Fold text into last tool message to preserve reasoning continuity

Author: capitanfeeder

src/api/providers/__tests__/mimo.spec.ts

@@ -376,7 +376,7 @@ describe("MimoHandler", () => {
 			expect(result[0].content).toBe("(empty)")
 		})
 
-		it("should separate tool_results from text in user messages", () => {
+		it("should merge text into last tool message when both exist in same turn", () => {
 			const messages: Anthropic.Messages.MessageParam[] = [
 				{
 					role: "user",
@@ -386,15 +386,28 @@ describe("MimoHandler", () => {
 							tool_use_id: "call_1",
 							content: "result",
 						},
-						{ type: "text" as const, text: "Here are the results" },
+						{ type: "text" as const, text: "<environment_details>..." },
 					],
 				},
 			]
 			const result = (handler as any).convertMessagesForMiMo(messages)
-			expect(result).toHaveLength(2)
+			expect(result).toHaveLength(1)
 			expect(result[0].role).toBe("tool")
-			expect(result[1].role).toBe("user")
-			expect(result[1].content).toBe("Here are the results")
+			expect(result[0].content).toContain("result")
+			expect(result[0].content).toContain("<environment_details>...")
+		})
+
+		it("should keep text as separate user message when no tool_results present", () => {
+			const messages: Anthropic.Messages.MessageParam[] = [
+				{
+					role: "user",
+					content: [{ type: "text" as const, text: "Hello" }],
+				},
+			]
+			const result = (handler as any).convertMessagesForMiMo(messages)
+			expect(result).toHaveLength(1)
+			expect(result[0].role).toBe("user")
+			expect(result[0].content).toBe("Hello")
 		})
 
 		it("should handle user message with string content", () => {

src/api/providers/mimo.ts

@@ -181,7 +181,6 @@ export class MimoHandler extends OpenAiHandler {
 					} else if (block.type === "text") {
 						textBlocks.push(block.text)
 					} else if (block.type === "image") {
-						// Convert Anthropic image block to OpenAI image_url format
 						const src = (block as any).source
 						if (src?.type === "base64" && src?.media_type) {
 							mediaParts.push({
@@ -195,10 +194,9 @@ export class MimoHandler extends OpenAiHandler {
 							})
 						}
 					}
-					// audio/video blocks are not supported in OpenAI chat format — skip silently
 				}
 
-				// Add tool results as role:"tool" messages (MiMo supports this)
+				// Add tool results as role:"tool" messages
 				for (const tr of toolResults) {
 					let content: string
 					if (typeof tr.content === "string") {
@@ -216,8 +214,18 @@ export class MimoHandler extends OpenAiHandler {
 					})
 				}
 
-				// Build user message content — plain string or multimodal array
-				if (mediaParts.length > 0) {
+				// When both tool_results and text exist in the same user turn (happens
+				// during resume/delegate flows where <environment_details> is appended),
+				// fold the text into the last tool message instead of creating a separate
+				// user message. A fresh role:"user" after tool results drops reasoning
+				// continuity for thinking models.
+				if (toolResults.length > 0 && textBlocks.length > 0 && mediaParts.length === 0) {
+					const lastToolIdx = converted.length - 1
+					const lastTool = converted[lastToolIdx] as any
+					if (lastTool?.role === "tool") {
+						lastTool.content += "\n" + textBlocks.join("\n")
+					}
+				} else if (mediaParts.length > 0) {
 					const content: any[] = []
 					if (textBlocks.length > 0) {
 						content.push({ type: "text", text: textBlocks.join("\n") })