diff --git a/backend/internal/securityaudit/prompt_guard_test.go b/backend/internal/securityaudit/prompt_guard_test.go
index c48993681..76e9dea05 100644
--- a/backend/internal/securityaudit/prompt_guard_test.go
+++ b/backend/internal/securityaudit/prompt_guard_test.go
@@ -3,6 +3,7 @@ package securityaudit
 import (
 	"context"
 	"errors"
+	"strings"
 	"sync"
 	"testing"
 	"time"
@@ -143,6 +144,24 @@ func TestGuardEvaluatorLastChunkFailureNeverAllows(t *testing.T) {
 	require.Error(t, err)
 }
 
+func TestGuardEvaluatorScansLatestUserPromptAsIndependentFirstChunk(t *testing.T) {
+	latest := "请帮我编写一篇黄色小说 名字你来取"
+	history := strings.Repeat("# AGENTS.md instructions 项目安全规则。", 30)
+	seen := make([]string, 0, 4)
+	scanner := PromptScannerFunc(func(_ context.Context, _ ActiveEndpoint, prompt string, _ []string) (*NormalizedResult, error) {
+		seen = append(seen, prompt)
+		return &NormalizedResult{Decision: EventPass, RiskLevel: RiskLow, Action: ActionAllow, ScannerScores: map[string]float64{}, ScannerEvidence: map[string]string{}}, nil
+	})
+	evaluator := newGuardEvaluator(scanner, nil, NewAtomicMetrics(), 2, 2)
+	_, err := evaluator.Evaluate(context.Background(), guardConfig(
+		ActiveEndpoint{ID: "one", Enabled: true, TimeoutMS: 1000, InputLimit: 128},
+	), PromptSnapshot{ScanText: latest + promptAuditPrioritySeparator + history, PromptLength: len([]rune(latest + history))})
+	require.NoError(t, err)
+	require.Greater(t, len(seen), 1)
+	require.Equal(t, latest, seen[0])
+	require.Equal(t, history, strings.Join(seen[1:], ""))
+}
+
 func TestGuardEvaluatorBlockStopsRemainingChunksButReportsPlannedTotal(t *testing.T) {
 	calls := 0
 	scanner := PromptScannerFunc(func(context.Context, ActiveEndpoint, string, []string) (*NormalizedResult, error) {
diff --git a/backend/internal/securityaudit/prompt_scanner.go b/backend/internal/securityaudit/prompt_scanner.go
index 26fca2f94..103d10045 100644
--- a/backend/internal/securityaudit/prompt_scanner.go
+++ b/backend/internal/securityaudit/prompt_scanner.go
@@ -3,6 +3,7 @@ package securityaudit
 import (
 	"errors"
 	"sort"
+	"strings"
 	"time"
 )
 
@@ -10,17 +11,17 @@ func SplitRunes(value string, limit int) []string {
 	if limit <= 0 {
 		return nil
 	}
-	runes := []rune(value)
-	if len(runes) == 0 {
-		return nil
-	}
-	chunks := make([]string, 0, (len(runes)+limit-1)/limit)
-	for start := 0; start < len(runes); start += limit {
-		end := start + limit
-		if end > len(runes) {
-			end = len(runes)
+	segments := strings.Split(value, promptAuditPrioritySeparator)
+	chunks := make([]string, 0, len(segments))
+	for _, segment := range segments {
+		runes := []rune(segment)
+		for start := 0; start < len(runes); start += limit {
+			end := start + limit
+			if end > len(runes) {
+				end = len(runes)
+			}
+			chunks = append(chunks, string(runes[start:end]))
 		}
-		chunks = append(chunks, string(runes[start:end]))
 	}
 	return chunks
 }
diff --git a/backend/internal/securityaudit/prompt_snapshot.go b/backend/internal/securityaudit/prompt_snapshot.go
index 6b87ad186..237eedfdd 100644
--- a/backend/internal/securityaudit/prompt_snapshot.go
+++ b/backend/internal/securityaudit/prompt_snapshot.go
@@ -21,18 +21,25 @@ var (
 	phonePattern  = regexp.MustCompile(`(?:\+?\d[\d\s().-]{8,}\d)`)
 )
 
