@yemi33/minions 0.1.369 → 0.1.371

package/CHANGELOG.md

@@ -1,6 +1,11 @@
 # Changelog
 
-## 0.1.369 (2026-04-06)
+## 0.1.371 (2026-04-06)
+
+### Features
+- Fix review re-dispatch infinite loop (#226)
+- Add subagent guidance and health check preamble to playbooks (#225)
+- Fix 2 failing source-pattern test assertions (#224)
 
 ### Fixes
 -  copy button preserves markdown formatting

package/engine/github.js

@@ -141,6 +141,13 @@ async function pollPrStatus(config) {
     else if (prData.state === 'closed') newStatus = PR_STATUS.ABANDONED;
     else if (prData.state === 'open') newStatus = PR_STATUS.ACTIVE;
 
+    // Track head SHA changes to detect new pushes (used for review re-dispatch gating)
+    if (prData.head?.sha && pr.headSha !== prData.head.sha) {
+      pr.headSha = prData.head.sha;
+      pr.lastPushedAt = new Date().toISOString();
+      updated = true;
+    }
+
     if (pr.status !== newStatus) {
       log('info', `PR ${pr.id} status: ${pr.status} → ${newStatus}`);
       pr.status = newStatus;
@@ -191,6 +198,9 @@ async function pollPrStatus(config) {
       if (states.some(s => s === 'CHANGES_REQUESTED')) newReviewStatus = 'changes-requested';
       else if (states.some(s => s === 'APPROVED')) newReviewStatus = 'approved';
       else if (states.length > 0) newReviewStatus = 'pending';
+      // If all reviews were COMMENTED (filtered out), states is empty but reviews exist.
+      // Set to 'waiting' instead of leaving as 'pending' to prevent infinite review re-dispatch.
+      else if (states.length === 0 && reviews.length > 0 && newReviewStatus === 'pending') newReviewStatus = 'waiting';
 
       if (pr.reviewStatus !== newReviewStatus) {
         log('info', `PR ${pr.id} reviewStatus: ${pr.reviewStatus} → ${newReviewStatus}`);

package/engine/lifecycle.js

@@ -727,6 +727,9 @@ function updatePrAfterReview(agentId, pr, project) {
 
   // Set reviewStatus to 'waiting' (single source of truth — synced from ADO/GitHub votes on next poll)
   target.reviewStatus = 'waiting';
+  // Track when this PR was last reviewed — used by engine.js to prevent re-dispatch
+  // until new commits are pushed (lastPushedAt > lastReviewedAt)
+  target.lastReviewedAt = ts();
   target.minionsReview = {
     reviewer: reviewerName,
     reviewedAt: ts(),

package/engine.js

@@ -1268,9 +1268,11 @@ function discoverFromPrs(config, project) {
     // minionsReview tracks metadata (reviewer, note) but not the authoritative status
     const reviewStatus = pr.reviewStatus || 'pending';
 
-    // PRs needing review: pending or waiting (review dispatched but no verdict yet)
+    // PRs needing review: pending review status and not already reviewed without new commits
     const autoReview = config.engine?.autoReview !== false;
-    const needsReview = autoReview && reviewStatus === 'pending';
+    // Skip re-dispatch if already reviewed and no new commits pushed since last review
+    const alreadyReviewed = pr.lastReviewedAt && (!pr.lastPushedAt || pr.lastPushedAt <= pr.lastReviewedAt);
+    const needsReview = autoReview && reviewStatus === 'pending' && !alreadyReviewed;
     if (needsReview) {
       const key = `review-${project?.name || 'default'}-${pr.id}`;
       if (isAlreadyDispatched(key) || isOnCooldown(key, cooldownMs)) continue;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@yemi33/minions",
-  "version": "0.1.369",
+  "version": "0.1.371",
   "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"

package/playbooks/decompose.md

@@ -21,6 +21,10 @@ A work item has been flagged as too large for a single agent dispatch. Analyze t
 {{acceptance_criteria}}
 {{/acceptance_criteria}}
 
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks. For codebase exploration, reading files, and writing the decomposition output, work directly — do not spawn subagents.
+
 ## Instructions
 
 1. **Explore the codebase** at `{{project_path}}` — understand the existing structure, patterns, and dependencies

package/playbooks/explore.md

@@ -52,6 +52,10 @@ If the task is purely exploratory (no deliverable requested), skip this step.
 ### 6. Status
 **Note:** Do NOT write to `agents/*/status.json` — the engine manages your status automatically.
 
