Fix ai_tools apps collector matching AI apps by raw substring

orbit/pkg/table/ai_tools/app_row_test.go

@@ -0,0 +1,46 @@
+package ai_tools
+
+import (
+	"encoding/json"
+	"testing"
+
+	"github.com/fleetdm/fleet/v4/orbit/pkg/table/ai_tools/internal/apps"
+)
+
+// TestAppRow locks in the apps column semantics: `name` carries the installed
+// program's real display name and `identifier` carries the known-app key, so a
+// wrong match shows the real program instead of masquerading as the known app;
+// the bundle id lives in `detail`.
+func TestAppRow(t *testing.T) {
+	a := apps.App{
+		Name:           "claude-desktop", // known-app key
+		DisplayName:    "Claude",
+		Vendor:         "Anthropic",
+		BundleID:       "com.anthropic.claude",
+		Version:        "1.2.3",
+		Path:           "/Applications/Claude.app",
+		PlatformSource: "applications",
+		Scope:          "system",
+		Running:        1,
+		PID:            42,
+		SHA256:         "deadbeef",
+	}
+	r := appRow(a)
+
+	if r["type"] != "apps" || r["name"] != "Claude" ||
+		r["identifier"] != "claude-desktop" ||
+		r["location"] != "local" || r["source"] != "applications" ||
+		r["version"] != "1.2.3" || r["path"] != "/Applications/Claude.app" ||
+		r["running"] != "1" || r["pid"] != "42" || r["sha256"] != "deadbeef" {
+		t.Errorf("row columns wrong: %+v", r)
+	}
+
+	var detail map[string]string
+	if err := json.Unmarshal([]byte(r["detail"]), &detail); err != nil {
+		t.Fatalf("detail not valid JSON: %v (%q)", err, r["detail"])
+	}
+	if detail["vendor"] != "Anthropic" || detail["bundle_id"] != "com.anthropic.claude" ||
+		detail["scope"] != "system" {
+		t.Errorf("detail wrong: %+v", detail)
+	}
+}

orbit/pkg/table/ai_tools/internal/apps/apps.go

@@ -15,6 +15,7 @@ import (
 // App is a detected AI desktop application.
 type App struct {
 	Name           string
+	DisplayName    string
 	Vendor         string
 	Path           string
 	BundleID       string
@@ -43,17 +44,17 @@ func knownApps() []knownApp {
 		{"chatgpt", []string{"chatgpt"}, []string{"chatgpt"}, 0},
 		{"ollama", []string{"ollama"}, []string{"ollama"}, 11434},
 		{"lm-studio", []string{"lm studio", "lmstudio", "lm-studio"}, []string{"lm studio", "lm-studio", "lmstudio"}, 1234},
-		{"jan", []string{"jan.app", "jan ", "/jan"}, []string{"jan"}, 1337},
+		{"jan", []string{"jan"}, []string{"jan"}, 1337},
 		{"gpt4all", []string{"gpt4all"}, []string{"gpt4all"}, 0},
 		{"msty", []string{"msty"}, []string{"msty"}, 0},
 		{"anythingllm", []string{"anythingllm", "anything llm"}, []string{"anythingllm"}, 0},
-		{"comet", []string{"comet.app", "comet "}, []string{"comet"}, 0}, // Perplexity Comet (AI browser)
-		{"dia", []string{"dia.app", "dia "}, []string{"dia"}, 0},         // Browser Company Dia (AI browser)
+		{"comet", []string{"comet"}, []string{"comet"}, 0}, // Perplexity Comet (AI browser)
+		{"dia", []string{"dia"}, []string{"dia"}, 0},       // Browser Company Dia (AI browser)
 		{"perplexity", []string{"perplexity"}, []string{"perplexity"}, 0},
 		{"cursor", []string{"cursor"}, []string{"cursor"}, 0},
 		{"windsurf", []string{"windsurf"}, []string{"windsurf"}, 0},
 		{"antigravity", []string{"antigravity"}, []string{"antigravity"}, 0},
-		{"trae", []string{"trae.app", "trae "}, []string{"trae"}, 0},
+		{"trae", []string{"trae"}, []string{"trae"}, 0},
 		{"lm-studio-cli", []string{"lms"}, []string{"lms"}, 0},
 	}
 }
@@ -62,6 +63,11 @@ func knownApps() []knownApp {
 func Scan(homesList []homes.Home, snap *proc.Snapshot) []App {
 	out := scanApps(homesList) // platform-specific (build-tagged)
 	for i := range out {
+		// Some discoveries carry no display name (e.g. a bare service binary);
+		// fall back to the known-app key rather than reporting a nameless row.
+		if out[i].DisplayName == "" {
+			out[i].DisplayName = out[i].Name
+		}
 		k, ok := knownByName(out[i].Name)
 		if !ok {
 			continue
@@ -83,11 +89,12 @@ func Scan(homesList []homes.Home, snap *proc.Snapshot) []App {
 }
 
