nothumanallowed 11.3.0 → 11.5.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "nothumanallowed",
-  "version": "11.3.0",
+  "version": "11.5.0",
   "description": "NotHumanAllowed — 38 AI agents, 53 tools. Email, calendar, browser automation, screen capture, canvas, cron/heartbeat, GitHub, Notion, Slack, voice chat, 28 languages. Zero-dependency CLI.",
   "type": "module",
   "bin": {

package/src/config.mjs

@@ -283,6 +283,9 @@ export function setConfigValue(key, value) {
     'my-role': 'profile.role',
     'role': 'profile.role',
     'profile-notes': 'profile.notes',
+    'thinking': 'thinking',
+    'extended-thinking': 'thinking',
+    'language': 'language',
   };
 
   const resolved = aliases[key] || key;

package/src/constants.mjs

@@ -5,7 +5,7 @@ import { fileURLToPath } from 'url';
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = path.dirname(__filename);
 
-export const VERSION = '11.3.0';
+export const VERSION = '11.5.0';
 export const BASE_URL = 'https://nothumanallowed.com/cli';
 export const API_BASE = 'https://nothumanallowed.com/api/v1';
 

package/src/services/llm.mjs

@@ -234,18 +234,33 @@ export async function streamSSE(res, format) {
 
 /**
  * NHA Free (Liara) — free LLM tier, no API key required.
- * Connects to the NHA-hosted Qwen3 + LoRA on Hetzner GPU.
- * OpenAI-compatible format. Slower than paid providers (5-15 seconds).
+ * Qwen3 32B on Hetzner RTX 6000 Pro 96GB. Supports thinking mode.
  */
 export async function callNHA(apiKey, model, systemPrompt, userMessage, stream = false) {
+  // Read thinking preference from config
+  let thinkingEnabled = false; // OFF by default for speed
+  try {
+    const fs = await import('fs');
+    const path = await import('path');
+    const os = await import('os');
+    const cfgFile = path.default.join(os.default.homedir(), '.nha', 'config.json');
+    if (fs.default.existsSync(cfgFile)) {
+      const cfg = JSON.parse(fs.default.readFileSync(cfgFile, 'utf-8'));
+      if (cfg.thinking === true || cfg.thinking === 'on' || cfg.thinking === 'true') {
+        thinkingEnabled = true;
+      }
+    }
+  } catch {}
+
   const body = {
-    model: model || 'nha-v1',
-    max_tokens: 4096,
+    model: model || '/opt/models/qwen3-32b',
+    max_tokens: thinkingEnabled ? 8192 : 4096,
     messages: [
       { role: 'system', content: systemPrompt },
       { role: 'user', content: userMessage },
     ],
     stream,
+    chat_template_kwargs: { enable_thinking: thinkingEnabled },
   };
   const res = await fetch('https://liara.nothumanallowed.com/v1/chat/completions', {
     method: 'POST',
@@ -258,9 +273,12 @@ export async function callNHA(apiKey, model, systemPrompt, userMessage, stream =
     const err = await res.text();
     throw new Error(`NHA Free ${res.status}: ${err}`);
   }
-  if (stream) return streamSSE(res, 'openai'); // OpenAI-compatible SSE format
+  if (stream) return streamSSE(res, 'openai');
   const data = await res.json();
-  return data.choices?.[0]?.message?.content || '';
+  let content = data.choices?.[0]?.message?.content || '';
+  // Strip thinking tags if present
+  content = content.replace(/<think>[\s\S]*?<\/think>/g, '').trim();
+  return content;
 }
 
 const PROVIDERS = {
@@ -414,6 +432,22 @@ export async function callLLMStream(config, systemPrompt, userMessage, onToken,
   const callFn = getProviderCall(provider);
   if (!callFn) throw new Error(`Unknown provider: ${provider}`);
 
+  // NHA Free: add thinking config to streaming request
+  if (provider === 'nha') {
+    try {
+      const fs2 = await import('fs');
+      const path2 = await import('path');
+      const os2 = await import('os');
+      const cfgFile2 = path2.default.join(os2.default.homedir(), '.nha', 'config.json');
+      if (fs2.default.existsSync(cfgFile2)) {
+        const cfg2 = JSON.parse(fs2.default.readFileSync(cfgFile2, 'utf-8'));
+        const thinkOn = cfg2.thinking === true || cfg2.thinking === 'on' || cfg2.thinking === 'true';
+        // Will be added to body below via buildRequestBody override
+        opts._thinkingEnabled = thinkOn;
+      }
+    } catch {}
+  }
+
   // Gemini and Cohere don't support streaming — fall back to non-streaming
   if (provider === 'gemini' || provider === 'cohere') {
     const text = await callFn(apiKey, model, systemPrompt, userMessage, false);
@@ -452,13 +486,13 @@ function buildRequestBody(provider, model, systemPrompt, userMessage, stream) {
   }
   // OpenAI-compatible format (OpenAI, DeepSeek, Grok, Mistral)
   const modelDefaults = {
-    nha: 'nha-v1',
+    nha: '/opt/models/qwen3-32b',
     openai: 'gpt-4o',
     deepseek: 'deepseek-chat',
     grok: 'grok-3-latest',
     mistral: 'mistral-large-latest',
   };
-  return {
+  const req = {
     model: model || modelDefaults[provider] || 'gpt-4o',
     max_tokens: provider === 'nha' ? 4096 : 8192,
     messages: [
@@ -467,6 +501,11 @@ function buildRequestBody(provider, model, systemPrompt, userMessage, stream) {
     ],
     stream,
   };
+  // NHA: add thinking control
+  if (provider === 'nha') {
+    req.chat_template_kwargs = { enable_thinking: false };
+  }
+  return req;
 }
 
 /** Get provider API URL */

package/src/services/web-ui.mjs

@@ -1984,6 +1984,7 @@ function renderSettings(el) {
       ['provider', 'Provider', 'nha (free) / anthropic / openai / gemini / deepseek / grok / mistral'],
       ['key', 'API Key', 'Not needed for NHA Free', true],
       ['model', 'Model', 'Leave empty for default'],
+      ['thinking', 'Extended Thinking', 'on / off — Qwen3 reasoning mode (NHA Free only)'],
     ]) +
     settingsSection('responder', 'Message Responder', 'Auto-reply to Telegram and Discord messages.', [
       ['telegram-bot-token', 'Telegram Bot Token', 'Get from @BotFather', true],