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks. For reading files, exploring directories, and writing findings, work directly — do not spawn subagents.
+
 ## Rules
 - Do NOT modify existing code unless the task explicitly asks for it.
 - Use the appropriate MCP tools for PR creation — check available tools before starting.

package/playbooks/fix.md

@@ -17,6 +17,14 @@ Branch: `{{pr_branch}}`
 
 {{review_note}}
 
+## Health Check
+
+Before starting work, run `git status` and verify the worktree is clean and on the expected branch (`{{pr_branch}}`). If the worktree is dirty or on the wrong branch, report the issue and stop.
+
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks. For sequential work, single-file edits, searches, and file reads, work directly — do not spawn subagents.
+
 ## How to Fix
 
 1. You are already in the correct worktree on branch `{{pr_branch}}`. Do NOT create additional worktrees.

package/playbooks/implement-shared.md

@@ -45,6 +45,14 @@ git push origin {{branch_name}}
 - Remove the worktree — the next plan item needs it
 - Create a new worktree — one already exists at `{{worktree_path}}`
 
+## Health Check
+
+Before starting work, run `git status` and verify the worktree is clean and on the expected branch (`{{branch_name}}`). If the worktree is dirty or on the wrong branch, report the issue and stop.
+
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks. For sequential work, single-file edits, searches, and file reads, work directly — do not spawn subagents.
+
 ## Instructions
 
 1. Read relevant source code and reference implementations before writing anything

package/playbooks/implement.md

@@ -30,6 +30,14 @@ If this feature spans multiple projects, you may need to:
 3. If changes are needed in other projects, create separate worktrees and PRs for each
 4. Note cross-repo dependencies in PR descriptions (e.g., "Requires office-bohemia PR #123")
 
+## Health Check
+
+Before starting work, run `git status` and verify the worktree is clean and on the expected branch. If the worktree is dirty or on the wrong branch, report the issue and stop.
+
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks (e.g., editing files in unrelated modules simultaneously). For sequential work, single-file edits, searches, and file reads, work directly — do not spawn subagents.
+
 ## Instructions
 
 1. Read relevant source code and reference implementations before writing anything

package/playbooks/review.md

@@ -11,6 +11,10 @@ Repo: {{repo_name}} | Org: {{ado_org}} | Project: {{ado_project}}
 Review **{{pr_id}}**: {{pr_title}}
 Branch: `{{pr_branch}}`
 
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks (e.g., reviewing unrelated files simultaneously). For reading diffs, checking patterns, and writing the review, work directly — do not spawn subagents.
+
 ## How to Review
 
 1. Fetch latest and read the diff:

package/playbooks/test.md

@@ -26,6 +26,10 @@ This is a **test/build/run task**. Your goal is to build, run, test, or verify s
 5. **Test** if the task asks for it (e.g., `yarn test`, `pytest`, etc.)
 6. **Report results** — what worked, what failed, build output, test results, localhost URL if running
 
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks. For building, testing, and reporting results, work directly — do not spawn subagents.
+
 ## Rules
 
 - **Do NOT create pull requests** — this is a test/verification task only

package/playbooks/verify.md

@@ -209,6 +209,10 @@ For each project worktree:
 
 4. Note the E2E PR URLs in the testing guide.
 
+## Working Style
+
+Use subagents only for genuinely parallel, independent tasks (e.g., building separate project worktrees simultaneously). For sequential steps like reading docs, running tests, and writing the report, work directly — do not spawn subagents.
+
 ## Rules
 
 - **Read the project docs first** — never assume a build system, language, or framework