fix(sse): anuncia role assistant no 1º delta do stream Responses→Chat

config/quality/file-size-baseline.json

@@ -8,6 +8,7 @@
   "_rebaseline_2026_06_15_3907_zai_validator": "PR #3907 own growth: validation.ts 4348->4394 (+46 = a zai (z.ai/glm) validator using directHttpsRequest to bypass the undici pool — same anti-zombie-socket pattern as nvidia #3226 — since z.ai drops idle keep-alive sockets without TCP RST after 502s. Anthropic wire format, x-api-key, 401/403->invalid else->valid). Cohesive validator branch; not extractable. (The global keepAlive 4000->1000 change from the original PR was reverted on review — scoped to the zai validator only.)",
   "_rebaseline_2026_06_15_3906_proxy_direct_fallback": "PR #3906 own growth: oauth/[provider]/[action]/route.ts 916->918 (+2 = swap runWithProxyContext -> runWithProxyContextOrDirect at the control-plane callsites so a dead pinned proxy degrades to a direct connection instead of a 500; the longer call name + a prettier reflow add the lines). Cohesive control-plane change; not extractable.",
   "_rebaseline_2026_06_15_3871_empty_pool": "PR #3871 own growth: combo.ts 5203->5204 (+1 = guard expandAutoComboCandidatePool against an empty candidatePool array — Array.isArray(pool) && pool.length > 0 so [] falls through to active-connection expansion instead of early-returning). One-line correctness fix; not extractable.",
+  "_rebaseline_2026_06_15_3911_sse_role": "PR #3911 own growth: openai-responses.ts 878->903 (+25 = withAssistantRoleOnFirstDelta wrapper that synthesizes role=assistant on the first Responses->Chat delta so strict streaming clients (langchain/n8n) do not drop tool_call deltas). Cohesive translator fix; not extractable.",
   "cap": 800,
   "frozen": {
     "open-sse/config/providerRegistry.ts": 4708,
@@ -41,7 +42,7 @@
     "open-sse/services/tokenRefresh.ts": 1997,
     "open-sse/services/usage.ts": 3408,
     "open-sse/translator/request/openai-to-gemini.ts": 844,
-    "open-sse/translator/response/openai-responses.ts": 878,
+    "open-sse/translator/response/openai-responses.ts": 903,
     "open-sse/utils/cursorAgentProtobuf.ts": 1521,
     "open-sse/utils/stream.ts": 2710,
     "src/app/(dashboard)/dashboard/HomePageClient.tsx": 1385,

open-sse/translator/response/openai-responses.ts

@@ -490,11 +490,36 @@ function normalizeUpstreamFailure(data, fallbackType = "server_error") {
   };
 }
 
+/**
+ * OpenAI Chat Completions streams announce the assistant role on the FIRST delta
+ * (e.g. `{ "role": "assistant", "content": "" }` or `{ "role": "assistant",
+ * "tool_calls": [...] }`). The Responses API has no role-announcement event, so when
+ * translating Responses → Chat we must synthesize it on the first emitted chunk.
+ *
+ * Strict streaming clients — notably @langchain/openai's `_convertDeltaToMessageChunk`
+ * (used by n8n's AI Agent) — key off the first chunk's role to build an AIMessageChunk.
+ * Without it, streamed tool_call deltas are dropped and the agent returns an empty
+ * response, even though the underlying tool call is well-formed.
+ */
+function withAssistantRoleOnFirstDelta(state, result) {
+  if (!result || state.roleEmitted) return result;
+  const delta = result.choices?.[0]?.delta;
+  if (delta && typeof delta === "object" && !Array.isArray(delta)) {
+    delta.role = "assistant";
+    state.roleEmitted = true;
+  }
+  return result;
+}
+
 /**
  * Translate OpenAI Responses API chunk to OpenAI Chat Completions format
  * This is for when Codex returns data and we need to send it to an OpenAI-compatible client
  */
 export function openaiResponsesToOpenAIResponse(chunk, state) {
+  return withAssistantRoleOnFirstDelta(state, openaiResponsesToOpenAIResponseStream(chunk, state));
+}
+
+function openaiResponsesToOpenAIResponseStream(chunk, state) {
   if (!chunk) {
     // Flush: send final chunk with finish_reason
     if (!state.finishReasonSent && state.started) {

tests/unit/responses-chat-assistant-role-first-chunk.test.ts

@@ -0,0 +1,70 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+const { openaiResponsesToOpenAIResponse } =
+  await import("../../open-sse/translator/response/openai-responses.ts");
+
+/**
+ * Regression: Responses → Chat Completions streaming must announce the assistant
+ * role on the FIRST emitted delta.
+ *
+ * OpenAI Chat Completions streams put `role: "assistant"` on the first chunk's
+ * delta. The Responses API has no role-announcement event, so the translator must
+ * synthesize it. Strict streaming clients — notably @langchain/openai's
+ * `_convertDeltaToMessageChunk` (used by n8n's AI Agent) — key off the first
+ * chunk's role to build an AIMessageChunk. Without it, streamed tool_call deltas
+ * are dropped and the agent receives an empty response.
+ */
+
+test("Responses->Chat: first tool_call chunk announces role=assistant", () => {
+  const state: Record<string, unknown> = {};
+
+  const first = openaiResponsesToOpenAIResponse(
+    {
+      type: "response.output_item.added",
+      item: {
+        type: "function_call",
+        call_id: "call_abc",
+        name: "get_weather",
+        arguments: "",
+      },
+    },
+    state
+  );
+
+  assert.ok(first, "should emit a chunk for output_item.added");
+  assert.equal(
+    first.choices[0].delta.role,
+    "assistant",
+    "first emitted delta must carry role=assistant"
+  );
+  assert.equal(first.choices[0].delta.tool_calls[0].function.name, "get_weather");
+
+  // Subsequent argument deltas must NOT repeat the role announcement.
+  const next = openaiResponsesToOpenAIResponse(
+    { type: "response.function_call_arguments.delta", delta: '{"x":1}' },
+    state
+  );
+  assert.ok(next, "should emit a chunk for arguments.delta");
+  assert.equal(next.choices[0].delta.role, undefined, "only the first delta announces the role");
+});
+
+test("Responses->Chat: first text chunk announces role=assistant", () => {
+  const state: Record<string, unknown> = {};
+
+  const first = openaiResponsesToOpenAIResponse(
+    { type: "response.output_text.delta", delta: "Olá" },
+    state
+  );
+
+  assert.ok(first, "should emit a chunk for output_text.delta");
+  assert.equal(first.choices[0].delta.role, "assistant");
+  assert.equal(first.choices[0].delta.content, "Olá");
+
+  const next = openaiResponsesToOpenAIResponse(
+    { type: "response.output_text.delta", delta: " mundo" },
+    state
+  );
+  assert.equal(next.choices[0].delta.role, undefined);
+  assert.equal(next.choices[0].delta.content, " mundo");
+});