@xn-intenton-z2a/agentic-lib 7.4.31 → 7.4.32

package/.github/workflows/agentic-lib-test.yml

@@ -40,6 +40,10 @@ on:
         type: string
         required: false
         default: "false"
+      log-branch:
+        type: string
+        required: false
+        default: "agentic-lib-logs"
       ref:
         type: string
         required: false

package/.github/workflows/agentic-lib-workflow.yml

@@ -809,6 +809,21 @@ jobs:
           instructions: ".github/agents/agent-implementation-review.md"
           model: ${{ needs.params.outputs.model }}
 
+      # Commit MISSION.md checkbox updates (from Fix #6) so downstream jobs see them
+      - name: Commit acceptance criteria updates
+        if: github.repository != 'xn-intenton-z2a/agentic-lib' && needs.params.outputs.dry-run != 'true'
+        run: |
+          if git diff --quiet MISSION.md 2>/dev/null; then
+            echo "No MISSION.md changes to commit"
+          else
+            git config user.name "github-actions[bot]"
+            git config user.email "github-actions[bot]@users.noreply.github.com"
+            git add MISSION.md
+            git commit -m "agentic-step: update acceptance criteria checkboxes [skip ci]"
+            git pull --rebase origin main 2>/dev/null || true
+            git push origin HEAD:main || echo "::warning::Could not push MISSION.md updates"
+          fi
+
       - name: Push log to log branch
         if: github.repository != 'xn-intenton-z2a/agentic-lib' && needs.params.outputs.dry-run != 'true'
         env:
@@ -1741,6 +1756,16 @@ jobs:
           echo "- Website: [${SITE_URL}](${SITE_URL})" >> $GITHUB_STEP_SUMMARY
 
       # W14: Post-merge director check — re-evaluate mission status after dev/PR merges
+      # Pull latest main to include dev merge (checkout was at maintain commit)
+      - name: Pull latest main (include dev merge)
+        if: |
+          needs.params.outputs.mission-complete != 'true' &&
+          needs.params.outputs.dry-run != 'true' &&
+          needs.director.outputs.decision != 'mission-complete' &&
+          needs.director.outputs.decision != 'mission-failed' &&
+          github.repository != 'xn-intenton-z2a/agentic-lib'
+        run: git pull origin main --ff-only 2>/dev/null || true
+
       - name: Fetch log and screenshot from log branch (post-merge director)
         if: |
           needs.params.outputs.mission-complete != 'true' &&
@@ -1831,6 +1856,8 @@ jobs:
     uses: ./.github/workflows/agentic-lib-test.yml
     with:
       ref: ${{ needs.maintain.outputs.commit-sha || github.ref }}
+      push-screenshot: "true"
+      log-branch: ${{ needs.params.outputs.log-branch }}
     secrets: inherit
 
   # ─── Schedule change (if requested) ────────────────────────────────

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xn-intenton-z2a/agentic-lib",
-  "version": "7.4.31",
+  "version": "7.4.32",
   "description": "Agentic-lib Agentic Coding Systems SDK powering automated GitHub workflows.",
   "type": "module",
   "scripts": {

package/src/actions/agentic-step/tasks/direct.js

@@ -65,7 +65,7 @@ function detectDedicatedTests() {
  * Build the metric-based mission-complete advisory string.
  * This is the mechanical check — purely rule-based, no LLM.
  */
-function buildMetricAssessment(ctx, config) {
+async function buildMetricAssessment(ctx, config) {
   const thresholds = config.missionCompleteThresholds || {};
   const minResolved = thresholds.minResolvedIssues ?? 3;
   const maxTodos = thresholds.maxSourceTodos ?? 0;
@@ -78,6 +78,12 @@ function buildMetricAssessment(ctx, config) {
   } catch { /* ignore parse errors */ }
   const criticalGaps = reviewGaps.filter((g) => g.severity === "critical");
 