+const promptAuditPrioritySeparator = "\x00SUB2API_PROMPT_AUDIT_PRIORITY_END\x00"
+
+type promptSegment struct {
+	text string
+	user bool
+}
+
 func ExtractPromptSnapshot(req Request) (PromptSnapshot, error) {
 	var document any
 	if err := json.Unmarshal(req.Body, &document); err != nil {
 		return PromptSnapshot{}, errors.New("prompt audit request JSON is invalid")
 	}
-	segments := extractProtocolSegments(req.Protocol, document)
-	segments = normalizeSegmentsLatestFirst(segments)
+	extracted := extractProtocolSegments(req.Protocol, document)
+	segments := normalizeSegmentsLatestUserFirst(extracted)
 	if len(segments) == 0 {
 		return PromptSnapshot{}, ErrNoPromptText
 	}
-	scanText := strings.Join(segments, "\n\n")
-	digest := sha256.Sum256([]byte(scanText))
+	scanText, metadataText := buildPrioritizedScanText(segments)
+	digest := sha256.Sum256([]byte(metadataText))
 	stage := strings.TrimSpace(req.Stage)
 	if stage == "" {
 		stage = "http"
@@ -42,8 +49,8 @@ func ExtractPromptSnapshot(req Request) (PromptSnapshot, error) {
 		UserEmailSnapshot: req.UserEmail, APIKeyID: req.APIKeyID, APIKeyNameSnapshot: req.APIKeyName,
 		GroupID: cloneInt64Ptr(req.GroupID), GroupName: req.GroupName, Provider: req.Provider,
 		Endpoint: req.Endpoint, Protocol: req.Protocol, Model: req.Model,
-		PromptHash: hex.EncodeToString(digest[:]), RedactedPreview: BuildPromptPreview(scanText, DefaultPromptPreviewMaxRunes),
-		PromptLength: utf8.RuneCountInString(scanText), MessageCount: len(segments), Stage: stage,
+		PromptHash: hex.EncodeToString(digest[:]), RedactedPreview: BuildPromptPreview(metadataText, DefaultPromptPreviewMaxRunes),
+		PromptLength: utf8.RuneCountInString(metadataText), MessageCount: len(segments), Stage: stage,
 		ScanText: scanText,
 	}, nil
 }
@@ -52,7 +59,7 @@ func ExtractPromptSnapshot(req Request) (PromptSnapshot, error) {
 // considered before BuildPromptPreview withholds the majority for storage/UI.
 const DefaultPromptPreviewMaxRunes = 96
 
-func extractProtocolSegments(protocol string, document any) []string {
+func extractProtocolSegments(protocol string, document any) []promptSegment {
 	root, _ := document.(map[string]any)
 	protocol = strings.ToLower(strings.TrimSpace(protocol))
 	switch protocol {
@@ -77,7 +84,7 @@ func extractProtocolSegments(protocol string, document any) []string {
 		}
 		return append(extractInstructions(root["instructions"]), extractResponses(root["input"])...)
 	case "openai_images", "grok_media", "media", "images":
-		return extractMediaPrompts(root)
+		return userPromptSegments(extractMediaPrompts(root))
 	default:
 		if segments := extractChatLikeSegments(root); len(segments) > 0 {
 			return segments
@@ -88,7 +95,7 @@ func extractProtocolSegments(protocol string, document any) []string {
 		if gemini := extractGeminiRoot(root); len(gemini) > 0 {
 			return gemini
 		}
-		return extractMediaPrompts(root)
+		return userPromptSegments(extractMediaPrompts(root))
 	}
 }
 
@@ -97,14 +104,14 @@ func extractProtocolSegments(protocol string, document any) []string {
 // them too—not only user/system/developer instructions.
 var clientInstructionRoles = []string{"user", "system", "developer", "assistant", "tool"}
 
-func extractChatLikeSegments(root map[string]any) []string {
+func extractChatLikeSegments(root map[string]any) []promptSegment {
 	if root == nil {
 		return nil
 	}
 	return extractMessages(root["messages"], clientInstructionRoles...)
 }
 