 // containsWordBoundary reports whether want appears in pn delimited by
-// non-alphanumeric characters (or the string edges). Several processNames tokens
-// are short, common substrings (e.g. "dia", "jan"); a plain strings.Contains
-// would falsely match unrelated processes like "mediaanalysisd", marking an app
-// Running with an unrelated PID. It scans every occurrence so a bounded match
-// later in the string is still found.
+// non-alphanumeric characters (or the string edges). Several match and
+// processNames tokens are short, common substrings (e.g. "dia", "jan", "lms");
+// a plain strings.Contains would falsely match unrelated programs like "NVIDIA
+// Control Panel" or processes like "mediaanalysisd", reporting an AI app that
+// was never installed or marking one Running with an unrelated PID. It scans
+// every occurrence so a bounded match later in the string is still found.
 func containsWordBoundary(pn, want string) bool {
 	if want == "" {
 		return false
@@ -112,23 +119,22 @@ func isAlnum(b byte) bool {
 	return (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z') || (b >= '0' && b <= '9')
 }
 
-// matchKnown finds the AI app a set of identifying strings belongs to.
+// matchKnown finds the AI app a set of identifying strings belongs to. Tokens
+// are matched at word boundaries, never mid-word: the joining NUL and the
+// non-alphanumeric characters of display names, bundle ids, and file names all
+// delimit words.
 func matchKnown(tokens ...string) (knownApp, bool) {
 	hay := strings.ToLower(strings.Join(tokens, "\x00"))
 	for _, k := range knownApps() {
 		for _, m := range k.match {
-			if strings.Contains(hay, m) {
+			if containsWordBoundary(hay, m) {
 				return k, true
 			}
 		}
 	}
 	return knownApp{}, false
 }
 
-// firstNonEmpty is used by scanApps on darwin and linux; the windows build has
-// no caller, so it is exempt from the unused check there.
-//
-//nolint:unused
 func firstNonEmpty(vals ...string) string {
 	for _, v := range vals {
 		if v != "" {

orbit/pkg/table/ai_tools/internal/apps/apps_darwin.go

@@ -49,6 +49,7 @@ func scanApps(homesList []homes.Home) []App {
 			}
 			out = append(out, App{
 				Name:           k.name,
+				DisplayName:    firstNonEmpty(info.BundleName, strings.TrimSuffix(e.Name(), ".app")),
 				BundleID:       info.BundleID,
 				Version:        firstNonEmpty(info.ShortVersion, info.BundleVersion),
 				Path:           appPath,

orbit/pkg/table/ai_tools/internal/apps/apps_darwin_test.go

@@ -46,4 +46,7 @@ func TestScanAppsRejectsExecutableTraversal(t *testing.T) {
 	if msty.execPath != "" {
 		t.Errorf("execPath = %q, want empty (traversal CFBundleExecutable must be rejected)", msty.execPath)
 	}
+	if msty.DisplayName != "Msty" {
+		t.Errorf("DisplayName = %q, want the bundle's real name \"Msty\"", msty.DisplayName)
+	}
 }

