---|---|
+| CCD spawn function | `BPt` | `"Launching MCP Server:"` |
+| CCD shutdown function | `CPt` | `"Shutting down MCP Server:"` |
+| CCD per-key promise queue | `dPt` | called by CCD spawn fn: `await dPt(e, async () => {...})` |
+| CCD server registry Map | `xX` | `.get()` immediately preceding the CCD shutdown log line |
+| Shared transport factory | `oPt` | `"built-in-node"` literal in factory body |
+| LAM manager class | `p0A` | `"[LocalMcpServerManager]"` or `"local-mcp-server-cleanup"` |
+| SSH manager class | `Rde` | `"[SshMcpServerManager]"` or `"ssh-mcp-server-cleanup"` |
+| `utilityProcess.fork` wrapper | `mFr` | constructed in shared factory's `built-in-node` branch |
+| `child_process.spawn` wrapper | `tFr` | constructed in shared factory's default branch |
+
+Extraction commands (verified against 1.5354.0):
+
+```bash
+cd build-reference/app-extracted/.vite/build
+
+# CCD spawn function name
+grep -Pzo 'async function \K\w+(?=\(\w*\)\s*\{(?s).{0,800}?Launching MCP Server)' index.js | tr '\0' '\n'
+
+# Shared transport factory (anchored on the unique 'built-in-node' string)
+grep -Pzo 'async function \K\w+(?=\([^)]*\)\s*\{(?s).{0,400}?built-in-node)' index.js | tr '\0' '\n'
+
+# All coordinator classes following the per-coordinator-registry pattern
+grep -Pzo 'class \K\w+(?=\s*\{(?s).{0,300}?this\.connections\s*=\s*new Map)' index.js | tr '\0' '\n'
+
+# LAM manager class specifically
+grep -Pzo 'class \K\w+(?=\s*\{(?s).{0,500}?local-mcp-server-cleanup)' index.js | tr '\0' '\n'
+```
+
+Two questions where a one-line answer from the team would help us route this downstream:
+
+1. Is per-coordinator isolated state intentional, or is it legacy drift from when each coordinator instantiated its transport inline?
+2. Is the recent extraction of the shared transport factory (`oPt`) the start of a dedup refactor, or incidental cleanup?
+
+If (1) is "intentional," we'll point users at the lockfile workaround as the supported path. If (2) is "in progress," this report saves you the duplicate analysis work.
+
+Full provenance: [aaddrick/claude-desktop-debian#546](https://github.com/aaddrick/claude-desktop-debian/issues/546). Related learnings doc updates: [#527](https://github.com/aaddrick/claude-desktop-debian/pull/527) and [#547](https://github.com/aaddrick/claude-desktop-debian/pull/547).
+
+## Filing checklist
+
+When you're ready to file:
+
+1. Open https://github.com/anthropics/claude-code/issues/new?template=bug_report.yml
+2. Paste each section above into the matching form field
+3. Submit
+4. Drop the GitHub issue URL as a comment on [#546](https://github.com/aaddrick/claude-desktop-debian/issues/546) so the trail is bidirectional
+
+Note: there is no in-app engineering bug-report path in Claude Desktop. `/bug` and `/feedback` are inert. The Help menu has "Get Support" (routes to the support chat, wrong queue for engineering) and "Troubleshooting" (self-diagnostic — useful for attaching `Copy Installation ID` or `Show Logs in File Manager` output to a GitHub issue, but not a reporting step on its own).
+
+## Voice and authorship
+
+Drafted using the [aaddrick-voice](https://github.com/aaddrick/written-voice-replication/blob/78f178dcf832943bcf1d5a65bf7627c3a20053a6/.claude/agents/aaddrick-voice.md) style profile against the form schema in `anthropics/claude-code/.github/ISSUE_TEMPLATE/bug_report.yml`.
+
+---
+Written by Claude Opus 4.7 via [Claude Code](https://claude.ai/code)

nix/claude-desktop.nix

@@ -245,6 +245,7 @@ cleanup_stale_cowork_socket
 log_message '--- Claude Desktop Launcher Start (NixOS) ---'
 log_message "Timestamp: $(date)"
 log_message "Arguments: $@"
+log_session_env
 
