chore: add real content that was sitting untracked

Surfaced by today's untracked-files audit. None of these are accidents —
multiple are referenced by name in CLAUDE.md and memory files but were
never added.

Categories:
- docs/PHASE_AUDIT_GUIDE.md (106 LOC) — Claude Code phase audit guidance
- ops/systemd/lakehouse-langfuse-bridge.service — Langfuse bridge unit
- package.json — top-level npm manifest
- scripts/e2e_pipeline_check.sh + production_smoke.sh — real test scripts
- reports/kimi/audit-last-week*.md — the "Two reports live" CLAUDE.md cites
- tests/multi-agent/scenarios/ — 44 staffing scenarios (cutover decision A)
- tests/multi-agent/playbooks/ — 102 playbook records
- tests/battery/, tests/agent_test/PRD.md, tests/real-world/* — real tests
- sidecar/sidecar/{lab_ui,pipeline_lab}.py — 888 LOC dev-only UIs that
  remain in service post-sidecar-drop (commit ba928b1 explicitly kept them)

Sensitivity check: scenarios use synthetic company names ("Heritage Foods",
"Cornerstone Fabrication"); audit reports describe code findings only;
no PII or secrets surfaced.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/multi-agent/playbooks/scenario-2026-04-21T01-28-36/prior_lessons.json

@@ -0,0 +1,54 @@
+[
+  {
+    "date": "2026-04-21",
+    "client": "Riverfront Steel",
+    "cities": "Toledo",
+    "states": "OH",
+    "events_total": 5,
+    "events_ok": 4,
+    "checkpoint_count": 2,
+    "model": "gpt-oss:20b",
+    "cloud": false,
+    "lesson": "**  \nBefore launching an expansion, always verify the pool size and certification status for the role; if the pool is unknown or certifications are missing, pre‑fetch the data or secure temporary hires to avoid a fail. For any misplacement or recurring assignment, cross‑check the booking system to confirm that staff such as Patrick Ross are not double‑booked. Maintain a buffer of at least three extra candidates for emergency roles to keep gaps at a minimum. This proactive verification will reduce drift risk and scheduling conflicts in future runs.",
+    "checkpoints": [
+      {
+        "after": "12:15",
+        "risk": "Forklift Operator drift risk in Toledo",
+        "hint": "Verify forklift operator certifications in Toledo; consider temporary hires if drift persists."
+      },
+      {
+        "after": "15:45",
+        "risk": "Double-booking risk for Patrick Ross",
+        "hint": "Verify Patrick Ross’s schedule before assigning; use the booking system to prevent overlapping shifts."
+      }
+    ],
+    "created_at": "2026-04-21T01:13:14.575Z",
+    "file": "2026-04-21_Riverfront_Steel_1776733994575.json"
+  },
+  {
+    "date": "2026-04-21",
+    "client": "Riverfront Steel",
+    "cities": "Toledo",
+    "states": "OH",
+    "events_total": 5,
+    "events_ok": 4,
+    "checkpoint_count": 2,
+    "model": "gpt-oss:20b",
+    "cloud": false,
+    "lesson": "**  \nWhen expanding a shift that requires multiple forklift operators, pull a fresh subset of operators from the pool that excludes those already assigned to the Toledo shift; maintain a temporary exclusion list to avoid duplicate assignments. For any misplacement or re‑assignment, lock the worker’s schedule before assigning a new shift to prevent double‑booking—especially for high‑demand staff like Ryan Hughes. Verify the pool size and current turns before processing an emergency request; if the pool is insufficient, trigger a contingency plan. Keep a real‑time log of all assignments to catch overlaps early.",
+    "checkpoints": [
+      {
+        "after": "12:15",
+        "risk": "Duplicate worker assignments risk",
+        "hint": "Avoid double booking: use different forklift operators for Toledo shift."
+      },
+      {
+        "after": "15:45",
+        "risk": "Double-book Ryan Hughes",
+        "hint": "Verify Ryan Hughes' schedule before assigning any new shift to avoid overlapping bookings."
+      }
+    ],
+    "created_at": "2026-04-21T01:26:26.982Z",
+    "file": "2026-04-21_Riverfront_Steel_1776734786982.json"
+  }
+]