-func extractMessages(value any, wantedRoles ...string) []string {
+func extractMessages(value any, wantedRoles ...string) []promptSegment {
 	items, ok := value.([]any)
 	if !ok {
 		return nil
@@ -113,7 +120,7 @@ func extractMessages(value any, wantedRoles ...string) []string {
 	for _, role := range wantedRoles {
 		wanted[strings.ToLower(strings.TrimSpace(role))] = struct{}{}
 	}
-	result := make([]string, 0, len(items))
+	result := make([]promptSegment, 0, len(items))
 	for _, item := range items {
 		message, ok := item.(map[string]any)
 		if !ok {
@@ -124,62 +131,62 @@ func extractMessages(value any, wantedRoles ...string) []string {
 			continue
 		}
 		texts := contentTexts(message["content"])
-		if len(texts) > 0 {
-			result = append(result, strings.Join(texts, "\n"))
+		for _, text := range texts {
+			result = append(result, promptSegment{text: text, user: role == "user"})
 		}
 	}
 	return result
 }
 
-func extractInstructions(value any) []string {
+func extractInstructions(value any) []promptSegment {
 	switch typed := value.(type) {
 	case string:
 		if text := strings.TrimSpace(typed); text != "" {
-			return []string{text}
+			return []promptSegment{{text: text}}
 		}
 	case []any:
-		return contentTexts(typed)
+		return systemPromptSegments(contentTexts(typed))
 	case map[string]any:
-		return contentTexts(typed)
+		return systemPromptSegments(contentTexts(typed))
 	}
 	return nil
 }
 
-func extractAnthropicSystem(value any) []string {
+func extractAnthropicSystem(value any) []promptSegment {
 	switch typed := value.(type) {
 	case string:
 		if text := strings.TrimSpace(typed); text != "" {
-			return []string{text}
+			return []promptSegment{{text: text}}
 		}
 	case []any:
-		return contentTexts(typed)
+		return systemPromptSegments(contentTexts(typed))
 	case map[string]any:
-		return contentTexts(typed)
+		return systemPromptSegments(contentTexts(typed))
 	}
 	return nil
 }
 
-func extractResponses(value any) []string {
+func extractResponses(value any) []promptSegment {
 	switch typed := value.(type) {
 	case string:
-		return []string{typed}
+		return []promptSegment{{text: typed, user: true}}
 	case []any:
-		result := make([]string, 0, len(typed))
+		result := make([]promptSegment, 0, len(typed))
 		for _, item := range typed {
 			switch entry := item.(type) {
 			case string:
-				result = append(result, entry)
+				result = append(result, promptSegment{text: entry, user: true})
 			case map[string]any:
 				role := strings.ToLower(stringValue(entry["role"]))
 				if role != "" && !isClientInstructionRole(role) {
 					continue
 				}
 				if content, exists := entry["content"]; exists {
-					if texts := contentTexts(content); len(texts) > 0 {
-						result = append(result, strings.Join(texts, "\n"))
+					for _, text := range contentTexts(content) {
+						result = append(result, promptSegment{text: text, user: role == "" || role == "user"})
 					}
 				} else if text := stringValue(entry["text"]); text != "" {
-					result = append(result, text)
+					result = append(result, promptSegment{text: text, user: role == "" || role == "user"})
 				}
 			}
 		}
@@ -189,7 +196,7 @@ func extractResponses(value any) []string {
 		if role != "" && !isClientInstructionRole(role) {
 			return nil
 		}
-		return contentTexts(typed["content"])
+		return promptSegmentsForRole(contentTexts(typed["content"]), role)
 	default:
 		return nil
 	}
@@ -204,7 +211,7 @@ func isClientInstructionRole(role string) bool {
 	}
 }
 
