@yemi33/minions 0.1.1659 → 0.1.1660

package/CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## 0.1.1660 (2026-05-01)
+
+### Fixes
+- less rigid agent orphan detection
+
 ## 0.1.1659 (2026-05-01)
 
 ### Other

package/engine/copilot-models.json

@@ -1,5 +1,5 @@
 {
   "runtime": "copilot",
   "models": null,
-  "cachedAt": "2026-05-01T05:34:43.680Z"
+  "cachedAt": "2026-05-01T05:58:00.505Z"
 }

package/engine/shared.js

@@ -721,7 +721,19 @@ const ENGINE_DEFAULTS = {
   inboxConsolidateThreshold: 5,
   agentTimeout: 18000000,  // 5h
   heartbeatTimeout: 300000, // 5min — stale-orphan grace after process tracking is lost
-  heartbeatTimeouts: {}, // optional per-type stale-orphan overrides; merged at runtime (see timeout.js)
+  // Per-type stale-orphan overrides (merged with config.engine.heartbeatTimeouts at runtime — see timeout.js).
+  // Heavy work types (multi-file edits, builds, test suites, full verify cycles) routinely go quiet for
+  // longer than the 5-min default when the engine has lost their tracked handle (e.g. across an engine
+  // restart). We give them headroom up to a typical build+tests cycle. Short-running types
+  // (decompose / meeting / etc.) keep the 5-min default by simply not appearing here.
+  heartbeatTimeouts: {
+    implement:       900000, // 15min — refactors, multi-file edits, builds
+    'implement:large': 900000, // 15min — same class of work, larger scope
+    fix:             900000, // 15min — fix runs often include builds + retries
+    test:            900000, // 15min — build-and-test against existing PR
+    verify:          900000, // 15min — full project verification cycle
+    plan:            600000, // 10min — research-heavy
+  },
   maxTurns: 100,
   worktreeCreateTimeout: 300000, // 5min for git worktree add on large Windows repos
   worktreeCreateRetries: 1, // retry once on transient timeout/lock races
@@ -785,7 +797,6 @@ const ENGINE_DEFAULTS = {
   copilotReasoningSummaries: false,  // Copilot --enable-reasoning-summaries (Anthropic-family models only)
   maxBudgetUsd: undefined,       // fleet USD ceiling for --max-budget-usd (per-agent override: agents.<id>.maxBudgetUsd). Honors 0 via ?? so a literal cap of $0 works
   disableModelDiscovery: false,  // skip runtime.listModels() REST calls fleet-wide (settings UI falls back to free-text)
-  heartbeatTimeouts: {},
   maxPendingContexts: 20, // cap pendingContexts arrays in cooldowns.json to prevent unbounded growth
   maxPendingContextEntryBytes: 256 * 1024, // 256 KB — cap each pendingContexts entry to prevent huge PR comments from bloating cooldowns.json
   maxDispatchPromptBytes: 1024 * 1024, // 1 MB — dispatch items with prompts larger than this sidecar to engine/contexts/ to prevent dispatch.json OOM (#1167)

package/engine/timeout.js

@@ -9,7 +9,7 @@ const queries = require('./queries');
 const steering = require('./steering');
 
 const { safeRead, safeWrite, safeJson, mutateJsonFileLocked, getProjects, projectWorkItemsPath, log, ts,
-  ENGINE_DEFAULTS, WI_STATUS, WORK_TYPE, DISPATCH_RESULT, AGENT_STATUS } = shared;
+  ENGINE_DEFAULTS, ENGINE_DIR, WI_STATUS, WORK_TYPE, DISPATCH_RESULT, AGENT_STATUS } = shared;
 const { getDispatch, getAgentStatus } = queries;
 const AGENTS_DIR = queries.AGENTS_DIR;
 const MINIONS_DIR = shared.MINIONS_DIR;
@@ -142,6 +142,23 @@ function isTrackedProcessAlive(procInfo) {
   }
 }
 
+// Last-resort liveness check via the on-disk PID file (engine/tmp/pid-<safeId>.pid).
+// Used by orphan detection to avoid false-positive kills when the engine has lost the
+// tracked process handle (engine restart, never-tracked spawn, etc.) but the OS-level
+// child process is still alive and healthy. The safeId here mirrors engine.js spawn
+// (id.replace(/[:\\/*?"<>|]/g, '-')) — same pattern engine/cli.js uses to re-attach.
+function isOsPidAliveForDispatch(itemId) {
+  const safeId = String(itemId || '').replace(/[:\\/*?"<>|]/g, '-');
+  const pidPath = path.join(ENGINE_DIR, 'tmp', `pid-${safeId}.pid`);
+  let raw;
+  try { raw = fs.readFileSync(pidPath, 'utf8'); }
+  catch { return false; }
+  const pid = parseInt(String(raw).trim(), 10);
+  if (!Number.isFinite(pid) || pid <= 0) return false;
+  try { process.kill(pid, 0); return true; }
+  catch { return false; }
+}
+
 function checkTimeouts(config) {
   const activeProcesses = engine().activeProcesses;
   const engineRestartGraceUntil = engine().engineRestartGraceUntil;
@@ -335,6 +352,11 @@ function checkTimeouts(config) {
     } catch { /* ENOENT — keep default */ }
 
     if (!processAlive && silentMs > staleOrphanTimeout && (Date.now() > engineRestartGraceUntil || engineRestartGraceExempt?.has(item.id))) {
+      // Last-resort PID check: lost tracked handle but OS process may still be alive.
+      if (isOsPidAliveForDispatch(item.id)) {
+        log('info', `Orphan check: ${item.agent} (${item.id}) silent ${silentSec}s but OS PID is alive — keeping [${_logState}]`);
+        continue;
+      }
       // No tracked process AND no recent output past stale-orphan timeout AND (grace period expired OR confirmed-dead at restart) → orphaned
       log('warn', `Orphan detected: ${item.agent} (${item.id}) — no live process tracked, silent for ${silentSec}s [${_logState}]`);
       dispatch().updateAgentStatus(item.id, AGENT_STATUS.TIMED_OUT, `Orphaned — no process, silent for ${silentSec}s`);
@@ -424,4 +446,5 @@ module.exports = {
   checkTimeouts,
   checkSteering,
   checkIdleThreshold,
+  isOsPidAliveForDispatch,
 };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.1659",
+  "version": "0.1.1660",
   "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"