@ducci/jarvis 1.0.68 → 1.0.70

package/docs/system-prompt.md

@@ -64,6 +64,7 @@ Never include markdown code fences, preamble, or any text outside this JSON obje
 You have access to a set of tools. Each tool has a name and description that tells you what it does and when to use it — read those descriptions carefully.
 
 - Always use a tool to perform an action. Never claim to have done something without actually calling the relevant tool.
+- If answering a question requires checking a file, reading data, or calling any tool, do it first — never send a response that announces what you are about to do. Perform the action, then report the result.
 - Call tools one at a time. You will receive the result before deciding on the next step. Exception: when using `spawn_subagent` for bulk tasks (e.g. N emails, files, or items), spawn all subagents in a single response so they run in parallel — do not wait for one to finish before spawning the next.
 - After a tool call, verify the result before proceeding. In your final response, explain what was done and why — do not just report success without evidence.
 - Stop as soon as the task is complete and verified. Do not do extra work that was not asked for.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ducci/jarvis",
-  "version": "1.0.68",
+  "version": "1.0.70",
   "description": "A fully automated agent system that lives on a server.",
   "main": "./src/index.js",
   "type": "module",

package/src/channels/telegram/index.js

@@ -154,9 +154,15 @@ export async function startTelegramChannel(config) {
         }
       }
 
+      let lastCheckpointSent = null;
       let result;
       try {
-        result = await handleChat(config, sessionId, userText, allAttachments);
+        result = await handleChat(config, sessionId, userText, allAttachments, async (checkpointResponse) => {
+          const text = typeof checkpointResponse === 'string' ? checkpointResponse : JSON.stringify(checkpointResponse);
+          lastCheckpointSent = text;
+          await appendTelegramChatLog(chatId, sessions[chatId] || null, 'JARVIS', text);
+          await sendMessage(api, chatId, text, sessions[chatId] || null);
+        });
       } catch (e) {
         console.error(`[telegram] agent error chat_id=${chatId}: ${e.message}`);
         const errText = e.message
@@ -185,9 +191,16 @@ export async function startTelegramChannel(config) {
           : result.response != null ? JSON.stringify(result.response, null, 2) : '';
         const text = rawResponse.trim()
           || 'The agent encountered an error and could not produce a response. Please try again.';
-        await appendTelegramChatLog(chatId, result.sessionId, 'JARVIS', text);
-        await sendMessage(api, chatId, text, result.sessionId);
-        console.log(`[telegram] response sent chat_id=${chatId} length=${text.length}`);
+        // Skip sending if this response was already sent as a checkpoint update —
+        // intervention_required and zero-progress reuse the last checkpoint response
+        // as their finalResponse, which would otherwise cause a duplicate message.
+        if (text !== lastCheckpointSent) {
+          await appendTelegramChatLog(chatId, result.sessionId, 'JARVIS', text);
+          await sendMessage(api, chatId, text, result.sessionId);
+          console.log(`[telegram] response sent chat_id=${chatId} length=${text.length}`);
+        } else {
+          console.log(`[telegram] skipped duplicate final response chat_id=${chatId}`);
+        }
       } catch (e) {
         console.error(`[telegram] delivery error chat_id=${chatId}: ${e.message}`);
         await api.sendMessage(chatId, 'Sorry, something went wrong sending the response. Please try again.').catch(() => {});

package/src/scripts/onboarding.js

@@ -390,16 +390,42 @@ async function run() {
   console.log(chalk.green(`\nModel ${chalk.bold(selectedModel)} saved to settings.`));
 
   // --- VISION MODEL STEP (OPTIONAL) ---
-  const { configureVision } = await inquirer.prompt([
-    {
-      type: 'confirm',
-      name: 'configureVision',
-      message: 'Do you want to configure a separate vision model for image analysis (e.g. for Telegram photos)?',
-      default: !!settings.visionModel,
+  let skipVision = false;
+  if (settings.visionModel) {
+    const { visionAction } = await inquirer.prompt([
+      {
+        type: 'list',
+        name: 'visionAction',
+        message: `Vision model is configured (${chalk.yellow(settings.visionModel)}). What do you want to do?`,
+        choices: [
+          { name: 'Keep current vision model', value: 'keep' },
+          { name: 'Change vision model', value: 'change' },
+          { name: 'Disable vision', value: 'disable' },
+        ],
+      }
+    ]);
+    if (visionAction === 'keep') {
+      skipVision = true;
+    } else if (visionAction === 'disable') {
+      delete settings.visionProvider;
+      delete settings.visionModel;
+      saveSettings(settings);
+      console.log(chalk.yellow('Vision model disabled.'));
+      skipVision = true;
     }
-  ]);
+  } else {
+    const { configureVision } = await inquirer.prompt([
+      {
+        type: 'confirm',
+        name: 'configureVision',
+        message: 'Do you want to configure a separate vision model for image analysis (e.g. for Telegram photos)?',
+        default: false,
+      }
+    ]);
+    if (!configureVision) skipVision = true;
+  }
 
