ThePalaceProject
diff --git a/‎.github/COMMIT_AND_PR_FOR_JIRA.md‎
Lines changed: 6 additions & 0 deletions b/‎.github/COMMIT_AND_PR_FOR_JIRA.md‎
Lines changed: 6 additions & 0 deletions
diff --git a/‎.github/PULL_REQUEST_TEMPLATE.md‎
Lines changed: 48 additions & 0 deletions b/‎.github/PULL_REQUEST_TEMPLATE.md‎
Lines changed: 48 additions & 0 deletions
diff --git a/‎.github/workflows/ledger.yml‎
Lines changed: 42 additions & 16 deletions b/‎.github/workflows/ledger.yml‎
Lines changed: 42 additions & 16 deletions
diff --git a/‎.github/workflows/unit-testing.yml‎
Lines changed: 57 additions & 27 deletions b/‎.github/workflows/unit-testing.yml‎
Lines changed: 57 additions & 27 deletions
@@ -1,5 +1,11 @@
 # Commit and PR format for JIRA
 
+> This is the **authored layer** of the [PR Report Contract](../docs/architecture/pr-report-contract.md).
+> The contract explains *why* the format exists (durable forensic record + fleet telemetry for an
+> agent-authored, human-orchestrated codebase); this file is the concrete commit/PR format. The PR
+> **body** is now driven by [`.github/PULL_REQUEST_TEMPLATE.md`](./PULL_REQUEST_TEMPLATE.md), which adds
+> `Evidence` / `Repro` / `Class` / `Obligations` to the Root-cause/Solution/How-to-verify below.
+
 When commits or PRs are linked to JIRA, the default post often shows only hash, author, date, one-line message, and file list. To make JIRA posts **useful** (root cause, what changed, how to verify), use the formats below.
 
 ---
 
@@ -0,0 +1,48 @@
+<!--
+  Palace iOS PR body — the durable record + fleet telemetry surface.
+  Contract: docs/architecture/pr-report-contract.md   Format: .github/COMMIT_AND_PR_FOR_JIRA.md
+  Fill every section. Delete a section ONLY if it genuinely does not apply, and say why.
+  The goal is that six months from now, git + this page alone reconstruct the whole change.
+-->
+
+## Summary
+<!-- One or two sentences. What changes, in patron-visible terms where possible. -->
+
+## Root cause
+<!-- WHY this change is needed: the mechanism of the bug, or the gap being filled.
+     Describe behavior and the call path — not a file list. -->
+
+## Solution
+<!-- WHAT this PR does: the behavior change, and the key technical decision. -->
+
+## Evidence
+<!-- Grounded-verifiability (contract §3). Every weight-bearing claim → where CI or an
+     artifact reproduces it. A number with no pointer is a hypothesis, not a verification.
+     Mark each as CI-reproduced or self-attested — do not hide the difference. -->
+- <claim, e.g. "PalaceTests full suite green"> → <CI job / artifact / exact command + exit status>
+- <claim, e.g. "TriageBotCore swift test 258/258"> → <CI job, or "self-attested — not yet CI-reproduced (Obligation below)">
+
+## Repro
+<!-- Durable reconstruction (contract §3). How the scenario was reproduced, so a future
+     reader can re-run it. Link the simdrive replay / fixture in-repo, not a dead run URL. -->
+- Scenario: <one sentence, patron-visible>   Jira: PP-XXXX
+- Replay / fixture: <.simdrive path or in-repo test> | none — <why>
+
+## Class
+<!-- Fleet-legibility (contract §4). Is this a recurrence of a known failure class?
+     Use an id from docs/regressions/recurrence-classes.md, or declare a NEW class there. -->
+- Recurrence-of: <registry-id, e.g. unbounded-live-dependency | debounce-missing | none — new class: <slug> (added to registry)>
+
+## Obligations
+<!-- Anything promised-but-not-done becomes a tracked checkbox with an owner, so a
+     roll-up can audit un-discharged promises. "will verify before merge" lives here. -->
+- [ ] <deferred / promised follow-up> — owner: <@handle or workstream>
+
+## Scope
+<!-- What this PR touches, and what it deliberately does not. -->
+
+## Not done / Deferred
+<!-- Hook-enforced for changes ≥50 prod LOC. Honest boundaries of the change. -->
+
+---
+🤖 Generated with [Claude Code](https://claude.com/claude-code)
@@ -59,9 +59,11 @@ jobs:
 
