@yemi33/minions 0.1.872 → 0.1.873

package/CHANGELOG.md

@@ -1,6 +1,6 @@
 # Changelog
 
-## 0.1.872 (2026-04-11)
+## 0.1.873 (2026-04-11)
 
 ### Features
 -  show last task duration in thought-process tab when agent is idle
@@ -10,6 +10,7 @@
 -  optimistic running state on pipeline Run Now click
 
 ### Fixes
+- short-circuit restart grace period for verifiably-dead agent processes (closes #869) (#886)
 -  orphan recovery misclassifies mid-run agents as failed on engine restart
 - dep merge failures identify conflicting branch and auto-queue fix (closes #814) (#882)
 -  always show doc-chat expand bar when thread is collapsed

package/engine/cli.js

@@ -161,6 +161,7 @@ const commands = {
           }
         }
 
+        const hadPid = agentPid && agentPid > 0; // track before liveness check
         if (agentPid && agentPid > 0) {
           try {
             if (process.platform === 'win32') {
@@ -172,6 +173,11 @@ const commands = {
           } catch { agentPid = null; }
         }
 
+        // PID was found but confirmed dead — exempt from restart grace period (#869)
+        if (hadPid && !agentPid) {
+          e.engineRestartGraceExempt.add(item.id);
+        }
+
         if (agentPid) {
           // Load sessionId from session.json for steering support
           let sessionId = null;

package/engine/timeout.js

@@ -127,6 +127,7 @@ function checkSteering(config) {
 function checkTimeouts(config) {
   const activeProcesses = engine().activeProcesses;
   const engineRestartGraceUntil = engine().engineRestartGraceUntil;
+  const engineRestartGraceExempt = engine().engineRestartGraceExempt;
   const { completeDispatch } = dispatch();
   const { runPostCompletionHooks } = require('./lifecycle');
 
@@ -307,8 +308,8 @@ function checkTimeouts(config) {
     const procInfo = activeProcesses.get(item.id);
     if (procInfo?._steeringAt && Date.now() - procInfo._steeringAt < 60000) continue;
 
-    if (!hasProcess && silentMs > effectiveTimeout && Date.now() > engineRestartGraceUntil) {
-      // No tracked process AND no recent output past effective timeout AND grace period expired → orphaned
+    if (!hasProcess && silentMs > effectiveTimeout && (Date.now() > engineRestartGraceUntil || engineRestartGraceExempt?.has(item.id))) {
+      // No tracked process AND no recent output past effective timeout AND (grace period expired OR confirmed-dead at restart) → orphaned
       log('warn', `Orphan detected: ${item.agent} (${item.id}) — no process tracked, silent for ${silentSec}s${isBlocking ? ' (blocking timeout exceeded)' : ''}`);
       dispatch().updateAgentStatus(item.id, AGENT_STATUS.TIMED_OUT, `Orphaned — no process, silent for ${silentSec}s`);
       // Clear session so retry starts fresh

package/engine.js

@@ -146,6 +146,7 @@ const activeProcesses = new Map(); // dispatchId → { proc, agentId, startedAt
 const realActivityMap = new Map(); // dispatchId → timestamp of last REAL agent output (not engine heartbeat)
 // tempAgents imported from engine/routing.js
 let engineRestartGraceUntil = 0; // timestamp — suppress orphan detection until this time
+const engineRestartGraceExempt = new Set(); // dispatch IDs with confirmed-dead PIDs at restart — bypass grace period
 
 // Per-tick cache of refs that failed to fetch — avoids repeating 30s ETIMEDOUT for same missing ref
 // Cleared at the start of each tick cycle (see tickInner)
@@ -3155,7 +3156,8 @@ module.exports = {
 
   // Dispatch management (re-exported from engine/dispatch.js)
   mutateDispatch, addToDispatch, isRetryableFailureReason, completeDispatch, writeInboxAlert, updateAgentStatus,
-  activeProcesses, realActivityMap, get engineRestartGraceUntil() { return engineRestartGraceUntil; },
+  activeProcesses, realActivityMap, engineRestartGraceExempt,
+  get engineRestartGraceUntil() { return engineRestartGraceUntil; },
   set engineRestartGraceUntil(v) { engineRestartGraceUntil = v; },
 
   // Agent lifecycle

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.872",
+  "version": "0.1.873",
   "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"