 # Check for display
 if ! check_display; then

scripts/launcher-common.sh

@@ -16,6 +16,39 @@ log_message() {
 	echo "$1" >> "$log_file"
 }
 
+# Log the session/IME environment vars that drive display and input
+# decisions, so bug reports include enough context to reason about
+# them without round-trip env-dump requests (#548).
+#
+# Emits one block:
+#     env={
+#       KEY=value
+#       ...
+#     }
+#
+# Empty or unset values are emitted as `KEY=` so absence is
+# unambiguous (vs. silently omitted). Caller must run setup_logging
+# first.
+log_session_env() {
+	local key
+	log_message 'env={'
+	for key in \
+		XDG_SESSION_TYPE \
+		WAYLAND_DISPLAY \
+		DISPLAY \
+		XDG_CURRENT_DESKTOP \
+		GTK_IM_MODULE \
+		XMODIFIERS \
+		QT_IM_MODULE \
+		CLAUDE_USE_WAYLAND \
+		CLAUDE_TITLEBAR_STYLE \
+		CLAUDE_GTK_IM_MODULE
+	do
+		log_message "  $key=${!key:-}"
+	done
+	log_message '}'
+}
+
 # Detect display backend (Wayland vs X11)
 # Sets: is_wayland, use_x11_on_wayland
 detect_display_backend() {

scripts/packaging/appimage.sh

@@ -98,6 +98,7 @@ log_message '--- Claude Desktop AppImage Start ---'
 log_message "Timestamp: $(date)"
 log_message "Arguments: $@"
 log_message "APPDIR: $appdir"
+log_session_env
 
 # Path to the bundled Electron executable and app
 electron_exec="$appdir/usr/lib/node_modules/electron/dist/electron"

scripts/packaging/deb.sh

@@ -114,6 +114,7 @@ cleanup_stale_cowork_socket
 log_message '--- Claude Desktop Launcher Start ---'
 log_message "Timestamp: \$(date)"
 log_message "Arguments: \$@"
+log_session_env
 
 # Check for display
 if ! check_display; then

scripts/packaging/rpm.sh

@@ -97,6 +97,7 @@ cleanup_stale_cowork_socket
 log_message '--- Claude Desktop Launcher Start ---'
 log_message "Timestamp: \$(date)"
 log_message "Arguments: \$@"
+log_session_env
 
 # Check for display
 if ! check_display; then

scripts/patches/app-asar.sh

@@ -37,16 +37,21 @@ EOFENTRY
 
 	# Update package.json
 	echo 'Modifying package.json to load frame fix and add node-pty...'
+	local desktop_name='claude-desktop.desktop'
+	if [[ ${build_format:-} == 'appimage' ]]; then
+		desktop_name='io.github.aaddrick.claude-desktop-debian.desktop'
+	fi
 	node -e "
 const fs = require('fs');
 const pkg = require('./app.asar.contents/package.json');
 pkg.originalMain = pkg.main;
 pkg.main = 'frame-fix-entry.js';
+pkg.desktopName = process.argv[1];
 pkg.optionalDependencies = pkg.optionalDependencies || {};
 pkg.optionalDependencies['node-pty'] = '^1.0.0';
 fs.writeFileSync('./app.asar.contents/package.json', JSON.stringify(pkg, null, 2));
-console.log('Updated package.json: main entry and node-pty dependency');
-"
+console.log('Updated package.json: main entry, desktopName, and node-pty dependency');
+" "$desktop_name"
 
 	# Create stub native module
 	echo 'Creating stub native module...'

tests/launcher-common.bats

@@ -35,10 +35,15 @@ setup() {
 	unset CLAUDE_USE_WAYLAND
 	unset NIRI_SOCKET
 	unset XDG_CURRENT_DESKTOP
+	unset XDG_SESSION_TYPE
 	unset CLAUDE_MENU_BAR
 	unset CLAUDE_TITLEBAR_STYLE
 	unset COWORK_VM_BACKEND
 	unset ELECTRON_USE_SYSTEM_TITLE_BAR
+	unset GTK_IM_MODULE
+	unset XMODIFIERS
+	unset QT_IM_MODULE
+	unset CLAUDE_GTK_IM_MODULE
 
 	# shellcheck source=scripts/launcher-common.sh
 	source "$SCRIPT_DIR/../scripts/launcher-common.sh"
@@ -86,6 +91,63 @@ teardown() {
 	[[ "${lines[1]}" == "test message two" ]]
 }
 