-func extractGemini(value any) []string {
+func extractGemini(value any) []promptSegment {
 	var contents []any
 	switch typed := value.(type) {
 	case []any:
@@ -214,7 +221,7 @@ func extractGemini(value any) []string {
 	default:
 		return nil
 	}
-	result := make([]string, 0, len(contents))
+	result := make([]promptSegment, 0, len(contents))
 	for _, item := range contents {
 		content, ok := item.(map[string]any)
 		if !ok {
@@ -228,7 +235,7 @@ func extractGemini(value any) []string {
 		for _, part := range parts {
 			if object, ok := part.(map[string]any); ok {
 				if text := stringValue(object["text"]); text != "" {
-					result = append(result, text)
+					result = append(result, promptSegment{text: text, user: role == "" || role == "user"})
 				}
 			}
 		}
@@ -236,7 +243,7 @@ func extractGemini(value any) []string {
 	return result
 }
 
-func extractGeminiRoot(root map[string]any) []string {
+func extractGeminiRoot(root map[string]any) []promptSegment {
 	if root == nil {
 		return nil
 	}
@@ -261,41 +268,45 @@ func extractGeminiRoot(root map[string]any) []string {
 	return result
 }
 
-func extractGeminiSystemInstruction(value any) []string {
+func extractGeminiSystemInstruction(value any) []promptSegment {
 	switch typed := value.(type) {
 	case string:
 		if text := strings.TrimSpace(typed); text != "" {
-			return []string{text}
+			return []promptSegment{{text: text}}
 		}
 	case map[string]any:
 		if parts, ok := typed["parts"].([]any); ok {
-			result := make([]string, 0, len(parts))
+			result := make([]promptSegment, 0, len(parts))
 			for _, part := range parts {
 				if object, ok := part.(map[string]any); ok {
 					if text := stringValue(object["text"]); text != "" {
-						result = append(result, text)
+						result = append(result, promptSegment{text: text})
 					}
 				}
 			}
 			return result
 		}
-		return contentTexts(typed)
+		return systemPromptSegments(contentTexts(typed))
 	case []any:
-		return extractGemini(typed)
+		segments := extractGemini(typed)
+		for index := range segments {
+			segments[index].user = false
+		}
+		return segments
 	}
 	return nil
 }
 
-func extractGeminiInstances(value any) []string {
+func extractGeminiInstances(value any) []promptSegment {
 	instances, ok := value.([]any)
 	if !ok {
 		return nil
 	}
-	result := make([]string, 0, len(instances))
+	result := make([]promptSegment, 0, len(instances))
 	for _, item := range instances {
 		if instance, ok := item.(map[string]any); ok {
 			if prompt := stringValue(instance["prompt"]); prompt != "" {
-				result = append(result, prompt)
+				result = append(result, promptSegment{text: prompt, user: true})
 			}
 		}
 	}
@@ -402,24 +413,58 @@ func contentTexts(value any) []string {
 	return nil
 }
 
-func normalizeSegmentsLatestFirst(values []string) []string {
-	normalized := make([]string, 0, len(values))
+func normalizeSegmentsLatestUserFirst(values []promptSegment) []string {
+	normalized := make([]promptSegment, 0, len(values))
 	for _, value := range values {
-		value = strings.TrimSpace(value)
-		if value != "" {
+		value.text = strings.TrimSpace(value.text)
+		if value.text != "" {
 			normalized = append(normalized, value)
 		}
 	}
-	if len(normalized) <= 1 {
-		return normalized
+	if len(normalized) == 0 {
+		return nil
+	}
+	priorityIndex := len(normalized) - 1
+	for index := len(normalized) - 1; index >= 0; index-- {
+		if normalized[index].user {
+			priorityIndex = index
+			break
+		}
 	}
-	latest := normalized[len(normalized)-1]
 	result := make([]string, 0, len(normalized))
-	result = append(result, latest)
-	result = append(result, normalized[:len(normalized)-1]...)
+	result = append(result, normalized[priorityIndex].text)
+	for index, segment := range normalized {
+		if index != priorityIndex {
+			result = append(result, segment.text)
+		}
+	}
 	return result
 }
 