orbit/pkg/table/ai_tools/internal/apps/apps_linux.go

@@ -32,6 +32,7 @@ func scanApps(homesList []homes.Home) []App {
 			seen[ka.name] = struct{}{}
 			out = append(out, App{
 				Name:           ka.name,
+				DisplayName:    name,
 				Path:           firstNonEmpty(exec, e.Name()),
 				PlatformSource: "desktop-file",
 				Scope:          scope,

orbit/pkg/table/ai_tools/internal/apps/apps_test.go

@@ -21,6 +21,29 @@ func TestMatchKnown(t *testing.T) {
 		{[]string{"Comet.app", "Comet", "ai.perplexity.comet"}, true, "comet", 0},
 		{[]string{"Dia.app", "Dia", "company.thebrowser.dia"}, true, "dia", 0},
 		{[]string{"Perplexity.app", "Perplexity", "ai.perplexity.macos"}, true, "perplexity", 0},
+
+		// Windows uninstall entries and MSIX packages carry a bare display name
+		// with no trailing delimiter; a genuine install must still match.
+		{[]string{"Dia"}, true, "dia", 0},
+		{[]string{"Jan"}, true, "jan", 1337},
+		{[]string{"Comet"}, true, "comet", 0},
+		{[]string{"Trae"}, true, "trae", 0},
+		{[]string{"lms"}, true, "lm-studio-cli", 0},
+
+		// A bounded occurrence after a mid-word one must still match: the
+		// boundary scan may not stop at the first hit.
+		{[]string{"NVIDIA Dia"}, true, "dia", 0},
+
+		// Short tokens ("dia", "lms") must not match mid-word inside unrelated
+		// software names.
+		{[]string{"NVIDIA Control Panel"}, false, "", 0},
+		{[]string{"NVIDIA Graphics Driver 591.86"}, false, "", 0},
+		{[]string{"NVIDIA Install Application"}, false, "", 0},
+		{[]string{"VLC media player"}, false, "", 0},
+		{[]string{"Plex Media Server 1.43.1.10611 (x64)"}, false, "", 0},
+		{[]string{"vs_minshellmsi"}, false, "", 0},
+		{[]string{"vs_minshellmsires"}, false, "", 0},
+		{[]string{"Windows Media Player.app"}, false, "", 0},
 	}
 	for _, c := range cases {
 		k, ok := matchKnown(c.tokens...)

orbit/pkg/table/ai_tools/internal/apps/apps_windows.go

@@ -101,6 +101,7 @@ func scanApps(homesList []homes.Home) []App {
 			}
 			c.add(appCandidate{
 				MatchTokens: []string{display},
+				DisplayName: display,
 				Vendor:      pub,
 				Version:     version,
 				Path:        loc,

orbit/pkg/table/ai_tools/internal/apps/appx.go

@@ -159,6 +159,7 @@ func scanAppxDirs(c *appCollector, installRoot string, userPkgDirs []string) {
 		}
 		c.add(appCandidate{
 			MatchTokens: []string{pkg.Name, appxLiteral(man.Properties.DisplayName)},
+			DisplayName: firstNonEmpty(appxLiteral(man.Properties.DisplayName), pkg.Name),
 			Vendor:      appxVendor(man.Properties.PublisherDisplayName, man.Identity.Publisher),
 			Version:     pkg.Version,
 			Path:        dir,

orbit/pkg/table/ai_tools/internal/apps/appx_scan_test.go

@@ -49,7 +49,7 @@ func TestScanAppxDirsInstallRoot(t *testing.T) {
 		t.Fatalf("got %d apps, want 1: %+v", len(got), got)
 	}
 	want := App{
-		Name: "chatgpt", Vendor: "OpenAI", Version: "1.2026.190.0",
+		Name: "chatgpt", DisplayName: "ChatGPT", Vendor: "OpenAI", Version: "1.2026.190.0",
 		Path: dir, Scope: "system", PlatformSource: "appx",
 	}
 	if got[0] != want {
@@ -198,10 +198,19 @@ func TestScanAppxDirsStaleUserDirIsNotAnInstall(t *testing.T) {
 	}
 }
 
