open-agents-ai 0.103.73 → 0.103.74

package/dist/index.js

@@ -12394,6 +12394,7 @@ async function handleCmd(cmd) {
           }
           if (args.temperature !== undefined) riData.temperature = Number(args.temperature);
           if (args.max_tokens) riData.max_tokens = Number(args.max_tokens);
+          if (args.think !== undefined) riData.think = args.think === 'true' || args.think === true;
         } else {
           riData = { prompt: riPrompt };
         }
@@ -12984,6 +12985,7 @@ async function handleCmd(cmd) {
                   }
                   if (parsedReq.temperature !== undefined) chatBody.temperature = parsedReq.temperature;
                   if (parsedReq.max_tokens) chatBody.max_tokens = parsedReq.max_tokens;
+                  if (parsedReq.think !== undefined) chatBody.think = parsedReq.think;
 
                   var chatHeaders = { 'Content-Type': 'application/json' };
                   if (isPassthrough && endpointAuth) chatHeaders['Authorization'] = 'Bearer ' + endpointAuth;
@@ -18637,6 +18639,7 @@ var init_agenticRunner = __esm({
           deepContext: options?.deepContext ?? false,
           dynamicContext: options?.dynamicContext ?? "",
           streamEnabled: options?.streamEnabled ?? false,
+          thinking: options?.thinking ?? true,
           bruteForce: options?.bruteForce ?? true,
           bruteForceMaxCycles: options?.bruteForceMaxCycles ?? 100,
           modelTier: options?.modelTier ?? "large",
@@ -20953,10 +20956,12 @@ ${transcript}`
       baseUrl;
       model;
       apiKey;
-      constructor(baseUrl, model, apiKey) {
+      thinking;
+      constructor(baseUrl, model, apiKey, thinking) {
         this.baseUrl = normalizeBaseUrl(baseUrl);
         this.model = model;
         this.apiKey = apiKey ?? "";
+        this.thinking = thinking ?? true;
       }
       /** Build auth headers — all providers use standard Bearer token auth. */
       authHeaders() {
@@ -20973,7 +20978,7 @@ ${transcript}`
           tools: request.tools,
           temperature: request.temperature,
           max_tokens: request.maxTokens,
-          think: false
+          think: this.thinking
         };
         const resp = await fetch(`${this.baseUrl}/v1/chat/completions`, {
           method: "POST",
@@ -21023,7 +21028,7 @@ ${transcript}`
       }
       /**
        * SSE streaming variant — yields StreamChunks as tokens arrive.
-       * Uses `stream: true` and `think: true` so thinking tokens are visible.
+       * Uses `stream: true` and the current thinking setting.
        * The existing chatCompletion() method is completely unmodified.
        */
       async *chatCompletionStream(request) {
@@ -21035,7 +21040,7 @@ ${transcript}`
           max_tokens: request.maxTokens,
           stream: true,
           stream_options: { include_usage: true },
-          think: true
+          think: this.thinking
         };
         const resp = await fetch(`${this.baseUrl}/v1/chat/completions`, {
           method: "POST",
@@ -21121,11 +21126,13 @@ var init_nexusBackend = __esm({
       authKey;
       consecutiveFailures = 0;
       static MAX_CONSECUTIVE_FAILURES = 3;
-      constructor(sendFn, model, targetPeer, authKey) {
+      thinking;
+      constructor(sendFn, model, targetPeer, authKey, thinking) {
         this.sendFn = sendFn;
         this.model = model;
         this.targetPeer = targetPeer || "";
         this.authKey = authKey || "";
+        this.thinking = thinking ?? true;
       }
       async chatCompletion(request) {
         if (this.consecutiveFailures >= _NexusAgenticBackend.MAX_CONSECUTIVE_FAILURES) {
@@ -21146,6 +21153,7 @@ var init_nexusBackend = __esm({
         if (this.authKey) {
           daemonArgs.auth_key = this.authKey;
         }
+        daemonArgs.think = String(this.thinking);
         let rawResult;
         try {
           rawResult = await this.sendFn("remote_infer", daemonArgs, request.timeoutMs || 12e4);
@@ -26944,6 +26952,7 @@ function renderSlashHelp() {
     ["/voice clone <file>", "Set voice clone reference audio (wav/mp3/ogg/flac)"],
     ["/voice clone glados", "Generate clone ref from GLaDOS for LuxTTS"],
     ["/voice clone overwatch", "Generate clone ref from Overwatch for LuxTTS"],
+    ["/think", "Toggle thinking/reasoning mode (Qwen3, DeepSeek-R1, etc.)"],
     ["/stream", "Toggle real-time token streaming (pastel syntax highlighting)"],
     ["/dream", "Start dream mode \u2014 creative idle exploration (writes to .oa/dreams/)"],
     ["/dream deep", "Deep dream \u2014 multi-cycle exploration with sleep architecture"],
@@ -34148,6 +34157,13 @@ async function handleSlashCommand(input, ctx) {
       renderInfo(`Token streaming: ${isOn ? "on" : "off"}${hasLocal ? " (project-local)" : ""}` + (isOn ? " \u2014 thinking tokens in grey italics, responses with pastel syntax highlighting" : ""));
       return "handled";
     }
+    case "think": {
+      const isOn = ctx.thinkToggle();