-  if (configureVision) {
+  if (!skipVision) {
     const { visionProvider } = await inquirer.prompt([
       {
         type: 'list',
@@ -558,11 +584,6 @@ async function run() {
     settings.visionModel = visionModel;
     saveSettings(settings);
     console.log(chalk.green(`Vision model ${chalk.bold(visionModel)} saved.`));
-  } else {
-    // Clear vision config if user opts out
-    delete settings.visionProvider;
-    delete settings.visionModel;
-    saveSettings(settings);
   }
 
   // --- TELEGRAM CHANNEL STEP (OPTIONAL) ---
@@ -640,44 +661,6 @@ async function run() {
     }
   }
 
-  // --- PERPLEXITY STEP (OPTIONAL) ---
-  const existingPerplexityKey = loadEnvVar('PERPLEXITY_API_KEY');
-  const { configurePerplexity } = await inquirer.prompt([
-    {
-      type: 'confirm',
-      name: 'configurePerplexity',
-      message: 'Do you want to configure Perplexity web search?',
-      default: !!existingPerplexityKey
-    }
-  ]);
-
-  if (configurePerplexity) {
-    let keepPerplexityKey = false;
-    if (existingPerplexityKey) {
-      const { keep } = await inquirer.prompt([
-        {
-          type: 'confirm',
-          name: 'keep',
-          message: 'A PERPLEXITY_API_KEY is already configured. Do you want to keep it?',
-          default: true
-        }
-      ]);
-      keepPerplexityKey = keep;
-    }
-    if (!keepPerplexityKey) {
-      const { perplexityKey } = await inquirer.prompt([
-        {
-          type: 'password',
-          name: 'perplexityKey',
-          message: 'Enter your Perplexity API key (from perplexity.ai/settings/api):',
-          validate: (input) => input.trim().length > 0 || 'API key cannot be empty.'
-        }
-      ]);
-      saveEnvVar('PERPLEXITY_API_KEY', perplexityKey.trim());
-      console.log(chalk.green('Perplexity API key saved.'));
-    }
-  }
-
   // --- PM2 + LOG ROTATION STEP ---
   const pm2Check = spawnSync('pm2', ['--version'], { stdio: 'pipe' });
   if (pm2Check.status !== 0) {

package/src/server/agent.js

@@ -618,7 +618,7 @@ export async function withSessionLock(sessionId, fn) {
  * Main entry point: handles a single POST /api/chat request.
  * Manages the handoff loop across multiple agent runs.
  */
-export async function handleChat(config, requestSessionId, userMessage, attachments = []) {
+export async function handleChat(config, requestSessionId, userMessage, attachments = [], onCheckpoint = null) {
   const sessionId = requestSessionId || crypto.randomUUID();
 
   // Serialize concurrent requests for the same session. Each request registers
@@ -632,7 +632,7 @@ export async function handleChat(config, requestSessionId, userMessage, attachme
   await previous;
 
   try {
-    return await _runHandleChat(config, sessionId, userMessage, attachments);
+    return await _runHandleChat(config, sessionId, userMessage, attachments, onCheckpoint);
   } finally {
     releaseLock();
     // Clean up only if no one else has queued behind us
@@ -646,7 +646,7 @@ export async function handleChat(config, requestSessionId, userMessage, attachme
  * The actual chat logic, extracted so handleChat can wrap it cleanly with the
  * session lock.
  */
-async function _runHandleChat(config, sessionId, userMessage, attachments = []) {
+async function _runHandleChat(config, sessionId, userMessage, attachments = [], onCheckpoint = null) {
   const client = createClient(config);
 
   const systemPromptTemplate = loadSystemPrompt();
@@ -802,7 +802,9 @@ async function _runHandleChat(config, sessionId, userMessage, attachments = [])
         break;
       }
 
