Agent: Max Output Tokens + Thinking deaktivieren einstellbar

Zwei neue Agent-Einstellungen (Frontend, DB-Migration additiv):
- MaxOutputTokens: hartes Ausgabelimit. CLI: CLAUDE_CODE_MAX_OUTPUT_TOKENS,
  API: max_tokens (statt hart 16384). 0 = aus.
- DisableThinking: CLI: CLAUDE_CODE_DISABLE_THINKING=1 + MAX_THINKING_TOKENS=0;
  entfernt CLAUDE_CODE_EFFORT_LEVEL (Explizit-Aus schlaegt Effort).
  API-Pfad ohne Aequivalent (Hint im UI).

Die einzigen zwei Parameter, die z.ai/GLM nachweislich honoriert
(Sampling-Parameter wie temperature/top_p werden dort still verworfen).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Endpoints/AgentEndpoint.cs

@@ -72,6 +72,8 @@ public static class AgentEndpoint
                 a.AllowUserChoice,
                 a.TaskInstruction,
                 a.Effort,
+                a.MaxOutputTokens,
+                a.DisableThinking,
                 a.AvatarScale,
                 a.AvatarOffsetX,
                 a.AvatarOffsetY,
@@ -87,6 +89,18 @@ public static class AgentEndpoint
         return results;
     }
 
+    /// <summary>
+    /// Max-Output-Tokens robust parsen: Zahl, Zahl-als-String oder leer/null (= 0 = aus).
+    /// Negative Werte werden auf 0 geklemmt; das Frontend schickt bei leerem Feld null.
+    /// </summary>
+    private static int ParseMaxOutputTokens(JsonElement el)
+    {
+        int val = 0;
+        if (el.ValueKind == JsonValueKind.Number) el.TryGetInt32(out val);
+        else if (el.ValueKind == JsonValueKind.String) int.TryParse(el.GetString(), out val);
+        return val > 0 ? val : 0;
+    }
+
     private static async Task HandleGet(HttpContext ctx, DatabaseService db)
     {
         var agentId = ctx.Request.RouteValues["agentId"]?.ToString();
@@ -139,6 +153,8 @@ public static class AgentEndpoint
             AllowUserChoice = !body.TryGetProperty("allowUserChoice", out var auc) || auc.GetBoolean(),
             TaskInstruction = body.TryGetProperty("taskInstruction", out var ti) ? ti.GetString() : null,
             Effort = body.TryGetProperty("effort", out var ef) ? ef.GetString() ?? "" : "",
+            MaxOutputTokens = body.TryGetProperty("maxOutputTokens", out var mot) ? ParseMaxOutputTokens(mot) : 0,
+            DisableThinking = body.TryGetProperty("disableThinking", out var dt) && dt.ValueKind == JsonValueKind.True,
             AvatarScale = body.TryGetProperty("avatarScale", out var asc) ? asc.GetDouble() : 1.0,
             AvatarOffsetX = body.TryGetProperty("avatarOffsetX", out var aox) ? aox.GetDouble() : 0,
             AvatarOffsetY = body.TryGetProperty("avatarOffsetY", out var aoy) ? aoy.GetDouble() : 0,
@@ -187,6 +203,8 @@ public static class AgentEndpoint
         if (body.TryGetProperty("allowUserChoice", out var auc)) agent.AllowUserChoice = auc.GetBoolean();
         if (body.TryGetProperty("taskInstruction", out var ti)) agent.TaskInstruction = ti.GetString();
         if (body.TryGetProperty("effort", out var ef)) agent.Effort = ef.GetString() ?? "";
+        if (body.TryGetProperty("maxOutputTokens", out var mot)) agent.MaxOutputTokens = ParseMaxOutputTokens(mot);
+        if (body.TryGetProperty("disableThinking", out var dt)) agent.DisableThinking = dt.ValueKind == JsonValueKind.True;
         if (body.TryGetProperty("avatarScale", out var asc)) agent.AvatarScale = asc.GetDouble();
         if (body.TryGetProperty("avatarOffsetX", out var aox)) agent.AvatarOffsetX = aox.GetDouble();
         if (body.TryGetProperty("avatarOffsetY", out var aoy)) agent.AvatarOffsetY = aoy.GetDouble();

