diff --git a/internal/converter/epub.go b/internal/converter/epub.go
index 8a203766..bf929f9f 100644
--- a/internal/converter/epub.go
+++ b/internal/converter/epub.go
@@ -5,6 +5,7 @@ package converter
 import (
 	"crypto/md5"
 	"fmt"
+	"html"
 	"net/url"
 	"os"
 	"path/filepath"
@@ -123,9 +124,7 @@ func ConvertHTMLToEPUB(htmlContent string, outputPath string, options EPUBOption
 
 	// Prepend source URL if provided
 	if options.SourceURL != "" {
-		sourceHeader := fmt.Sprintf(`<p style="font-size: 0.85em; color: #666; margin-bottom: 1.5em;">Source: <a href="%s">%s</a></p>`,
-			options.SourceURL, options.SourceURL)
-		processedHTML = sourceHeader + processedHTML
+		processedHTML = sourceHeaderHTML(options.SourceURL) + processedHTML
 	}
 
 	// Add the main content section
@@ -144,6 +143,13 @@ func ConvertHTMLToEPUB(htmlContent string, outputPath string, options EPUBOption
 	return nil
 }
 
+// sourceHeaderHTML renders the attribution line linking back to the source.
+func sourceHeaderHTML(sourceURL string) string {
+	escaped := html.EscapeString(sourceURL)
+	return fmt.Sprintf(`<p style="font-size: 0.85em; color: #666; margin-bottom: 1.5em;">Source: <a href="%s">%s</a></p>`,
+		escaped, escaped)
+}
+
 // ConvertHTMLFileToEPUB reads an HTML file and converts it to EPUB.
 func ConvertHTMLFileToEPUB(htmlPath string, options EPUBOptions) (string, error) {
 	logging.Logf("[EPUB] ConvertHTMLFileToEPUB: processing %s", htmlPath)
diff --git a/internal/converter/epub_test.go b/internal/converter/epub_test.go
new file mode 100644
index 00000000..48d9de00
--- /dev/null
+++ b/internal/converter/epub_test.go
@@ -0,0 +1,51 @@
+package converter
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestSourceHeaderHTMLEscapesURL(t *testing.T) {
+	tests := []struct {
+		name      string
+		sourceURL string
+		unwanted  string
+	}{
+		{
+			name:      "quote closing the href attribute",
+			sourceURL: `https://example.com/x"><script>alert(1)</script>`,
+			unwanted:  "<script>",
+		},
+		{
+			name:      "quote opening an attribute on the anchor",
+			sourceURL: `https://example.com/x" onmouseover="alert(1)`,
+			unwanted:  `onmouseover="`,
+		},
+		{
+			name:      "angle brackets in the link text",
+			sourceURL: `https://example.com/<img src=x onerror=alert(1)>`,
+			unwanted:  "<img",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := sourceHeaderHTML(tt.sourceURL)
+			if strings.Contains(got, tt.unwanted) {
+				t.Errorf("sourceHeaderHTML(%q) leaked %q:\n%s", tt.sourceURL, tt.unwanted, got)
+			}
+			if strings.Count(got, `<a href="`) != 1 {
+				t.Errorf("expected exactly one anchor, got:\n%s", got)
+			}
+		})
+	}
+}
+
+func TestSourceHeaderHTMLKeepsOrdinaryURL(t *testing.T) {
+	url := "https://example.com/article?id=1&page=2"
+	got := sourceHeaderHTML(url)
+
+	if !strings.Contains(got, "https://example.com/article?id=1&amp;page=2") {
+		t.Errorf("expected the URL preserved with escaping, got:\n%s", got)
+	}
+}