fix: DeepSeek reasoning_content dropped on multi-step tool calls

- Base URL corrected: zen/v1 -> zen/go/v1 (opencode-go provider)
- Model changed: deepseek-v4-flash-free -> deepseek-v4-flash
- Added wrapLanguageModel middleware to inject reasoning_content via
  providerMetadata.openaiCompatible before each stream call
- Fixed test mocks: removed vi.importActual (unsupported in Bun), added
  tool factory mocks, preserved real tool export in ai mock
- Added 11 tests for the normalizeMessages middleware

1 files changed, 66 insertions, 1 deletions

diff --git a/packages/core/src/llm/provider.ts b/packages/core/src/llm/provider.ts
index 06f7b72..2a917b2 100644
--- a/packages/core/src/llm/provider.ts
+++ b/packages/core/src/llm/provider.ts
@@ -1,9 +1,74 @@
 import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
+import { wrapLanguageModel } from "ai";
+import type { LanguageModelV1Middleware, LanguageModelV1Prompt } from "ai";
+
+/**
+ * Normalize messages for interleaved reasoning providers (DeepSeek).
+ * Extracts { type: "reasoning" } / { type: "redacted-reasoning" } parts from
+ * assistant message content arrays and moves them into
+ * providerMetadata.openaiCompatible.reasoning_content so
+ * @ai-sdk/openai-compatible serializes them correctly for the API.
+ *
+ * IMPORTANT: The messages in params.prompt are already in LanguageModelV1Prompt
+ * format (post-conversion by the AI SDK). They use `providerMetadata`, NOT
+ * `providerOptions` — that conversion happens before the middleware runs.
+ */
+function normalizeMessages(msgs: unknown[]): unknown[] {
+	return msgs.map((msg: unknown) => {
+		const message = msg as Record<string, unknown>;
+		if (message.role !== "assistant" || !Array.isArray(message.content)) return message;
+
+		const content = message.content as Array<Record<string, unknown>>;
+		const reasoningParts = content.filter(
+			(p) => p.type === "reasoning" || p.type === "redacted-reasoning",
+		);
+		const reasoningText = reasoningParts.map((p) => p.text ?? "").join("");
+
+		const filteredContent = content.filter(
+			(p) => p.type !== "reasoning" && p.type !== "redacted-reasoning",
+		);
+
+		const existingMetadata = (message.providerMetadata ?? {}) as Record<string, unknown>;
+		const existingOpenAICompat = (existingMetadata.openaiCompatible ?? {}) as Record<string, unknown>;
+
+		return {
+			...message,
+			content: filteredContent,
+			providerMetadata: {
+				...existingMetadata,
+				openaiCompatible: {
+					...existingOpenAICompat,
+					reasoning_content: reasoningText,
+				},
+			},
+		};
+	});
+}
 
 export function createProvider(config: { apiKey: string; baseURL: string }) {
-	return createOpenAICompatible({
+	const provider = createOpenAICompatible({
 		name: "opencode-zen",
 		apiKey: config.apiKey,
 		baseURL: config.baseURL,
 	});
+
+	return (modelId: string) => {
+		const middleware: LanguageModelV1Middleware = {
+			middlewareVersion: "v1" as const,
+			transformParams: async ({ type, params }) => {
+				if (type === "stream" && params.prompt) {
+					return {
+						...params,
+						prompt: normalizeMessages(params.prompt as unknown[]) as LanguageModelV1Prompt,
+					};
+				}
+				return params;
+			},
+		};
+
+		return wrapLanguageModel({
+			model: provider(modelId),
+			middleware: [middleware],
+		});
+	};
 }