Endpoints/ChatEndpoint.cs

@@ -294,7 +294,7 @@ public static class ChatEndpoint
 
                     if (anthropic.IsCliAvailable)
                     {
-                        await anthropic.StreamChatViaCli(res, ctx.ModelId, ctx.SystemPrompt, messages, ctx.ChainId, workspace: ctx.Workspace, allowUserChoice: ctx.Agent?.AllowUserChoice ?? true, planMode: body.PlanMode, ultracode: body.Ultracode, webSearch: !string.IsNullOrEmpty(ctx.WebSearchVersion), sessionId: sessionId, agentId: ctx.AgentId, remoteMgr: remoteMgr, mcpServers: ctx.Session?.McpServers, useAutoContext: ctx.Agent?.UseAutoContext ?? true, effort: ctx.Effort, seedUsage: panelUsage, workCt: workCt, sseCt: sseCt);
+                        await anthropic.StreamChatViaCli(res, ctx.ModelId, ctx.SystemPrompt, messages, ctx.ChainId, workspace: ctx.Workspace, allowUserChoice: ctx.Agent?.AllowUserChoice ?? true, planMode: body.PlanMode, ultracode: body.Ultracode, webSearch: !string.IsNullOrEmpty(ctx.WebSearchVersion), sessionId: sessionId, agentId: ctx.AgentId, remoteMgr: remoteMgr, mcpServers: ctx.Session?.McpServers, useAutoContext: ctx.Agent?.UseAutoContext ?? true, effort: ctx.Effort, maxOutputTokens: ctx.MaxOutputTokens, disableThinking: ctx.DisableThinking, seedUsage: panelUsage, workCt: workCt, sseCt: sseCt);
                     }
                     else
                     {
@@ -304,7 +304,7 @@ public static class ChatEndpoint
                     return;
                 }
 
-                await anthropic.StreamChat(res, ctx.ModelId, ctx.SystemPrompt, messages, ctx.ChainId, ctx.Workspace, apiToken: anthropicToken, sessionId: sessionId, agentId: ctx.AgentId, userId: userId, useMemory: ctx.UseMemory, webSearchVersion: ctx.WebSearchVersion, maxIterations: ctx.MaxIterations, toolChoice: effectiveToolChoice, review: ctx.Review, finalizer: ctx.Finalizer, allowUserChoice: ctx.Agent?.AllowUserChoice ?? true, planMode: ctx.PlanMode, useAutoContext: ctx.Agent?.UseAutoContext ?? true, useGitHubSearch: ctx.Agent?.UseGitHubSearch ?? false, useSerperSearch: ctx.Agent?.UseSerperSearch ?? false, effort: ctx.Effort, seedUsage: panelUsage, workCt: workCt, hardCt: hardCt, sseCt: sseCt, remoteMgr: remoteMgr);
+                await anthropic.StreamChat(res, ctx.ModelId, ctx.SystemPrompt, messages, ctx.ChainId, ctx.Workspace, apiToken: anthropicToken, sessionId: sessionId, agentId: ctx.AgentId, userId: userId, useMemory: ctx.UseMemory, webSearchVersion: ctx.WebSearchVersion, maxIterations: ctx.MaxIterations, toolChoice: effectiveToolChoice, review: ctx.Review, finalizer: ctx.Finalizer, allowUserChoice: ctx.Agent?.AllowUserChoice ?? true, planMode: ctx.PlanMode, useAutoContext: ctx.Agent?.UseAutoContext ?? true, useGitHubSearch: ctx.Agent?.UseGitHubSearch ?? false, useSerperSearch: ctx.Agent?.UseSerperSearch ?? false, effort: ctx.Effort, maxOutputTokens: ctx.MaxOutputTokens, seedUsage: panelUsage, workCt: workCt, hardCt: hardCt, sseCt: sseCt, remoteMgr: remoteMgr);
             }
         }
         finally