diff --git a/include/utf8.h b/include/utf8.h
index 3008a86..4dfc9c8 100644
--- a/include/utf8.h
+++ b/include/utf8.h
@@ -24,4 +24,7 @@ int utf8_strlen(const char *str);
 /* Remove last UTF-8 character from string */
 void utf8_remove_last_char(char *str);
 
+/* Validate a UTF-8 byte sequence */
+bool utf8_is_valid_sequence(const char *bytes, int len);
+
 #endif /* UTF8_H */
diff --git a/src/ssh_server.c b/src/ssh_server.c
index 4023611..c532d09 100644
--- a/src/ssh_server.c
+++ b/src/ssh_server.c
@@ -108,6 +108,12 @@ int client_printf(client_t *client, const char *fmt, ...) {
     va_start(args, fmt);
     int len = vsnprintf(buffer, sizeof(buffer), fmt, args);
     va_end(args);
+
+    /* Check for buffer overflow or encoding error */
+    if (len < 0 || len >= (int)sizeof(buffer)) {
+        return -1;
+    }
+
     return client_send(client, buffer, len);
 }
 
@@ -155,6 +161,10 @@ static int read_username(client_t *client) {
         } else {
             /* UTF-8 multi-byte */
             int len = utf8_byte_length(b);
+            if (len <= 0 || len > 4) {
+                /* Invalid UTF-8 start byte */
+                continue;
+            }
             buf[0] = b;
             if (len > 1) {
                 int read_bytes = ssh_channel_read(client->channel, &buf[1], len - 1, 0);
@@ -163,6 +173,11 @@ static int read_username(client_t *client) {
                     continue;
                 }
             }
+            /* Validate the complete UTF-8 sequence */
+            if (!utf8_is_valid_sequence(buf, len)) {
+                /* Invalid UTF-8 sequence */
+                continue;
+            }
             if (pos + len < MAX_USERNAME_LEN - 1) {
                 memcpy(username + pos, buf, len);
                 pos += len;
@@ -175,7 +190,8 @@ static int read_username(client_t *client) {
     client_printf(client, "\r\n");
 
     if (username[0] == '\0') {
-        strcpy(client->username, "anonymous");
+        strncpy(client->username, "anonymous", MAX_USERNAME_LEN - 1);
+        client->username[MAX_USERNAME_LEN - 1] = '\0';
     } else {
         strncpy(client->username, username, MAX_USERNAME_LEN - 1);
         /* Truncate to 20 characters */
@@ -420,7 +436,8 @@ void* client_handle_session(void *arg) {
     message_t join_msg = {
         .timestamp = time(NULL),
     };
-    strcpy(join_msg.username, "系统");
+    strncpy(join_msg.username, "系统", MAX_USERNAME_LEN - 1);
+    join_msg.username[MAX_USERNAME_LEN - 1] = '\0';
     snprintf(join_msg.content, MAX_MESSAGE_LEN, "%s 加入了聊天室", client->username);
     room_broadcast(g_room, &join_msg);
 
@@ -472,6 +489,10 @@ void* client_handle_session(void *arg) {
                     }
                 } else if (b >= 128) {  /* UTF-8 multi-byte */
                     int char_len = utf8_byte_length(b);
+                    if (char_len <= 0 || char_len > 4) {
+                        /* Invalid UTF-8 start byte */
+                        continue;
+                    }
                     buf[0] = b;
                     if (char_len > 1) {
                         int read_bytes = ssh_channel_read(client->channel, &buf[1], char_len - 1, 0);
@@ -480,6 +501,11 @@ void* client_handle_session(void *arg) {
                             continue;
                         }
                     }
+                    /* Validate the complete UTF-8 sequence */
+                    if (!utf8_is_valid_sequence(buf, char_len)) {
+                        /* Invalid UTF-8 sequence */
+                        continue;
+                    }
                     int len = strlen(input);
                     if (len + char_len < MAX_MESSAGE_LEN - 1) {
                         memcpy(input + len, buf, char_len);
@@ -507,7 +533,8 @@ cleanup:
         message_t leave_msg = {
             .timestamp = time(NULL),
         };
-        strcpy(leave_msg.username, "系统");
+        strncpy(leave_msg.username, "系统", MAX_USERNAME_LEN - 1);
+        leave_msg.username[MAX_USERNAME_LEN - 1] = '\0';
         snprintf(leave_msg.content, MAX_MESSAGE_LEN, "%s 离开了聊天室", client->username);
 
         client->connected = false;
diff --git a/src/utf8.c b/src/utf8.c
index 8afa505..f75db32 100644
--- a/src/utf8.c
+++ b/src/utf8.c
@@ -135,3 +135,55 @@ void utf8_remove_last_char(char *str) {
 
     str[i] = '\0';
 }
+
+/* Validate a UTF-8 byte sequence */
+bool utf8_is_valid_sequence(const char *bytes, int len) {
+    if (len <= 0 || len > 4 || !bytes) {
+        return false;
+    }
+
+    const unsigned char *b = (const unsigned char *)bytes;
+
+    /* Check first byte matches the expected length */
+    int expected_len = utf8_byte_length(b[0]);
+    if (expected_len != len) {
+        return false;
+    }
+
+    /* Validate continuation bytes (must be 10xxxxxx) */
+    for (int i = 1; i < len; i++) {
+        if ((b[i] & 0xC0) != 0x80) {
+            return false;
+        }
+    }
+
+    /* Validate codepoint ranges to prevent overlong encodings */
+    uint32_t codepoint = 0;
+    switch (len) {
+        case 1:
+            /* 0xxxxxxx - valid range: 0x00-0x7F */
+            codepoint = b[0];
+            if (codepoint > 0x7F) return false;
+            break;
+        case 2:
+            /* 110xxxxx 10xxxxxx - valid range: 0x80-0x7FF */
+            codepoint = ((b[0] & 0x1F) << 6) | (b[1] & 0x3F);
+            if (codepoint < 0x80 || codepoint > 0x7FF) return false;
+            break;
+        case 3:
+            /* 1110xxxx 10xxxxxx 10xxxxxx - valid range: 0x800-0xFFFF */
+            codepoint = ((b[0] & 0x0F) << 12) | ((b[1] & 0x3F) << 6) | (b[2] & 0x3F);
+            if (codepoint < 0x800 || codepoint > 0xFFFF) return false;
+            /* Reject UTF-16 surrogates (0xD800-0xDFFF) */
+            if (codepoint >= 0xD800 && codepoint <= 0xDFFF) return false;
+            break;
+        case 4:
+            /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx - valid range: 0x10000-0x10FFFF */
+            codepoint = ((b[0] & 0x07) << 18) | ((b[1] & 0x3F) << 12) |
+                       ((b[2] & 0x3F) << 6) | (b[3] & 0x3F);
+            if (codepoint < 0x10000 || codepoint > 0x10FFFF) return false;
+            break;
+    }
+
+    return true;
+}