diff --git a/claudetool/onstart/analyze_test.go b/claudetool/onstart/analyze_test.go
new file mode 100644
index 0000000..be70ce7
--- /dev/null
+++ b/claudetool/onstart/analyze_test.go
@@ -0,0 +1,157 @@
+package onstart
+
+import (
+	"context"
+	"slices"
+	"testing"
+)
+
+func TestAnalyzeCodebase(t *testing.T) {
+	t.Run("Basic Analysis", func(t *testing.T) {
+		// Test basic functionality with regular ASCII filenames
+		codebase, err := AnalyzeCodebase(context.Background(), ".")
+		if err != nil {
+			t.Fatalf("AnalyzeCodebase failed: %v", err)
+		}
+
+		if codebase == nil {
+			t.Fatal("Expected non-nil codebase")
+		}
+
+		if codebase.TotalFiles == 0 {
+			t.Error("Expected some files to be analyzed")
+		}
+
+		if len(codebase.ExtensionCounts) == 0 {
+			t.Error("Expected extension counts to be populated")
+		}
+	})
+
+	t.Run("Non-ASCII Filenames", func(t *testing.T) {
+		// Test with non-ASCII characters in filenames
+		testdataPath := "./testdata"
+		codebase, err := AnalyzeCodebase(context.Background(), testdataPath)
+		if err != nil {
+			t.Fatalf("AnalyzeCodebase failed with non-ASCII filenames: %v", err)
+		}
+
+		if codebase == nil {
+			t.Fatal("Expected non-nil codebase")
+		}
+
+		// We expect 8 files in our testdata directory
+		expectedFiles := 8
+		if codebase.TotalFiles != expectedFiles {
+			t.Errorf("Expected %d files, got %d", expectedFiles, codebase.TotalFiles)
+		}
+
+		// Verify extension counts include our non-ASCII files
+		expectedExtensions := map[string]int{
+			".go":            1, // 测试文件.go
+			".js":            1, // café.js
+			".py":            1, // русский.py
+			".md":            3, // 🚀rocket.md, readme-español.md, claude-한국어.md
+			".html":          1, // Übung.html
+			"<no-extension>": 1, // Makefile-日本語
+		}
+
+		for ext, expectedCount := range expectedExtensions {
+			actualCount, exists := codebase.ExtensionCounts[ext]
+			if !exists {
+				t.Errorf("Expected extension %s to be found", ext)
+				continue
+			}
+			if actualCount != expectedCount {
+				t.Errorf("Expected %d files with extension %s, got %d", expectedCount, ext, actualCount)
+			}
+		}
+
+		// Verify file categorization works with non-ASCII filenames
+		// Check build files
+		if !slices.Contains(codebase.BuildFiles, "Makefile-日本語") {
+			t.Error("Expected Makefile-日本語 to be categorized as a build file")
+		}
+
+		// Check documentation files
+		if !slices.Contains(codebase.DocumentationFiles, "readme-español.md") {
+			t.Error("Expected readme-español.md to be categorized as a documentation file")
+		}
+
+		// Check guidance files
+		if !slices.Contains(codebase.GuidanceFiles, "subdir/claude.한국어.md") {
+			t.Error("Expected subdir/claude.한국어.md to be categorized as a guidance file")
+		}
+	})
+}
+
+func TestCategorizeFile(t *testing.T) {
+	t.Run("Non-ASCII Filenames", func(t *testing.T) {
+		tests := []struct {
+			name     string
+			path     string
+			expected string
+		}{
+			{"Chinese Go file", "测试文件.go", ""},
+			{"French JS file", "café.js", ""},
+			{"Russian Python file", "русский.py", ""},
+			{"Emoji markdown file", "🚀rocket.md", ""},
+			{"German HTML file", "Übung.html", ""},
+			{"Japanese Makefile", "Makefile-日本語", "build"},
+			{"Spanish README", "readme-español.md", "documentation"},
+			{"Korean Claude file", "subdir/claude.한국어.md", "guidance"},
+			// Test edge cases with Unicode normalization and combining characters
+			{"Mixed Unicode file", "test中文🚀.txt", ""},
+			{"Combining characters", "filé̂.go", ""}, // file with combining acute and circumflex accents
+			{"Right-to-left script", "مرحبا.py", ""},  // Arabic "hello"
+		}
+
+		for _, tt := range tests {
+			t.Run(tt.name, func(t *testing.T) {
+				result := categorizeFile(tt.path)
+				if result != tt.expected {
+					t.Errorf("categorizeFile(%q) = %q, want %q", tt.path, result, tt.expected)
+				}
+			})
+		}
+	})
+}
+
+func TestTopExtensions(t *testing.T) {
+	t.Run("With Non-ASCII Files", func(t *testing.T) {
+		// Create a test codebase with known extension counts
+		codebase := &Codebase{
+			ExtensionCounts: map[string]int{
+				".md":   5, // Most common
+				".go":   3,
+				".js":   2,
+				".py":   1,
+				".html": 1, // Least common
+			},
+			TotalFiles: 12,
+		}
+
+		topExt := codebase.TopExtensions()