       - name: Install ledger
         run: |
-          # v0.9.6: Ledger 0.9.6 + QAAtlas 2.3.0
+          # v0.9.7: monolith app-target indexing (real blast-radius on the app
+          # target; sign-in path no longer false-LOW). See CodeAtlas
+          # fix/monolith-target-indexing. Install URL redirects to SyncTek-LLC.
           curl -fsSL https://raw.githubusercontent.com/mauricecarrier7/ledger-dist/main/install.sh \
-            | bash -s -- --version 0.9.6 --dir ./tools/bin
+            | bash -s -- --version 0.9.7 --dir ./tools/bin
           
           # Clear macOS quarantine attributes (fixes exit code 11)
           xattr -cr ./tools/bin/ 2>/dev/null || true
@@ -409,16 +411,16 @@ jobs:
             // ═══════════════════════════════════════════════════════════════
             // TEST COVERAGE SECTION (QAAtlas) - only if qa domain enabled
             // ═══════════════════════════════════════════════════════════════
-            body += '---\n';
-            body += '### 🧪 Test Coverage (via QAAtlas)\n\n';
-            
-            // Check if we have valid QA data
+            // TEST COVERAGE (QAAtlas): the `qa` domain is disabled
+            // (DOMAINS="arch,reach,a11y" at the top of this file), so this
+            // section stays SILENT rather than posting a standing "set
+            // OPENAI_API_KEY" banner on every PR — a constant banner is noise
+            // that trains readers to skim past warnings
+            // (pr-report-contract.md §2, delta-or-silent).
             const hasQAData = qaCoverage && parseFloat(qaCoverage) > 0 || (parseInt(qaTested) > 0 || parseInt(qaTotal) > 0);
-            
-            if (!hasQAData) {
-              body += '> ⚠️ **No coverage data available** - QAAtlas requires an OpenAI API key.\n';
-              body += '> Set `OPENAI_API_KEY` in repository secrets to enable AI-powered test coverage analysis.\n\n';
-            } else {
+            if (hasQAData) {
+              body += '---\n';
+              body += '### 🧪 Test Coverage (via QAAtlas)\n\n';
               const coverageNum = parseFloat(qaCoverage);
               const coverageIcon = coverageNum >= 50 ? '🟢' : coverageNum >= 30 ? '🟡' : '🔴';
               body += `**Coverage Estimate:** ${coverageIcon} **${qaCoverage}%**\n\n`;
@@ -519,15 +521,31 @@ jobs:
             body += `| Layer Violations | ${violations} | Dependencies that break architectural boundaries |\n`;
             body += `| Hotspots | ${hotspots} | Files with high complexity + frequent changes |\n`;
             body += `| Avg Coupling | ${avgCoupling} | How interconnected modules are (lower is better, <1.0 is good) |\n\n`;
-            
+
+            // Honesty note: these are repo-BASELINE metrics, not diff deltas —
+            // they read identically on most PRs. Don't misread a constant as
+            // PR-caused. A true base-vs-head delta is a tracked Obligation
+            // (pr-report-contract.md §2).
+            if (changedFiles === '0') {
+              body += `> ℹ️ This diff changed **0** architecture-relevant files — the numbers above are the repo baseline, unchanged by this PR.\n\n`;
+            } else {
+              body += `> ℹ️ Baseline metrics (whole-repo), not a diff delta. Watch for a value that *moves* vs. \`develop\`, not its absolute level.\n\n`;
+            }
+
             if (parseInt(archIssues) === 0 && cycles === '0' && violations === '0') {
               body += '✅ **Architecture looks healthy!**\n\n';
             } else {
               // Show issues summary
               if (parseInt(archIssues) > 0) {
                 body += `⚠️ Found **${archIssues}** architectural issue(s).\n\n`;
               }
-              
+              // A layer violation with no issue count still gets NAMED, not left
+              // as a bare number the reader can't act on (pr-report-contract.md
+              // gap #8). Surfaces the same findings file below.
+              if (parseInt(violations) > 0) {
+                body += `⚠️ **${violations}** layer violation(s) — see **Architecture Findings** below for the offending dependency edge(s).\n\n`;
+              }
+
               // Expandable: Dependency Cycles
               if (parseInt(cycles) > 0) {
                 body += '<details>\n<summary><strong>🔄 Dependency Cycles</strong></summary>\n\n';
@@ -544,8 +562,9 @@ jobs:
                 body += '\n</details>\n\n';
               }
               