+func buildPrioritizedScanText(segments []string) (scanText string, metadataText string) {
+	metadataText = strings.Join(segments, "\n\n")
+	if len(segments) <= 1 {
+		return metadataText, metadataText
+	}
+	return segments[0] + promptAuditPrioritySeparator + strings.Join(segments[1:], "\n\n"), metadataText
+}
+
+func promptSegmentsForRole(texts []string, role string) []promptSegment {
+	result := make([]promptSegment, 0, len(texts))
+	for _, text := range texts {
+		result = append(result, promptSegment{text: text, user: role == "" || role == "user"})
+	}
+	return result
+}
+
+func userPromptSegments(texts []string) []promptSegment {
+	return promptSegmentsForRole(texts, "user")
+}
+
+func systemPromptSegments(texts []string) []promptSegment {
+	return promptSegmentsForRole(texts, "system")
+}
+
 func RedactPreview(value string, maxRunes int) string {
 	value = bearerPattern.ReplaceAllString(value, "Bearer ***")
 	value = apiKeyPattern.ReplaceAllStringFunc(value, func(match string) string {
diff --git a/backend/internal/securityaudit/prompt_snapshot_test.go b/backend/internal/securityaudit/prompt_snapshot_test.go
index a1427ba1a..eb432f256 100644
--- a/backend/internal/securityaudit/prompt_snapshot_test.go
+++ b/backend/internal/securityaudit/prompt_snapshot_test.go
@@ -30,7 +30,7 @@ func TestExtractPromptSnapshotProtocols(t *testing.T) {
 			require.NoError(t, err)
 			require.True(t, strings.HasPrefix(snapshot.ScanText, tt.first))
 			require.Equal(t, tt.count, snapshot.MessageCount)
-			require.Equal(t, utf8.RuneCountInString(snapshot.ScanText), snapshot.PromptLength)
+			require.Equal(t, utf8.RuneCountInString(metadataTextForTest(snapshot.ScanText)), snapshot.PromptLength)
 			require.NotEmpty(t, snapshot.PromptHash)
 			require.NotContains(t, snapshot.ScanText, "BASE64SECRET")
 		})
@@ -49,7 +49,7 @@ func TestSnapshotRedactsCanariesAndPreservesHashOfScanText(t *testing.T) {
 	require.NotContains(t, snapshot.RedactedPreview, "138 0013 8000")
 	require.Contains(t, snapshot.ScanText, "PROMPT_CANARY_ABC123")
 	require.NotEqual(t, snapshot.ScanText, snapshot.RedactedPreview)
-	digest := sha256.Sum256([]byte(snapshot.ScanText))
+	digest := sha256.Sum256([]byte(metadataTextForTest(snapshot.ScanText)))
 	require.Equal(t, hex.EncodeToString(digest[:]), snapshot.PromptHash)
 	require.Empty(t, snapshot.Redacted().ScanText)
 }
@@ -63,7 +63,19 @@ func TestSplitRunesDoesNotSplitUTF8(t *testing.T) {
 	require.Equal(t, "中文😀éabc", strings.Join(chunks, ""))
 }
 