+		if len(topExt) != 5 {
+			t.Errorf("Expected 5 top extensions, got %d", len(topExt))
+		}
+
+		// Check that extensions are sorted by count (descending)
+		expected := []string{
+			".md: 5 (42%)",
+			".go: 3 (25%)",
+			".js: 2 (17%)",
+			".html: 1 (8%)",
+			".py: 1 (8%)",
+		}
+
+		for i, expectedExt := range expected {
+			if i >= len(topExt) {
+				t.Errorf("Missing expected extension at index %d: %s", i, expectedExt)
+				continue
+			}
+			if topExt[i] != expectedExt {
+				t.Errorf("Expected extension %q at index %d, got %q", expectedExt, i, topExt[i])
+			}
+		}
+	})
+}
diff --git "a/claudetool/onstart/testdata/Makefile-\346\227\245\346\234\254\350\252\236" "b/claudetool/onstart/testdata/Makefile-\346\227\245\346\234\254\350\252\236"
new file mode 100644
index 0000000..c78f45a
--- /dev/null
+++ "b/claudetool/onstart/testdata/Makefile-\346\227\245\346\234\254\350\252\236"
@@ -0,0 +1,10 @@
+# Makefile with Japanese characters in filename
+# This should be categorized as a build file
+
+all:
+	echo "Building with Japanese characters in Makefile name"
+
+clean:
+	rm -f *.o
+
+.PHONY: all clean
diff --git "a/claudetool/onstart/testdata/caf\303\251.js" "b/claudetool/onstart/testdata/caf\303\251.js"
new file mode 100644
index 0000000..6cd53e3
--- /dev/null
+++ "b/claudetool/onstart/testdata/caf\303\251.js"
@@ -0,0 +1,2 @@
+// JavaScript file with French accent in filename
+console.log('Hello from café.js');
diff --git "a/claudetool/onstart/testdata/readme-espa\303\261ol.md" "b/claudetool/onstart/testdata/readme-espa\303\261ol.md"
new file mode 100644
index 0000000..4232509
--- /dev/null
+++ "b/claudetool/onstart/testdata/readme-espa\303\261ol.md"
@@ -0,0 +1,8 @@
+# README Español
+
+This is a documentation file with Spanish characters in the filename.
+
+## Características
+
+- Soporte para Unicode
+- Caracteres españoles en nombres de archivo
diff --git "a/claudetool/onstart/testdata/subdir/claude.\355\225\234\352\265\255\354\226\264.md" "b/claudetool/onstart/testdata/subdir/claude.\355\225\234\352\265\255\354\226\264.md"
new file mode 100644
index 0000000..ab36b92
--- /dev/null
+++ "b/claudetool/onstart/testdata/subdir/claude.\355\225\234\352\265\255\354\226\264.md"
@@ -0,0 +1,8 @@
+# Claude Guidance with Korean Characters
+
+This file should be categorized as a guidance file since it starts with 'claude-' and ends with '.md'.
+
+## 지침
+
+- 한국어 문자 지원
+- 파일 이름에 유니코드 사용
diff --git "a/claudetool/onstart/testdata/\303\234bung.html" "b/claudetool/onstart/testdata/\303\234bung.html"
new file mode 100644
index 0000000..afc66ab
--- /dev/null
+++ "b/claudetool/onstart/testdata/\303\234bung.html"
@@ -0,0 +1,10 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>German Umlaut Test</title>
+</head>
+<body>
+    <h1>Übung HTML File</h1>
+    <p>This HTML file has German umlaut characters in the filename.</p>
+</body>
+</html>
diff --git "a/claudetool/onstart/testdata/\321\200\321\203\321\201\321\201\320\272\320\270\320\271.py" "b/claudetool/onstart/testdata/\321\200\321\203\321\201\321\201\320\272\320\270\320\271.py"
new file mode 100644
index 0000000..8110369
--- /dev/null
+++ "b/claudetool/onstart/testdata/\321\200\321\203\321\201\321\201\320\272\320\270\320\271.py"
@@ -0,0 +1,2 @@
+# Python file with Russian characters in filename
+print('Hello from русский.py')
diff --git "a/claudetool/onstart/testdata/\346\265\213\350\257\225\346\226\207\344\273\266.go" "b/claudetool/onstart/testdata/\346\265\213\350\257\225\346\226\207\344\273\266.go"
new file mode 100644
index 0000000..30a3d59
--- /dev/null
+++ "b/claudetool/onstart/testdata/\346\265\213\350\257\225\346\226\207\344\273\266.go"
@@ -0,0 +1,6 @@
+// Package test with Chinese characters in filename
+package test
+
+func TestFunction() {
+	// This is a test file with Chinese characters in the filename
+}
diff --git "a/claudetool/onstart/testdata/\360\237\232\200rocket.md" "b/claudetool/onstart/testdata/\360\237\232\200rocket.md"
new file mode 100644
index 0000000..6140a51
--- /dev/null
+++ "b/claudetool/onstart/testdata/\360\237\232\200rocket.md"
@@ -0,0 +1,8 @@
+# README with Emoji in filename
+
+This is a documentation file with an emoji character in the filename.
+
+## Features
+
+- Unicode support
+- Emoji in filenames