+  // Acceptance criteria from MISSION.md checkboxes
+  const { countAcceptanceCriteria } = await import("../../../copilot/telemetry.js");
+  const missionPath = config.paths?.mission?.path || "MISSION.md";
+  const acceptance = countAcceptanceCriteria(missionPath);
+  const acceptanceMet = acceptance.total > 0 && acceptance.met > acceptance.total / 2;
+
   // C6: Removed "Dedicated tests" metric; using cumulative transforms instead
   const metrics = [
     { metric: "Open issues", value: ctx.issuesSummary.length, target: 0, met: ctx.issuesSummary.length === 0 },
@@ -87,6 +93,7 @@ function buildMetricAssessment(ctx, config) {
     { metric: "Cumulative transforms", value: ctx.cumulativeTransformationCost, target: 1, met: ctx.cumulativeTransformationCost >= 1 },
     { metric: "Budget", value: ctx.cumulativeTransformationCost, target: ctx.transformationBudget || "unlimited", met: !(ctx.transformationBudget > 0 && ctx.cumulativeTransformationCost >= ctx.transformationBudget) },
     { metric: "Implementation review", value: criticalGaps.length === 0 ? "No critical gaps" : `${criticalGaps.length} critical gap(s)`, target: "No critical gaps", met: criticalGaps.length === 0 },
+    { metric: "Acceptance criteria", value: acceptance.total > 0 ? `${acceptance.met}/${acceptance.total}` : "N/A", target: "> 50%", met: acceptanceMet },
   ];
 
   const allMet = metrics.every((m) => m.met);
@@ -158,6 +165,9 @@ function buildPrompt(ctx, agentInstructions, metricAssessment) {
     "Consider the implementation review findings — if critical gaps exist, do NOT declare mission-complete.",
     "Check the acceptance criteria in the Mission section above. If all criteria are clearly satisfied by the current source code and tests (verified via read_file), you SHOULD declare mission-complete even if not all mechanical metrics are MET.",
     "For simple missions (few functions, clear acceptance criteria), do not require elaborate test coverage or documentation beyond what the acceptance criteria specify.",
+    "",
+    "**Post-merge evaluation context:** This director runs AFTER a dev transformation has been merged. The source code, tests, README, and website you see are the result of that merge. The acceptance criteria checkboxes in MISSION.md reflect the implementation review's findings. If the metrics show all conditions MET and the acceptance criteria are > 50% checked, you should declare mission-complete unless you find a critical implementation gap via read_file. Do not defer to a future run — the pipeline has a structural 2-run minimum, and this is your chance to complete in 1 run.",
+    "",
     "Then call report_director_decision with your determination.",
     "",
     "**You MUST call report_director_decision exactly once.**",
@@ -485,7 +495,7 @@ export async function direct(context) {
   };
 
   // Build metric-based advisory
-  const metricAssessment = buildMetricAssessment(ctx, config);
+  const metricAssessment = await buildMetricAssessment(ctx, config);
   core.info(`Metric assessment: ${metricAssessment.assessment}`);
 
