@yemi33/minions 0.1.263 → 0.1.265

package/CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## 0.1.265 (2026-04-03)
+
+### Features
+-  CC streaming shows tool use activity during multi-turn responses
+
+### Fixes
+-  clean up remaining evaluate references in comments and log messages
+
 ## 0.1.263 (2026-04-03)
 
 ### Features

package/dashboard/js/command-center.js

@@ -217,6 +217,9 @@ async function _ccDoSend(message, skipUserMsg) {
             streamedText = evt.text;
             streamDiv.innerHTML = renderMd(streamedText);
             if (msgs.scrollHeight - msgs.scrollTop - msgs.clientHeight < 150) msgs.scrollTop = msgs.scrollHeight;
+          } else if (evt.type === 'tool') {
+            streamDiv.innerHTML = '<span style="color:var(--blue);font-size:11px">\uD83D\uDD27 Using ' + escHtml(evt.name) + '...</span>';
+            if (msgs.scrollHeight - msgs.scrollTop - msgs.clientHeight < 150) msgs.scrollTop = msgs.scrollHeight;
           } else if (evt.type === 'done') {
             // Replace streaming div with a proper ccAddMessage
             streamDiv.remove();

package/dashboard.js

@@ -3013,6 +3013,9 @@ What would you like to discuss or change? When you're happy, say "approve" and I
           sessionId,
           onChunk: (text) => {
             try { res.write('data: ' + JSON.stringify({ type: 'chunk', text }) + '\n\n'); } catch {}
+          },
+          onToolUse: (name, input) => {
+            try { res.write('data: ' + JSON.stringify({ type: 'tool', name, input: typeof input === 'string' ? input.slice(0, 200) : JSON.stringify(input).slice(0, 200) }) + '\n\n'); } catch {}
           }
         });
         trackUsage('command-center', result.usage);

package/engine/lifecycle.js

@@ -1075,7 +1075,7 @@ function resolveWiPath(meta) {
 }
 
 /**
- * Parse structured eval verdict from evaluate agent output.
+ * Parse structured eval verdict from review agent output.
  * Looks for a JSON block with { pass, build, tests, criteria_met, criteria_failed, feedback }.
  * Returns parsed object or null if not found.
  */
@@ -1256,7 +1256,7 @@ function runPostCompletionHooks(dispatchItem, agentId, code, stdout, config) {
 
   if (isSuccess && meta?.item?.id && !skipDoneStatus) updateWorkItemStatus(meta, 'done', '');
 
-  // Auto-dispatch evaluate work item after implement or fix completes successfully
+  // Auto-dispatch review work item after implement or fix completes successfully
   if (isSuccess && !skipDoneStatus && (type === 'implement' || (type === 'fix' && meta?.item?._evalParentId)) && meta?.item?.id) {
     const evalLoop = config.engine?.evalLoop ?? shared.ENGINE_DEFAULTS.evalLoop;
     if (evalLoop) {
@@ -1269,7 +1269,7 @@ function runPostCompletionHooks(dispatchItem, agentId, code, stdout, config) {
             // Dedup: skip if a review item already exists for this parent
             const existing = items.find(i => i._evalParentId === evalTargetId && i.type === 'review' && i.status === 'pending');
             if (existing) {
-              log('info', `Eval loop: evaluate item ${existing.id} already exists for ${evalTargetId}, skipping`);
+              log('info', `Eval loop: review item ${existing.id} already exists for ${evalTargetId}, skipping`);
               return items;
             }
             const parentItem = items.find(i => i.id === evalTargetId);

package/engine/llm.js

@@ -113,7 +113,7 @@ function isResumeSessionStillValid(result) {
  * Returns the same result object as callLLM when the process completes.
  * onChunk(text) is called for each assistant text block as it arrives.
  */
-function callLLMStreaming(promptText, sysPromptText, { timeout = 120000, label = 'llm', model = 'sonnet', maxTurns = 1, allowedTools = '', sessionId = null, onChunk = () => {} } = {}) {
+function callLLMStreaming(promptText, sysPromptText, { timeout = 120000, label = 'llm', model = 'sonnet', maxTurns = 1, allowedTools = '', sessionId = null, onChunk = () => {}, onToolUse = null } = {}) {
   return new Promise((resolve) => {
     const id = uid();
     const tmpDir = path.join(ENGINE_DIR, 'tmp');
@@ -162,18 +162,21 @@ function callLLMStreaming(promptText, sysPromptText, { timeout = 120000, label =
           }
         } catch { /* incomplete JSON or non-JSON line */ }
       }
-      // Also try to extract partial text from incomplete line buffer (real-time streaming)
-      // Claude CLI writes large JSON objects incrementally — extract "text":"..." as it grows
-      try {
-        const textMatch = lineBuf.match(/"text"\s*:\s*"((?:[^"\\]|\\.)*)$/);
-        if (textMatch && textMatch[1].length > 20 && textMatch[1] !== lastTextSent) {
-          // Unescape basic JSON escapes for display
-          const partial = textMatch[1].replace(/\\n/g, '\n').replace(/\\"/g, '"').replace(/\\\\/g, '\\');
-          if (partial.length > lastTextSent.length) {
-            onChunk(partial);
+      // Also emit tool_use events so the frontend can show "Using tool: Read..."
+      for (const line of lines) {
+        const trimmed = line.trim();
+        if (!trimmed || !trimmed.startsWith('{')) continue;
+        try {
+          const obj = JSON.parse(trimmed);
+          if (obj.type === 'assistant' && obj.message?.content) {
+            for (const block of obj.message.content) {
+              if (block.type === 'tool_use' && block.name && onToolUse) {
+                onToolUse(block.name, block.input);
+              }
+            }
           }
-        }
-      } catch { /* best-effort partial extraction */ }
+        } catch {}
+      }
     });
     proc.stderr.on('data', d => { stderr += d.toString(); });
 

package/engine/shared.js

@@ -356,8 +356,8 @@ const ENGINE_DEFAULTS = {
   autoApprovePlans: false, // auto-approve PRDs without waiting for human approval
   autoReview: true, // auto-dispatch review agents for new PRs (disable for manual review workflow)
   meetingRoundTimeout: 600000, // 10min per meeting round before auto-advance
-  evalLoop: true, // enable evaluate→fix loop after implementation completes
-  evalMaxIterations: 3, // max evaluate→fix cycles before escalating to human
+  evalLoop: true, // enable review→fix loop after implementation completes
+  evalMaxIterations: 3, // max review→fix cycles before escalating to human
   evalMaxCost: null, // USD ceiling per work item across all eval iterations; null = no limit (gather baseline data first)
 };
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.263",
+  "version": "0.1.265",
   "description": "Multi-agent AI dev team that runs from ~/.minions/ — five autonomous agents share a single engine, dashboard, and knowledge base",
   "bin": {
     "minions": "bin/minions.js"