-              // Expandable: All Findings
-              if (parseInt(archIssues) > 0) {
+              // Expandable: All Findings — open when a violation exists so the
+              // offending edge is named, not just counted (gap #8).
+              if (parseInt(archIssues) > 0 || parseInt(violations) > 0) {
                 body += '<details>\n<summary><strong>📋 Architecture Findings</strong></summary>\n\n';
                 try {
                   const findings = fs.readFileSync('/tmp/arch_findings.md', 'utf8');
@@ -569,7 +588,14 @@ jobs:
             body += '> 💡 *Detects code that cannot be reached from entry points (dead code).*\n\n';
             
             if (parseInt(reachIssues) === 0) {
-              body += '✅ **No dead code detected**\n\n';
+              // Don't present a vacuous truth as a green check: "no dead code"
+              // over ZERO analyzed files is meaningless, and a fake ✅ trains
+              // readers to trust the real ones less (pr-report-contract.md §2).
+              if (changedFiles === '0') {
+                body += 'ℹ️ No architecture-relevant files in this diff — reachability not evaluated.\n\n';
+              } else {
+                body += '✅ **No dead code detected**\n\n';
+              }
             } else {
               body += `Found **${reachIssues}** potentially unreachable symbol(s).\n`;
               body += `> These functions/classes may be unused and could be candidates for removal.\n\n`;
 
@@ -762,25 +762,42 @@ jobs:
               }
               body += `**${tests} tests** | **${passed} passed** | **${failed} failed** | **${skipped} skipped** | ⏱️ ${duration} | 📊 ${passRate}${coverageDisplay}\n\n`;
               
-              // Class breakdown table — collapsed by default. Per-class
-              // rows are valuable when triaging a failure but noise on a
-              // green PR. Reviewers click through to expand.
+              // Full per-class matrix — ALWAYS collapsed, never dumped open
+              // (pr-report-contract.md §1, decision-first). ~900 green rows are
+              // noise a reviewer must not scroll past; failures are surfaced
+              // separately below. Skipped column added so an all-skipped class
+              // reconciles instead of rendering as a false-green `7|0|0`;
+              // ⚠️ flags any row where passed+failed+skipped != total.
               if (classSummary && classSummary.trim()) {
                 const classLines = classSummary.trim().split('\n').filter(l => l.split('|')[0]);
-                const failedClasses = classLines.filter(l => l.split('|')[3] !== '0').length;
+                const failedClasses = classLines.filter(l => (l.split('|')[3] || '0') !== '0').length;
                 const summaryLabel = failedClasses > 0
-                  ? `<strong>Tests by Class</strong> — ${classLines.length} classes, <strong>${failedClasses} with failures</strong>`
-                  : `<strong>Tests by Class</strong> — ${classLines.length} classes (click to expand)`;
-                body += `<details${failedClasses > 0 ? ' open' : ''}>\n<summary>${summaryLabel}</summary>\n\n`;
-                body += `| Class | Tests | Passed | Failed | Duration |\n`;
-                body += `|-------|-------|--------|--------|----------|\n`;
+                  ? `<strong>All ${classLines.length} classes</strong> — ${failedClasses} with failures (full matrix)`
+                  : `<strong>All ${classLines.length} classes</strong> (full matrix — click to expand)`;
+                body += `<details>\n<summary>${summaryLabel}</summary>\n\n`;
+                body += `| Class | Tests | Passed | Failed | Skipped | Duration |\n`;
+                body += `|-------|-------|--------|--------|---------|----------|\n`;
                 classLines.forEach(line => {
-                  const [cls, total, pass, fail, dur] = line.split('|');
-                  const icon = fail === '0' ? '✅' : '❌';
+                  // 6-field format (cls|total|pass|fail|skip|dur); tolerate the
+                  // legacy 5-field (cls|total|pass|fail|dur) during rollout.
+                  const parts = line.split('|');
+                  const cls = parts[0];
+                  const total = parts[1] || '0';
+                  const pass = parts[2] || '0';
+                  const fail = parts[3] || '0';
+                  const hasSkip = parts.length >= 6;
+                  const skip = hasSkip ? (parts[4] || '0') : '0';
+                  const dur = hasSkip ? parts[5] : (parts[4] || '-');
+                  const reconciles = (parseInt(pass) + parseInt(fail) + parseInt(skip)) === parseInt(total);
+                  const icon = parseInt(fail) > 0 ? '❌' : (reconciles ? '✅' : '⚠️');
                   const failCell = fail === '0' ? fail : `**${fail}**`;
-                  body += `| ${icon} ${cls} | ${total} | ${pass} | ${failCell} | ${dur} |\n`;
+                  body += `| ${icon} ${cls} | ${total} | ${pass} | ${failCell} | ${skip} | ${dur} |\n`;
                 });
-                body += `\n</details>\n\n`;
+                body += `\n`;
+                if (reportUrl) {
+                  body += `📊 Full interactive matrix: **[report](${reportUrl})**\n`;
+                }
+                body += `</details>\n\n`;
               }
             } else {
               if (testOutcome === 'success') {
@@ -795,10 +812,16 @@ jobs:
               }
             }
             