   // --- LLM decision via hybrid session ---

package/src/actions/agentic-step/tasks/implementation-review.js

@@ -48,7 +48,10 @@ function buildReviewPrompt(mission, config, agentInstructions, agentLogsSummary)
     "   - Tests that don't assert anything meaningful (empty/trivial)",
     "   - Features listed as done in docs but missing from code",
     "   - PRs merged without test coverage for the claimed feature",
-    "4. Call report_implementation_review with your findings.",
+    "4. Check the MISSION.md Acceptance Criteria checkboxes (`- [ ]`). For each criterion,",
+    "   if you verified it is implemented AND unit-tested, include its exact text in the",
+    "   `acceptanceCriteriaMet` array. Copy the criterion text exactly as it appears after `- [ ]`.",
+    "5. Call report_implementation_review with your findings.",
     "",
     "**You MUST call report_implementation_review exactly once.**",
   ].join("\n");
@@ -166,15 +169,47 @@ export async function implementationReview(context) {
             },
             description: "Metrics that may be misleading about actual progress",
           },
+          acceptanceCriteriaMet: {
+            type: "array",
+            items: { type: "string" },
+            description: "Exact text of each acceptance criterion from MISSION.md that is verified as implemented AND unit-tested. Copy the text after '- [ ]' exactly.",
+          },
         },
         required: ["elements", "gaps", "advice"],
       },
-      handler: async ({ elements, gaps, advice, misleadingMetrics }) => {
+      handler: async ({ elements, gaps, advice, misleadingMetrics, acceptanceCriteriaMet }) => {
         reviewResult.elements = elements || [];
         reviewResult.gaps = gaps || [];
         reviewResult.advice = advice || "";
         reviewResult.misleadingMetrics = misleadingMetrics || [];
-        return { textResultForLlm: `Review recorded: ${elements?.length || 0} elements traced, ${gaps?.length || 0} gaps found` };
+
+        // Update MISSION.md checkboxes based on verified acceptance criteria
+        const metCriteria = acceptanceCriteriaMet || [];
+        if (metCriteria.length > 0) {
+          try {
+            const missionPath = config.paths?.mission?.path || "MISSION.md";
+            const { readFileSync, writeFileSync } = await import("fs");
+            let missionContent = readFileSync(missionPath, "utf8");
+            let checkedCount = 0;
+            for (const criterionText of metCriteria) {
+              // Match the checkbox line containing this criterion text (fuzzy: trim whitespace)
+              const escaped = criterionText.replace(/[.*+?^${}()|[\]\\]/g, "\\$&").trim();
+              const re = new RegExp(`- \\[ \\] ${escaped}`);
+              if (re.test(missionContent)) {
+                missionContent = missionContent.replace(re, `- [x] ${criterionText.trim()}`);
+                checkedCount++;
+              }
+            }
+            if (checkedCount > 0) {
+              writeFileSync(missionPath, missionContent, "utf8");
+              core.info(`Updated ${checkedCount} acceptance criteria checkboxes in ${missionPath}`);
+            }
+          } catch (err) {
+            core.warning(`Could not update MISSION.md checkboxes: ${err.message}`);
+          }
+        }
+
+        return { textResultForLlm: `Review recorded: ${elements?.length || 0} elements traced, ${gaps?.length || 0} gaps found, ${metCriteria.length} criteria checked` };
       },
     });
 

package/src/scripts/push-to-logs.sh

@@ -132,6 +132,31 @@ else
         git rebase --abort 2>/dev/null || true
       }
     }
+
+    # W3b: After successful rebase (no conflict), re-apply our boolean true values.
+    # The rebase may have replayed our commit on top of a remote state with false values,
+    # causing our true values to be lost. Re-apply them from the saved copy.
+    STATE_FILE="agentic-lib-state.toml"
+    if [ -f "${TMPDIR}/ours-${STATE_FILE}" ] && [ -f "$STATE_FILE" ]; then
+      NEEDS_AMEND=false
+      while IFS='=' read -r key val; do
+        key=$(echo "$key" | xargs)
+        val=$(echo "$val" | xargs)
+        if [ "$val" = "true" ]; then
+          CURRENT=$(grep "^${key} = " "$STATE_FILE" 2>/dev/null | sed 's/.*= *//' | xargs || true)
+          if [ "$CURRENT" != "true" ]; then
+            sed -i "s/^${key} = .*/${key} = true/" "$STATE_FILE" 2>/dev/null || \
+              sed -i'' "s/^${key} = .*/${key} = true/" "$STATE_FILE" 2>/dev/null || true
+            NEEDS_AMEND=true
+          fi
+        fi
+      done < "${TMPDIR}/ours-${STATE_FILE}"
+      if [ "$NEEDS_AMEND" = "true" ]; then
+        echo "push-to-logs: re-applied boolean true values after rebase"
+        git add "$STATE_FILE"
+        git commit --amend --no-edit 2>/dev/null || true
+      fi
+    fi
     sleep $((attempt * 2))
     if [ "$attempt" -eq "$MAX_RETRIES" ]; then
       echo "::warning::push-to-logs: failed to push after $MAX_RETRIES attempts"

package/src/seeds/zero-package.json

@@ -17,7 +17,7 @@
   "author": "",
   "license": "MIT",
   "dependencies": {
-    "@xn-intenton-z2a/agentic-lib": "^7.4.31"
+    "@xn-intenton-z2a/agentic-lib": "^7.4.32"
   },
   "devDependencies": {
     "@playwright/test": "^1.58.0",