From 9b794a8aec2ae95ab29f50feb6ef4817099bfcf6 Mon Sep 17 00:00:00 2001
From: Futureppo <futureppo666@gmail.com>
Date: Mon, 13 Jul 2026 14:30:50 +0800
Subject: [PATCH] Fix responses assistant history content

---
 internal/openai/responses_compat.go      | 49 ++++++++++++++++++
 internal/openai/responses_compat_test.go | 65 ++++++++++++++++++++++++
 2 files changed, 114 insertions(+)

diff --git a/internal/openai/responses_compat.go b/internal/openai/responses_compat.go
index cd65fec..869503e 100644
--- a/internal/openai/responses_compat.go
+++ b/internal/openai/responses_compat.go
@@ -202,9 +202,58 @@ func sanitizeInputItem(item map[string]any) map[string]any {
 	out := clone(item)
 	delete(out, "internal_chat_message_metadata_passthrough")
 	delete(out, "phase")
+	if content, exists := out["content"]; exists && String(out, "type", "") == "message" {
+		out["content"] = sanitizeMessageContent(content)
+	}
+	return out
+}
+
+func sanitizeMessageContent(content any) any {
+	parts, ok := content.([]any)
+	if !ok {
+		return content
+	}
+	out := make([]any, 0, len(parts))
+	for _, raw := range parts {
+		part, ok := raw.(map[string]any)
+		if !ok {
+			out = append(out, raw)
+			continue
+		}
+		clean := clone(part)
+		switch String(clean, "type", "") {
+		case "output_text":
+			clean["type"] = "input_text"
+		case "image":
+			if image := anthropicImageContent(clean); image != nil {
+				clean = image
+			}
+		}
+		out = append(out, clean)
+	}
 	return out
 }
 
+func anthropicImageContent(part map[string]any) map[string]any {
+	source, ok := part["source"].(map[string]any)
+	if !ok {
+		return nil
+	}
+	var url string
+	switch String(source, "type", "") {
+	case "url":
+		url = String(source, "url", "")
+	case "base64":
+		url = "data:" + String(source, "media_type", "") + ";base64," + String(source, "data", "")
+	default:
+		return nil
+	}
+	if url == "" {
+		return nil
+	}
+	return map[string]any{"type": "input_image", "image_url": url, "detail": "auto"}
+}
+
 func hasReasoningText(item map[string]any) bool {
 	for _, key := range []string{"summary", "content"} {
 		if parts, ok := item[key].([]any); ok && len(parts) > 0 {
diff --git a/internal/openai/responses_compat_test.go b/internal/openai/responses_compat_test.go
index 87c4003..65cd99e 100644
--- a/internal/openai/responses_compat_test.go
+++ b/internal/openai/responses_compat_test.go
@@ -126,6 +126,71 @@ func TestPrepareCompatibleResponsesDropsUnknownInput(t *testing.T) {
 	}
 }
 
+func TestPrepareCompatibleResponsesRewritesAssistantOutputTextHistory(t *testing.T) {
+	wire, _, err := PrepareCompatibleResponses(map[string]any{
+		"model": "grok-4.5",
+		"input": []any{
+			map[string]any{
+				"type": "message", "role": "assistant", "id": "msg_1", "status": "completed",
+				"content": []any{map[string]any{"type": "output_text", "text": "first response"}},
+			},
+			map[string]any{"type": "message", "role": "user", "content": []any{map[string]any{"type": "input_text", "text": "continue"}}},
+		},
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	input := wire["input"].([]any)
+	message := input[0].(map[string]any)
+	content := message["content"].([]any)
+	part := content[0].(map[string]any)
+	if part["type"] != "input_text" || part["text"] != "first response" {
+		t.Fatalf("assistant history content = %#v", content)
+	}
+}
+
+func TestPrepareCompatibleResponsesDoesNotAddMissingMessageContent(t *testing.T) {
+	wire, _, err := PrepareCompatibleResponses(map[string]any{
+		"model": "grok-4.5",
+		"input": []any{
+			map[string]any{"type": "message", "role": "assistant", "tool_calls": []any{map[string]any{"id": "call_1", "type": "function"}}},
+		},
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	input := wire["input"].([]any)
+	message := input[0].(map[string]any)
+	if _, exists := message["content"]; exists {
+		t.Fatalf("missing content was added: %#v", message)
+	}
+}
+
+func TestPrepareCompatibleResponsesRewritesAnthropicImageContent(t *testing.T) {
+	wire, _, err := PrepareCompatibleResponses(map[string]any{
+		"model": "grok-4.5",
+		"input": []any{
+			map[string]any{
+				"type": "message", "role": "user",
+				"content": []any{
+					map[string]any{"type": "input_text", "text": "describe"},
+					map[string]any{"type": "image", "source": map[string]any{"type": "base64", "media_type": "image/png", "data": "AAAA"}},
+				},
+			},
+		},
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	input := wire["input"].([]any)
+	message := input[0].(map[string]any)
+	content := message["content"].([]any)
+	image := content[1].(map[string]any)
+	if image["type"] != "input_image" || image["image_url"] != "data:image/png;base64,AAAA" {
+		t.Fatalf("image content = %#v", image)
+	}
+}
+
 func TestPrepareCompatibleResponsesNormalizesCodexHostedTools(t *testing.T) {
 	wire, _, err := PrepareCompatibleResponses(map[string]any{
 		"model": "grok-4.5", "input": "hello",