-// TestScanAppxDirsMissingManifest covers the manifest being unreadable, which
-// must cost only the vendor, never the row.
+// TestScanAppxDirsMissingManifest covers a manifest that is unreadable or
+// carries no usable display name (including a MUI indirect string), which must
+// cost only the vendor and the friendly name, never the row: the display name
+// falls back to the package identity name.
 func TestScanAppxDirsMissingManifest(t *testing.T) {
-	for _, manifest := range []string{"", "not xml at all <<<", "<Package><Identity/></Package>"} {
+	const indirectManifest = `<?xml version="1.0" encoding="utf-8"?>
+<Package xmlns="http://schemas.microsoft.com/appx/manifest/foundation/windows10">
+  <Properties>
+    <DisplayName>@{OpenAI.ChatGPT-Desktop_1.2026.190.0_arm64__2p2nqsd0c76g0?ms-resource://OpenAI.ChatGPT-Desktop/Resources/AppName}</DisplayName>
+  </Properties>
+</Package>`
+
+	for _, manifest := range []string{"", "not xml at all <<<", "<Package><Identity/></Package>", indirectManifest} {
 		root := t.TempDir()
 		mkPackageDir(t, root, chatGPTPFN, manifest)
 
@@ -218,6 +227,9 @@ func TestScanAppxDirsMissingManifest(t *testing.T) {
 		if got[0].Vendor != "" {
 			t.Errorf("manifest=%q: got vendor %q, want empty", manifest, got[0].Vendor)
 		}
+		if got[0].DisplayName != "OpenAI.ChatGPT-Desktop" {
+			t.Errorf("manifest=%q: got display name %q, want the package identity name", manifest, got[0].DisplayName)
+		}
 	}
 }
 

orbit/pkg/table/ai_tools/internal/apps/appx_test.go

@@ -210,7 +210,9 @@ func TestMatchKnownPackageNames(t *testing.T) {
 	}{
 		{"OpenAI.ChatGPT-Desktop", true, "chatgpt"},
 		{"ElementLabs.LMStudio", true, "lm-studio"},
-		{"Perplexity.Comet", true, "perplexity"},
+		// The "." delimits a word, so the "comet" token sees the product segment
+		// and wins over the publisher-only "perplexity" match.
+		{"Perplexity.Comet", true, "comet"},
 
 		{"Microsoft.WindowsCalculator", false, ""},
 		{"Microsoft.VCLibs.140.00", false, ""},

orbit/pkg/table/ai_tools/internal/apps/collect.go

@@ -13,6 +13,7 @@ package apps
 // matched against the known-app list and deduplicated.
 type appCandidate struct {
 	MatchTokens []string // identifying strings matched against knownApps
+	DisplayName string
 	Vendor      string
 	Version     string
 	Path        string
@@ -66,6 +67,7 @@ func (c *appCollector) add(cand appCandidate) bool {
 	c.seen[k.name] = struct{}{}
 	c.out = append(c.out, App{
 		Name:           k.name,
+		DisplayName:    cand.DisplayName,
 		Vendor:         cand.Vendor,
 		Version:        cand.Version,
 		Path:           cand.Path,

orbit/pkg/table/ai_tools/internal/apps/collect_test.go

@@ -53,6 +53,43 @@ func TestAppCollectorSharedAcrossSources(t *testing.T) {
 	}
 }
 