-      // Checkpoint reached — log this run
+      // Checkpoint reached — log this run and notify the caller (e.g. Telegram adapter)
+      // so intermediate progress is visible to the user instead of being swallowed
+      // by the handoff loop until the final response.
       await appendLog(sessionId, {
         iteration: run.iteration,
         model: config.selectedModel,
@@ -812,6 +814,7 @@ async function _runHandleChat(config, sessionId, userMessage, attachments = [])
         logSummary: run.logSummary,
         status: 'checkpoint_reached',
       });
+      if (onCheckpoint) await onCheckpoint(run.response);
 
       // Accumulate failedApproaches from this run into session metadata so the
       // full history of failed strategies is available across all handoff runs.

package/src/server/tools.js

@@ -656,6 +656,71 @@ const SEED_TOOLS = {
       return { status: 'ok', name: args.name, content };
     `,
   },
+  analyze_image: {
+    definition: {
+      type: 'function',
+      function: {
+        name: 'analyze_image',
+        description: 'Fetch an image from a URL and analyze it using the configured vision model. Returns a detailed description of the image. Use this whenever a user shares an image URL and asks about its content.',
+        parameters: {
+          type: 'object',
+          properties: {
+            url: {
+              type: 'string',
+              description: 'The URL of the image to analyze (http or https).',
+            },
+            prompt: {
+              type: 'string',
+              description: 'Optional question or instruction for the vision model, e.g. "What text is visible?" or "Describe the chart". Defaults to a general description.',
+            },
+          },
+          required: ['url'],
+        },
+      },
+    },
+    code: `
+      const settingsPath = path.join(process.env.HOME, '.jarvis/data/config/settings.json');
+      const settings = JSON.parse(await fs.promises.readFile(settingsPath, 'utf8').catch(() => '{}'));
+      const visionModel = settings.visionModel;
+      const visionProvider = settings.visionProvider;
+      if (!visionModel || !visionProvider) {
+        return { status: 'error', message: 'No vision model configured. Set visionModel and visionProvider in settings.' };
+      }
+      let apiKey, baseURL;
+      if (visionProvider === 'z-ai') {
+        apiKey = process.env.ZAI_API_KEY;
+        baseURL = 'https://api.z.ai/api/coding/paas/v4/';
+      } else {
+        apiKey = process.env.OPENROUTER_API_KEY;
+        baseURL = 'https://openrouter.ai/api/v1';
+      }
+      if (!apiKey) return { status: 'error', message: 'No API key found for vision provider: ' + visionProvider };
+      const imgResponse = await fetch(args.url);
+      if (!imgResponse.ok) return { status: 'error', message: 'Failed to fetch image: HTTP ' + imgResponse.status };
+      const buffer = await imgResponse.arrayBuffer();
+      const base64 = Buffer.from(buffer).toString('base64');
+      const contentType = imgResponse.headers.get('content-type') || 'image/jpeg';
+      const dataUrl = 'data:' + contentType + ';base64,' + base64;
+      const textPrompt = args.prompt?.trim()
+        ? 'The user shared this image with the following question/context: "' + args.prompt.trim() + '"\\n\\nPlease describe what you see, paying special attention to anything relevant to their message.'
+        : 'Please describe this image in detail. Include all visible text, objects, colors, layout, and any other relevant details.';
+      const apiResponse = await fetch(baseURL + (baseURL.endsWith('/') ? '' : '/') + 'chat/completions', {
+        method: 'POST',
+        headers: { 'Authorization': 'Bearer ' + apiKey, 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+          model: visionModel,
+          messages: [{ role: 'user', content: [
+            { type: 'image_url', image_url: { url: dataUrl } },
+            { type: 'text', text: textPrompt },
+          ]}],
+        }),
+      });
+      const result = await apiResponse.json();
+      if (!apiResponse.ok) return { status: 'error', message: result.error?.message || 'Vision API error' };
+      const description = result.choices?.[0]?.message?.content?.trim() || '(no description returned)';
+      return { status: 'ok', description };
+    `,
+  },
 };
 
 export function seedTools() {