add hermes agent provider

server/src/llm/chat-tools.ts

@@ -385,6 +385,10 @@ function normalizeIncomingMessages(messages: ChatMessage[]) {
   return [{ role: "system", content: CHAT_TOOL_SYSTEM_PROMPT }, ...normalized];
 }
 
+function normalizePlainIncomingMessages(messages: ChatMessage[]) {
+  return messages.map((message) => buildOpenAIConversationMessage(message));
+}
+
 function normalizeIncomingResponsesInput(messages: ChatMessage[]) {
   const normalized = messages.map((message) => buildOpenAIResponsesInputMessage(message));
 
@@ -853,6 +857,20 @@ function extractResponsesText(response: any, fallback = "") {
   return parts.join("") || fallback;
 }
 
+function extractChatCompletionContent(message: any) {
+  if (typeof message?.content === "string") return message.content;
+  if (!Array.isArray(message?.content)) return "";
+
+  return message.content
+    .map((part: any) => {
+      if (typeof part === "string") return part;
+      if (typeof part?.text === "string") return part.text;
+      if (typeof part?.content === "string") return part.content;
+      return "";
+    })
+    .join("");
+}
+
 function getUnstreamedText(finalText: string, streamedText: string) {
   if (!finalText) return "";
   if (!streamedText) return finalText;
@@ -1093,6 +1111,26 @@ export async function runToolAwareChatCompletions(params: ToolAwareCompletionPar
   };
 }
 
+export async function runPlainChatCompletions(params: ToolAwareCompletionParams): Promise<ToolAwareCompletionResult> {
+  const completion = await params.client.chat.completions.create({
+    model: params.model,
+    messages: normalizePlainIncomingMessages(params.messages),
+    temperature: params.temperature,
+    max_tokens: params.maxTokens,
+  } as any);
+
+  const usageAcc: Required<ToolAwareUsage> = { inputTokens: 0, outputTokens: 0, totalTokens: 0 };
+  const sawUsage = mergeUsage(usageAcc, completion?.usage);
+  const message = completion?.choices?.[0]?.message;
+
+  return {
+    text: extractChatCompletionContent(message),
+    usage: sawUsage ? usageAcc : undefined,
+    raw: { response: completion, api: "chat.completions" },
+    toolEvents: [],
+  };
+}
+
 export async function* runToolAwareOpenAIChatStream(
   params: ToolAwareCompletionParams
 ): AsyncGenerator<ToolAwareStreamingEvent> {
@@ -1354,3 +1392,41 @@ export async function* runToolAwareChatCompletionsStream(
     },
   };
 }
+
+export async function* runPlainChatCompletionsStream(
+  params: ToolAwareCompletionParams
+): AsyncGenerator<ToolAwareStreamingEvent> {
+  const rawResponses: unknown[] = [];
+  const usageAcc: Required<ToolAwareUsage> = { inputTokens: 0, outputTokens: 0, totalTokens: 0 };
+  let sawUsage = false;
+  let text = "";
+
+  const stream = await params.client.chat.completions.create({
+    model: params.model,
+    messages: normalizePlainIncomingMessages(params.messages),
+    temperature: params.temperature,
+    max_tokens: params.maxTokens,
+    stream: true,
+  } as any);
+
+  for await (const chunk of stream as any as AsyncIterable<any>) {
+    rawResponses.push(chunk);
+    sawUsage = mergeUsage(usageAcc, chunk?.usage) || sawUsage;
+
+    const deltaText = chunk?.choices?.[0]?.delta?.content ?? "";
+    if (typeof deltaText === "string" && deltaText.length) {
+      text += deltaText;
+      yield { type: "delta", text: deltaText };
+    }
+  }
+
+  yield {
+    type: "done",
+    result: {
+      text,
+      usage: sawUsage ? usageAcc : undefined,
+      raw: { streamed: true, responses: rawResponses, api: "chat.completions" },
+      toolEvents: [],
+    },
+  };
+}