Snapshot WIP: solver HP epic progress, BPHX/HX physics, BMAD skill refresh.

Capture uncommitted solver robustness work (regularization, domain errors, linear solver lifecycle, tube DP/MSH), web workbench updates, and synced BMAD skills across IDE agent folders before starting BPHX pressure-drop.

Co-authored-by: Cursor <cursoragent@cursor.com>

.agents/skills/ui-ux-pro-max/scripts/validate_data.py

@@ -0,0 +1,114 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Data integrity guardrail for ui-ux-pro-max. Stdlib-only, no pytest dependency,
+so it can run as a standalone pre-publish/CI check:
+
+    python validate_data.py
+
+Checks, per configured domain/stack CSV:
+  - file exists
+  - header row contains every column referenced in search_cols/output_cols
+  - no duplicate primary-key values (first column) within a file
+  - any "Decision_Rules"-style JSON column parses as JSON
+
+Exits 0 with no output on success; exits 1 and prints every problem found
+on failure (fail-fast is the wrong call here -- a data change can break
+several files at once, so we want the full list in one run).
+"""
+
+import csv
+import json
+import sys
+from pathlib import Path
+
+from core import CSV_CONFIG, STACK_CONFIG, _STACK_COLS, DATA_DIR
+
+# REASONING_FILE lives in design_system.py, not core.py -- redeclared here to
+# avoid a circular import (design_system.py imports core.py).
+REASONING_FILE = "ui-reasoning.csv"
+JSON_COLUMNS = {"Decision_Rules"}
+
+
+def _read_rows(filepath):
+    with open(filepath, "r", encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        return reader.fieldnames or [], list(reader)
+
+
+def _check_file(label, filepath, search_cols, output_cols, problems):
+    if not filepath.exists():
+        problems.append(f"[{label}] missing file: {filepath}")
+        return
+
+    try:
+        headers, rows = _read_rows(filepath)
+    except (csv.Error, UnicodeDecodeError, OSError) as e:
+        problems.append(f"[{label}] failed to parse {filepath.name}: {e}")
+        return
+
+    header_set = set(headers)
+    for col in set(search_cols) | set(output_cols):
+        if col not in header_set:
+            problems.append(f"[{label}] {filepath.name}: expected column '{col}' not found in header")
+
+    # Only check for duplicates against an actual identifier column ("No" is
+    # the sequential-index convention used across this dataset). The first
+    # CSV column is not reliably a unique key -- e.g. stack files use
+    # "Category", which legitimately repeats across many guideline rows.
+    if "No" in header_set:
+        seen = {}
+        for i, row in enumerate(rows, start=2):  # +1 header, +1 to be 1-indexed
+            key = row.get("No", "")
+            if key in seen:
+                problems.append(
+                    f"[{label}] {filepath.name}: duplicate 'No' value '{key}' on rows {seen[key]} and {i}"
+                )
+            else:
+                seen[key] = i
+    elif label.startswith("stack:"):
+        problems.append(
+            f"[{label}] {filepath.name}: missing 'No' index column present in other stack files "
+            "(schema drift -- harmless for search, but inconsistent with the rest of data/stacks/)"
+        )
+
+    for row_idx, row in enumerate(rows, start=2):
+        for col in JSON_COLUMNS:
+            if col in row and row[col]:
+                try:
+                    json.loads(row[col])
+                except json.JSONDecodeError as e:
+                    problems.append(
+                        f"[{label}] {filepath.name} row {row_idx}: column '{col}' is not valid JSON: {e}"
+                    )
+
+
+def main():
+    problems = []
+
+    for domain, config in CSV_CONFIG.items():
+        _check_file(f"domain:{domain}", DATA_DIR / config["file"],
+                    config["search_cols"], config["output_cols"], problems)
+
+    for stack, config in STACK_CONFIG.items():
+        _check_file(f"stack:{stack}", DATA_DIR / config["file"],
+                    _STACK_COLS["search_cols"], _STACK_COLS["output_cols"], problems)
+
+    reasoning_path = DATA_DIR / REASONING_FILE
+    if reasoning_path.exists():
+        _check_file("reasoning", reasoning_path, ["UI_Category"], ["UI_Category", "Decision_Rules"], problems)
+    else:
+        problems.append(f"[reasoning] missing file: {reasoning_path}")
+
+    if problems:
+        print(f"FAILED: {len(problems)} data integrity issue(s) found:\n")
+        for p in problems:
+            print(f"  - {p}")
+        sys.exit(1)
+
+    print(f"OK: validated {len(CSV_CONFIG)} domain files, {len(STACK_CONFIG)} stack files, and ui-reasoning.csv")
+    sys.exit(0)
+
+
+if __name__ == "__main__":
+    main()