-            // Failed tests details
+            // Failing detail — OPEN when red so the reviewer lands on the
+            // actionable rows, not the collapsed matrix (contract §1).
             if (failedTests && failedTests.trim()) {
-              body += `<details>\n<summary><strong>Failed Tests (click to expand)</strong></summary>\n\n`;
+              body += `<details open>\n<summary>❌ <strong>Failed tests</strong></summary>\n\n`;
               body += '```\n' + failedTests.trim() + '\n```\n';
+              // Names only — parse-xcresult emits identifiers, not messages.
+              // Inlining the failure message + first stack frame is a tracked
+              // Obligation (pr-report-contract.md gap #3); until then, the
+              // interactive report carries the detail.
+              body += `\n_Names only — open the interactive report above for messages + stack frames._\n`;
               body += `</details>\n\n`;
             }
             
@@ -833,28 +856,33 @@ jobs:
             body += `not that we shipped less UI. Total coverage is kept for continuity. `;
             body += `The excluded paths are covered by simdrive E2E journeys (see \`chaos-replay-on-pr.yml\`).\n\n`;
             
-            // Trends
-            if (newFailures || fixedTests || flakyTests || testCountChange) {
-              body += `### 📈 Trends\n\n`;
-              
+            // What changed vs. base — the orchestrator's first question
+            // (contract §1). Delta-or-silent: this whole section renders only
+            // when the .test-history cache hits (populating deltas). When it's
+            // empty the section is silent by design, not a green claim — the
+            // develop cache-warm job that makes deltas reliable is a tracked
+            // Obligation (pr-report-contract.md gap #4).
+            if ((newFailures && newFailures.trim()) || (fixedTests && fixedTests.trim()) || (flakyTests && flakyTests.trim()) || testCountChange) {
+              body += `### 📈 What changed vs. base\n\n`;
+
               if (testCountChange) {
-                body += `Test count change: **${testCountChange}**\n\n`;
+                body += `Test count: **${testCountChange}**\n\n`;
               }
-              
+
               if (newFailures && newFailures.trim()) {
-                body += `<details>\n<summary>⚠️ <strong>New Failures</strong></summary>\n\n`;
+                body += `<details open>\n<summary>⚠️ <strong>New failures</strong> — passing on base, failing on this branch</summary>\n\n`;
                 body += '```\n' + newFailures.trim() + '\n```\n';
                 body += `</details>\n\n`;
               }
-              
+
               if (fixedTests && fixedTests.trim()) {
-                body += `<details>\n<summary>✅ <strong>Fixed Tests</strong></summary>\n\n`;
+                body += `<details>\n<summary>✅ <strong>Fixed</strong> — failing on base, passing on this branch</summary>\n\n`;
                 body += '```\n' + fixedTests.trim() + '\n```\n';
                 body += `</details>\n\n`;
               }
-              
+
               if (flakyTests && flakyTests.trim()) {
-                body += `<details>\n<summary>⚡ <strong>Flaky Tests Detected</strong></summary>\n\n`;
+                body += `<details>\n<summary>⚡ <strong>Flaky</strong> — inconsistent across recent runs</summary>\n\n`;
                 body += '```\n' + flakyTests.trim() + '\n```\n';
                 body += `</details>\n\n`;
               }
@@ -866,12 +894,14 @@ jobs:
               body += `Download the **snapshot-failures** artifact to view visual differences.\n\n`;
             }
             
-            // Links section
+            // Links + provenance. The counts above are produced by THIS CI run
+            // parsing the xcresult — not author-attested (contract §2,§3).
             body += `---\n`;
             if (reportUrl) {
               body += `🔗 **[Interactive HTML Report](${reportUrl})** | `;
             }
             body += `**[CI Run Details](${runUrl})**\n\n`;
+            body += `_Counts above were produced by this CI run's xcresult parse — reproduce via the run link._\n\n`;
             
             // Artifacts table
             body += `<details>\n<summary>📦 <strong>Downloadable Artifacts</strong></summary>\n\n`;