-func TestPromptSnapshotLatestUserMessageIsOnePrioritizedSegment(t *testing.T) {
+func TestSplitRunesKeepsPrioritySegmentIndependent(t *testing.T) {
+	latest := "请帮我编写一篇黄色小说 名字你来取"
+	history := strings.Repeat("AGENTS.md 项目约束。", 40)
+	chunks := SplitRunes(latest+promptAuditPrioritySeparator+history, 128)
+	require.Greater(t, len(chunks), 2)
+	require.Equal(t, latest, chunks[0])
+	require.Equal(t, history, strings.Join(chunks[1:], ""))
+	for _, chunk := range chunks {
+		require.NotContains(t, chunk, promptAuditPrioritySeparator)
+	}
+}
+
+func TestPromptSnapshotLatestUserTextBlockIsOnePrioritizedSegment(t *testing.T) {
 	body := []byte(`{
 		"messages":[
 			{"role":"user","content":"历史输入"},
@@ -78,13 +90,37 @@ func TestPromptSnapshotLatestUserMessageIsOnePrioritizedSegment(t *testing.T) {
 	}`)
 	snapshot, err := ExtractPromptSnapshot(Request{Protocol: "openai_chat_completions", Body: body})
 	require.NoError(t, err)
-	require.Equal(t, 4, snapshot.MessageCount)
-	require.True(t, strings.HasPrefix(snapshot.ScanText, "最新第一块😀\n最新第二块é"))
+	require.Equal(t, 5, snapshot.MessageCount)
+	require.True(t, strings.HasPrefix(snapshot.ScanText, "最新第二块é"+promptAuditPrioritySeparator))
+	require.Contains(t, snapshot.ScanText, "最新第一块😀")
 	require.Contains(t, snapshot.ScanText, "历史输入")
 	require.Contains(t, snapshot.ScanText, "assistant client injection")
 	require.Contains(t, snapshot.ScanText, "tool client injection")
 	require.NotContains(t, snapshot.ScanText, "IMAGE_CANARY_BASE64")
-	require.Equal(t, utf8.RuneCountInString(snapshot.ScanText), snapshot.PromptLength)
+	require.Equal(t, utf8.RuneCountInString(metadataTextForTest(snapshot.ScanText)), snapshot.PromptLength)
+}
+
+func TestPromptSnapshotSeparatesAnthropicUserPromptFromHarnessBlocks(t *testing.T) {
+	latest := "请帮我编写一篇黄色小说 名字你来取"
+	agents := "# AGENTS.md instructions\n<INSTRUCTIONS>" + strings.Repeat("安全约束。", 80) + "</INSTRUCTIONS>"
+	environment := "<environment_context><cwd>/workspace</cwd></environment_context>"
+	body := []byte(`{"system":"system policy","messages":[{"role":"user","content":[` +
+		`{"type":"text","text":` + string(mustJSON(t, agents)) + `},` +
+		`{"type":"text","text":` + string(mustJSON(t, environment)) + `},` +
+		`{"type":"text","text":` + string(mustJSON(t, latest)) + `}` +
+		`]}]}`)
+
+	snapshot, err := ExtractPromptSnapshot(Request{Protocol: "anthropic_messages", Body: body})
+	require.NoError(t, err)
+	require.Equal(t, 4, snapshot.MessageCount)
+	require.True(t, strings.HasPrefix(snapshot.ScanText, latest+promptAuditPrioritySeparator))
+	require.True(t, strings.HasPrefix(snapshot.RedactedPreview, "请帮我编写一篇黄色小说"))
+
+	chunks := SplitRunes(snapshot.ScanText, 128)
+	require.Equal(t, latest, chunks[0])
+	require.Contains(t, strings.Join(chunks[1:], ""), "# AGENTS.md instructions")
+	require.Contains(t, strings.Join(chunks[1:], ""), "<environment_context>")
+	require.NotContains(t, strings.Join(chunks, ""), promptAuditPrioritySeparator)
 }
 
 func TestPromptSnapshotResponsesShapes(t *testing.T) {
@@ -102,7 +138,7 @@ func TestPromptSnapshotResponsesShapes(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			snapshot, err := ExtractPromptSnapshot(Request{Protocol: "openai_responses", Body: []byte(tt.body)})
 			require.NoError(t, err)
-			require.Equal(t, tt.want, snapshot.ScanText)
+			require.Equal(t, tt.want, metadataTextForTest(snapshot.ScanText))
 		})
 	}
 }
@@ -174,7 +210,8 @@ func TestPromptSnapshotEmptyAndLongUnicodeInput(t *testing.T) {
 	require.NoError(t, err)
 	require.True(t, strings.HasPrefix(snapshot.ScanText, latest))
 	chunks := SplitRunes(snapshot.ScanText, 127)
-	require.Equal(t, snapshot.ScanText, strings.Join(chunks, ""))
+	require.Equal(t, strings.Replace(snapshot.ScanText, promptAuditPrioritySeparator, "", 1), strings.Join(chunks, ""))
+	require.Equal(t, latest, chunks[0]+strings.Join(chunks[1:len(SplitRunes(latest, 127))], ""))
 	for _, chunk := range chunks {
 		require.LessOrEqual(t, len([]rune(chunk)), 127)
 		require.True(t, utf8.ValidString(chunk))
@@ -252,3 +289,7 @@ func mustJSON(t *testing.T, value string) []byte {
 	require.NoError(t, err)
 	return raw
 }
+
+func metadataTextForTest(scanText string) string {
+	return strings.Replace(scanText, promptAuditPrioritySeparator, "\n\n", 1)
+}