+// TestAppCollectorImpostorDoesNotMaskRealInstall covers the masking hazard of
+// first-match-wins dedup: an unrelated system-wide program whose name merely
+// contains a known-app token must not match at all, so it can never claim the
+// slot of a genuine per-user install of that app scanned later.
+func TestAppCollectorImpostorDoesNotMaskRealInstall(t *testing.T) {
+	c := newAppCollector()
+	if c.add(appCandidate{
+		MatchTokens: []string{"NVIDIA Control Panel"},
+		Vendor:      "NVIDIA Corporation",
+		Version:     "8.1.969.0",
+		Path:        `C:\Program Files\NVIDIA Corporation\Control Panel Client`,
+		Scope:       "system",
+		Source:      "registry",
+	}) {
+		t.Error("NVIDIA Control Panel was collected as an AI app")
+	}
+	if !c.add(appCandidate{
+		MatchTokens: []string{"Dia"},
+		DisplayName: "Dia",
+		Vendor:      "The Browser Company",
+		Version:     "1.0.0",
+		Path:        `C:\Users\alice\AppData\Local\Programs\Dia`,
+		Scope:       "user",
+		Source:      "registry",
+	}) {
+		t.Error("genuine per-user Dia install was not collected")
+	}
+
+	got := c.apps()
+	if len(got) != 1 {
+		t.Fatalf("got %d apps, want 1: %+v", len(got), got)
+	}
+	if got[0].Name != "dia" || got[0].DisplayName != "Dia" || got[0].Scope != "user" || got[0].Vendor != "The Browser Company" {
+		t.Errorf("got %+v, want the genuine user-scoped Dia install", got[0])
+	}
+}
+
 func TestAppCollectorSkipsUnknown(t *testing.T) {
 	c := newAppCollector()
 	for _, tokens := range [][]string{
@@ -96,6 +133,7 @@ func TestAppCollectorPreservesOrderAndFields(t *testing.T) {
 	c := newAppCollector()
 	c.add(appCandidate{
 		MatchTokens: []string{"Ollama"},
+		DisplayName: "Ollama",
 		Vendor:      "Ollama Inc.",
 		Version:     "0.5.7",
 		Path:        `C:\Users\alice\AppData\Local\Programs\Ollama`,
@@ -112,7 +150,7 @@ func TestAppCollectorPreservesOrderAndFields(t *testing.T) {
 		t.Errorf("got order [%q %q], want [\"ollama\" \"cursor\"]", got[0].Name, got[1].Name)
 	}
 	want := App{
-		Name: "ollama", Vendor: "Ollama Inc.", Version: "0.5.7",
+		Name: "ollama", DisplayName: "Ollama", Vendor: "Ollama Inc.", Version: "0.5.7",
 		Path: `C:\Users\alice\AppData\Local\Programs\Ollama`, Scope: "user", PlatformSource: "registry",
 	}
 	if got[0] != want {

orbit/pkg/table/ai_tools/tables.go

@@ -48,7 +48,7 @@ var allTypes = []string{"mcp_server", "ide_plugins", "agents", "apps", "sockets"
 var columns = []string{
 	"type",       // mcp_server | ide_plugins | agents | apps | sockets | agent_instruction | browser_extension
 	"name",       // server/plugin/agent/app/process/instruction-file name
-	"identifier", // plugin_id | bundle_id | mcp server name | agent binary | socket service
+	"identifier", // plugin_id | known-app key | mcp server name | agent binary | socket service
 	"category",   // classification bucket (coding-assistant, agent-runtime, ai-api-egress, ...)
 	"location",   // local | remote
 	"source",     // provenance: client | editor | install_method | platform_source | direction | tool
@@ -301,8 +301,8 @@ func agentRow(a agents.Agent) map[string]string {
 func appRow(a apps.App) map[string]string {
 	return row(map[string]string{
 		"type":       "apps",
-		"name":       a.Name,
-		"identifier": a.BundleID,
+		"name":       a.DisplayName,
+		"identifier": a.Name,
 		"location":   "local",
 		"source":     a.PlatformSource,
 		"version":    a.Version,