@yemi33/minions 0.1.769 → 0.1.771

package/CHANGELOG.md

@@ -1,11 +1,17 @@
 # Changelog
 
-## 0.1.769 (2026-04-10)
+## 0.1.771 (2026-04-10)
+
+### Fixes
+-  clarify timeout.js error reason — "Exited with error" vs "Completed"
+
+## 0.1.770 (2026-04-10)
 
 ### Features
 - id/kill endpoint
 
 ### Fixes
+-  increase ADO pipeline/task caps from 3 to 10, update docs
 -  ADO build detection uses builds API, not unreliable status checks
 -  handle work type names as action aliases for dispatch
 -  align Agent Usage and Engine Usage table columns

package/docs/pr-review-fix-loop.md

@@ -41,7 +41,7 @@ How the engine manages the lifecycle of a PR from creation through review, fix,
 
 - Gate: `buildFixAttempts < maxBuildFixAttempts` (default 3) + grace period expired
 - **Grace period** (`_buildFixPushedAt`): after fix dispatches, waits `buildFixGracePeriod` (default 10min, configurable in `ENGINE_DEFAULTS`) for CI to run before re-dispatching. Cleared when poller detects build status transition (CI actually ran).
-- **Error logs**: GitHub fetches annotations (failures only, not warnings) + Actions job log (always). ADO fetches build timeline + failed task logs. Both fetch up to 3 failing pipelines.
+- **Error logs**: GitHub fetches annotations (failures only, not warnings) + Actions job log (always). ADO queries builds API directly (not status checks), fetches build timeline → failed task logs (up to 10 per build, up to 10 failing pipelines).
 - **Escalation**: after 3 failed attempts, writes inbox alert, sets `buildFixEscalated = true`, stops auto-dispatch. Counter resets when build recovers.
 
 ## 5. Fix completes
@@ -108,3 +108,20 @@ How the engine manages the lifecycle of a PR from creation through review, fix,
 | `lastReviewedAt` | `updatePrAfterReview()` | Prevents re-dispatch if reviewed |
 | `minionsReview` | Post-completion hooks | `{ reviewer, reviewedAt, note, fixedAt }` |
 | `humanFeedback` | `pollPrHumanComments()` | `{ pendingFix, feedbackContent, lastProcessedCommentDate }` |
+
+## Platform differences
+
+| | GitHub | ADO |
+|---|---|---|
+| **Build status API** | `/commits/{sha}/check-runs` | `_apis/build/builds` (not status checks — those show stale codecoverage postbacks) |
+| **Commit tracking** | `head.sha` (source branch tip) | `lastMergeCommit.commitId` (merge commit that builds use as sourceVersion) |
+| **Passing** | success, skipped, neutral | succeeded, partiallySucceeded (warnings, not failures) |
+| **Error logs** | Annotations (failures only) + Actions job log (always) | Build timeline → failed task logs (up to 10 tasks, up to 10 pipelines) |
+| **Push detection** | `head.sha` change | `lastMergeCommit.commitId` change |
+
+### ADO lessons learned
+
+- Don't trust `/pullRequests/{id}/statuses` — shows stale codecoverage postbacks, not actual build results
+- Builds use the merge commit hash as `sourceVersion`, not the source branch commit — compare against `lastMergeCommit.commitId`
+- `partiallySucceeded` counts as passing (warnings, not failures)
+- A stale but passing build is still valid — don't re-trigger builds that already passed

package/engine/ado.js

@@ -141,9 +141,9 @@ async function fetchAdoBuildErrorLog(orgBase, project, failedStatus, token, pr,
     );
     if (failedRecords.length === 0) return null;
 
-    // Fetch logs for failed tasks (cap at 3 to limit API calls)
+    // Fetch logs for failed tasks (cap at 10 to cover multi-stage pipelines)
     const logParts = [];
-    for (const record of failedRecords.slice(0, 3)) {
+    for (const record of failedRecords.slice(0, 10)) {
       try {
         const logUrl = `${orgBase}/${project.adoProject}/_apis/build/builds/${buildId}/logs/${record.log.id}?api-version=7.1`;
         const text = await adoFetchText(logUrl, token);
@@ -404,7 +404,7 @@ async function pollPrStatus(config) {
 
       // Fetch actual compiler/build error logs when transitioning to failing
       if (buildStatus === 'failing') {
-        const failedStatusObjs = buildStatuses.filter(s => s.state === 'failed' || s.state === 'error').slice(0, 3);
+        const failedStatusObjs = buildStatuses.filter(s => s.state === 'failed' || s.state === 'error').slice(0, 10);
         const logParts = [];
         const seenBuildIds = new Set();
         for (const failedStatusObj of failedStatusObjs) {

package/engine/timeout.js

@@ -206,7 +206,7 @@ function checkTimeouts(config) {
           safeWrite(outputLogPath, `# Output for dispatch ${item.id}\n# Exit code: ${isSuccess ? 0 : 1}\n# Completed: ${ts()}\n# Detected via output scan\n\n## Result\n${text || '(no text)'}\n`);
         } catch (e) { log('warn', 'parse output result: ' + e.message); }
 
-        completeDispatch(item.id, isSuccess ? DISPATCH_RESULT.SUCCESS : DISPATCH_RESULT.ERROR, 'Completed (detected from output)');
+        completeDispatch(item.id, isSuccess ? DISPATCH_RESULT.SUCCESS : DISPATCH_RESULT.ERROR, isSuccess ? 'Completed (detected from output)' : 'Exited with error (detected from output)');
 
         // Run post-completion hooks via shared helper (async — fire and forget in timeout context)
         const fullLogForHooks = safeRead(liveLogPath) || liveLog;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.769",
+  "version": "0.1.771",
   "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"