+      const save = hasLocal ? ctx.saveLocalSettings.bind(ctx) : ctx.saveSettings.bind(ctx);
+      save({ thinking: isOn });
+      renderInfo(`Thinking mode: ${isOn ? "on" : "off"}${hasLocal ? " (project-local)" : ""}` + (isOn ? " \u2014 models that support reasoning (Qwen3, DeepSeek-R1, etc.) will show their thinking chain" : " \u2014 reasoning chain suppressed, model responds directly"));
+      return "handled";
+    }
     case "tools": {
       const tools = listCustomToolFiles(ctx.repoRoot);
       if (tools.length === 0) {
@@ -46637,7 +46653,7 @@ function formatDMNToolArgs(toolName, args) {
       return "";
   }
 }
-function startTask(task, config, repoRoot, voice, stream, taskStores, bruteForce, statusBar, sudoCallback, costTracker, onComplete, taskType, contextWindowSize, modelCaps, personality, deepContext, onCompaction, emotionEngine, flowEnabled, slashCommandHandler) {
+function startTask(task, config, repoRoot, voice, stream, taskStores, bruteForce, statusBar, sudoCallback, costTracker, onComplete, taskType, contextWindowSize, modelCaps, personality, deepContext, onCompaction, emotionEngine, flowEnabled, slashCommandHandler, thinkingEnabled) {
   const voiceStyleMap = {
     concise: 1,
     balanced: 3,
@@ -46685,9 +46701,9 @@ function startTask(task, config, repoRoot, voice, stream, taskStores, bruteForce
       return baseSendFn(cmd, args, timeout);
     };
     const targetPeer = config.backendUrl.startsWith("peer://") ? config.backendUrl.slice(7) : void 0;
-    backend = new NexusAgenticBackend(sendFn, config.model, targetPeer, config.apiKey);
+    backend = new NexusAgenticBackend(sendFn, config.model, targetPeer, config.apiKey, thinkingEnabled);
   } else {
-    backend = new OllamaAgenticBackend(config.backendUrl, config.model, config.apiKey);
+    backend = new OllamaAgenticBackend(config.backendUrl, config.model, config.apiKey, thinkingEnabled);
   }
   try {
     const endpointHistory = loadUsageHistory("endpoint", repoRoot);
@@ -46753,6 +46769,7 @@ function startTask(task, config, repoRoot, voice, stream, taskStores, bruteForce
     dynamicContext,
     modelTier,
     streamEnabled: stream?.enabled ?? false,
+    thinking: thinkingEnabled,
     bruteForce: bruteForce ?? true,
     bruteForceMaxCycles: 100,
     // effectively unlimited — no hard timeout, agent runs until complete or aborted
@@ -46867,6 +46884,7 @@ ${entry.fullContent}`
           "update",
           "upgrade",
           "parallel",
+          "think",
           "telegram",
           "tg",
           "call",
@@ -47215,6 +47233,7 @@ async function startInteractive(config, repoPath) {
   if (savedSettings.dbPath)
     config = { ...config, dbPath: savedSettings.dbPath };
   let streamEnabled = savedSettings.stream ?? false;
+  let thinkingEnabled = savedSettings.thinking !== false;
   let bruteForceEnabled = savedSettings.bruteforce ?? true;
   let currentStyle = PRESET_NAMES.includes(savedSettings.style) ? savedSettings.style : "balanced";
   let deepContextEnabled = savedSettings.deepContext ?? false;
@@ -47896,6 +47915,10 @@ Rationale: ${proposal.rationale}${provenanceNote}`;
       streamEnabled = !streamEnabled;
       return streamEnabled;
     },
+    thinkToggle() {
+      thinkingEnabled = !thinkingEnabled;
+      return thinkingEnabled;
+    },
     bruteForceToggle() {
       bruteForceEnabled = !bruteForceEnabled;
       return bruteForceEnabled;
@@ -49049,7 +49072,7 @@ Execute this skill now. Follow the behavioral guidance above.`;
           }, bruteForceEnabled, statusBar, handleSudoRequest, costTracker, (summary, meta) => {
             lastCompletedSummary = summary;
             lastTaskMeta = meta ?? null;
-          }, currentTaskType, resolvedContextWindowSize, resolvedCaps, currentStyle, deepContextEnabled, compactionSNRCallback(), emotionEngine, flowEnabled, buildSlashCommandHandler());
+          }, currentTaskType, resolvedContextWindowSize, resolvedCaps, currentStyle, deepContextEnabled, compactionSNRCallback(), emotionEngine, flowEnabled, buildSlashCommandHandler(), thinkingEnabled);
           activeTask = task;
           showPrompt();
           await task.promise;
@@ -49281,7 +49304,7 @@ NEW TASK: ${fullInput}`;
       }, bruteForceEnabled, statusBar, handleSudoRequest, costTracker, (summary, meta) => {
         lastCompletedSummary = summary;
         lastTaskMeta = meta ?? null;
-      }, currentTaskType, resolvedContextWindowSize, resolvedCaps, currentStyle, deepContextEnabled, compactionSNRCallback(), emotionEngine, flowEnabled, buildSlashCommandHandler());
+      }, currentTaskType, resolvedContextWindowSize, resolvedCaps, currentStyle, deepContextEnabled, compactionSNRCallback(), emotionEngine, flowEnabled, buildSlashCommandHandler(), thinkingEnabled);
       activeTask = task;
       showPrompt();
       await task.promise;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.103.73",
+  "version": "0.103.74",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",