+# =============================================================================
+# log_session_env
+# =============================================================================
+
+@test "log_session_env: emits env={ ... } block with all required keys" {
+	setup_logging
+	XDG_SESSION_TYPE='wayland'
+	WAYLAND_DISPLAY='wayland-0'
+	DISPLAY=':0'
+	XDG_CURRENT_DESKTOP='KDE'
+	GTK_IM_MODULE='ibus'
+	XMODIFIERS='@im=ibus'
+	QT_IM_MODULE='ibus'
+	CLAUDE_USE_WAYLAND='1'
+	CLAUDE_TITLEBAR_STYLE='hybrid'
+	CLAUDE_GTK_IM_MODULE='xim'
+	log_session_env
+
+	run cat "$log_file"
+	# Exact-line match locks block structure (open/close braces on
+	# their own lines) and per-key formatting in one pass.
+	[[ "${lines[0]}"  == 'env={' ]]
+	[[ "${lines[1]}"  == '  XDG_SESSION_TYPE=wayland' ]]
+	[[ "${lines[2]}"  == '  WAYLAND_DISPLAY=wayland-0' ]]
+	[[ "${lines[3]}"  == '  DISPLAY=:0' ]]
+	[[ "${lines[4]}"  == '  XDG_CURRENT_DESKTOP=KDE' ]]
+	[[ "${lines[5]}"  == '  GTK_IM_MODULE=ibus' ]]
+	[[ "${lines[6]}"  == '  XMODIFIERS=@im=ibus' ]]
+	[[ "${lines[7]}"  == '  QT_IM_MODULE=ibus' ]]
+	[[ "${lines[8]}"  == '  CLAUDE_USE_WAYLAND=1' ]]
+	[[ "${lines[9]}"  == '  CLAUDE_TITLEBAR_STYLE=hybrid' ]]
+	[[ "${lines[10]}" == '  CLAUDE_GTK_IM_MODULE=xim' ]]
+	[[ "${lines[11]}" == '}' ]]
+}
+
+@test "log_session_env: unset/empty values render as 'KEY=' (no value)" {
+	setup_logging
+	# All vars unset by setup() except this one, which exercises the
+	# empty-string branch (must be indistinguishable from unset).
+	GTK_IM_MODULE=''
+	log_session_env
+
+	run cat "$log_file"
+	# Exact-line match proves the line ends right after '=' — a
+	# substring like *'KEY='* would also match 'KEY=value'.
+	[[ "${lines[1]}"  == '  XDG_SESSION_TYPE=' ]]
+	[[ "${lines[2]}"  == '  WAYLAND_DISPLAY=' ]]
+	[[ "${lines[3]}"  == '  DISPLAY=' ]]
+	[[ "${lines[4]}"  == '  XDG_CURRENT_DESKTOP=' ]]
+	[[ "${lines[5]}"  == '  GTK_IM_MODULE=' ]]
+	[[ "${lines[6]}"  == '  XMODIFIERS=' ]]
+	[[ "${lines[7]}"  == '  QT_IM_MODULE=' ]]
+	[[ "${lines[8]}"  == '  CLAUDE_USE_WAYLAND=' ]]
+	[[ "${lines[9]}"  == '  CLAUDE_TITLEBAR_STYLE=' ]]
+	[[ "${lines[10]}" == '  CLAUDE_GTK_IM_MODULE=' ]]
+}
+
 # =============================================================================
 # check_display
 # =============================================================================

tests/test-artifact-appimage.sh

@@ -94,7 +94,7 @@ assert_contains "$appdir/AppRun" 'build_electron_args' \
 
 # --- App contents (asar) ---
 resources_dir="$appdir/usr/lib/node_modules/electron/dist/resources"
-validate_app_contents "$resources_dir"
+validate_app_contents "$resources_dir" "${component_id}.desktop"
 
 # --- Cleanup ---
 rm -rf "$extract_dir"

tests/test-artifact-common.sh

@@ -59,8 +59,10 @@ assert_command_succeeds() {
 
 # Validate app contents inside an Electron resources directory.
 # $1 = path to the resources/ dir containing app.asar
+# $2 = expected desktopName in app/package.json
 validate_app_contents() {
 	local resources_dir="$1"
+	local expected_desktop_name="${2:-claude-desktop.desktop}"
 
 	assert_file_exists "$resources_dir/app.asar"
 	assert_dir_exists "$resources_dir/app.asar.unpacked"
@@ -95,6 +97,11 @@ validate_app_contents() {
 			'frame-fix-entry.js' \
 			"package.json main field references frame-fix-entry.js"
 
+		# package.json desktopName matches the installed desktop file
+		assert_contains "$extract_dir/app/package.json" \
+			"\"desktopName\": \"$expected_desktop_name\"" \
+			"package.json desktopName matches $expected_desktop_name"
+
 		# .vite/build/index.js exists (main process code)
 		assert_file_exists "$extract_dir/app/.vite/build/index.js"