codymaster 4.1.4 → 4.3.0

package/skills/marketplace-report-crawler/scripts/setup-folders.sh

@@ -0,0 +1,83 @@
+#!/bin/bash
+# ============================================================================
+# Marketplace Report Folder Structure Generator
+# ============================================================================
+# Usage: bash setup-folders.sh [output_root] [yearMonth]
+# Example: bash setup-folders.sh ./data 202602
+#
+# Creates organized folder tree:
+#   {root}/{Platform}/{Brand}/{report_type}/
+#   Files will be saved as: {YYYYMM}_{Brand}_{report_type}.{ext}
+# ============================================================================
+
+set -e
+
+# --- Configuration ---
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CONFIG_DIR="${SCRIPT_DIR}/../config"
+OUTPUT_ROOT="${1:-/Users/todyle/Builder/Boxme/AutomateReport/data}"
+YEAR_MONTH="${2:-$(date -v-1m +%Y%m 2>/dev/null || date -d '-1 month' +%Y%m)}"
+
+echo "=============================================="
+echo "📁 Marketplace Report Folder Generator"
+echo "=============================================="
+echo "📂 Output Root: ${OUTPUT_ROOT}"
+echo "📅 Period: ${YEAR_MONTH}"
+echo ""
+
+# --- Read configs using Node.js (cross-platform JSON parsing) ---
+if ! command -v node &> /dev/null; then
+  echo "❌ Node.js is required. Please install it first."
+  exit 1
+fi
+
+# Extract accounts and report types from JSON configs
+ACCOUNTS=$(node -e "
+const accounts = require('${CONFIG_DIR}/accounts.json').accounts;
+const reportTypes = require('${CONFIG_DIR}/report-types.json').reportTypes;
+
+// For each active account, list platform + report types
+accounts.filter(a => a.active).forEach(account => {
+  account.platforms.forEach(platform => {
+    const reports = reportTypes
+      .filter(r => r.platform === platform && r.status !== 'disabled')
+      .map(r => r.id);
+    // Deduplicate report IDs
+    const uniqueReports = [...new Set(reports)];
+    uniqueReports.forEach(reportId => {
+      console.log(\`\${platform}|\${account.brandSlug}|\${reportId}\`);
+    });
+  });
+});
+")
+
+# --- Create folders ---
+FOLDER_COUNT=0
+
+echo "$ACCOUNTS" | while IFS='|' read -r platform brand reportType; do
+  # Capitalize platform name for folder
+  PLATFORM_DIR=$(echo "$platform" | awk '{print toupper(substr($0,1,1)) tolower(substr($0,2))}')
+  
+  FOLDER_PATH="${OUTPUT_ROOT}/${PLATFORM_DIR}/${brand}/${reportType}"
+  
+  if [ ! -d "$FOLDER_PATH" ]; then
+    mkdir -p "$FOLDER_PATH"
+    echo "  ✅ Created: ${PLATFORM_DIR}/${brand}/${reportType}/"
+  else
+    echo "  ⏭️  Exists:  ${PLATFORM_DIR}/${brand}/${reportType}/"
+  fi
+  
+  FOLDER_COUNT=$((FOLDER_COUNT + 1))
+done
+
+echo ""
+echo "=============================================="
+echo "✅ Folder structure ready!"
+echo ""
+echo "📝 File naming convention:"
+echo "   {YYYYMM}_{Brand}_{report_type}.{ext}"
+echo "   Example: ${YEAR_MONTH}_Vita_Dairy_platform_income.xlsx"
+echo ""
+echo "📂 Full path example:"
+echo "   ${OUTPUT_ROOT}/Shopee/Vita_Dairy/platform_income/${YEAR_MONTH}_Vita_Dairy_platform_income.xlsx"
+echo "=============================================="

package/skills/medical-research/SKILL.md

@@ -0,0 +1,194 @@
+---
+name: medical-research
+description: Evidence-based medical writing protocol for OB/GYN research articles. Ensures citation standards, Level of Evidence grading, anti-hallucination rules, and clinical accuracy markers. Use when generating or reviewing medical content.
+allowed-tools: Read, Write, Edit, Bash
+version: 1.0
+priority: HIGH
+---
+
+# Medical Research — Evidence-Based Writing Protocol
+
+> **Mandatory** when writing medical articles for the OB/GYN RND system.
+
+---
+
+## 1. Evidence Grading — Oxford CEBM
+
+Every recommendation/protocol in articles **MUST** include Level of Evidence:
+
+| Level | Evidence Type | In-Article Notation |
+|-------|--------------|---------------------|
+| **I** | Systematic Review / Meta-analysis / Large RCT | `[LoE: I]` |
+| **II** | Small RCT / High-quality Cohort study | `[LoE: II]` |
+| **III** | Case-control study / Retrospective Cohort | `[LoE: III]` |
+| **IV** | Case series / Cross-sectional | `[LoE: IV]` |
+| **V** | Expert opinion / Consensus | `[LoE: V]` |
+
+**Rules:**
+- Each pathology article (Tier 2) must have **at least 3 recommendations with LoE**
+- Each technique article (Tier 3) must have **at least 2 recommendations with LoE**
+- Each foundation article (Tier 1) must have **at least 1 LoE**
+- Include LoE **inline** right after the recommendation, e.g.: `Aspirin 150mg from week 12 for high-risk group [LoE: I — ASPRE trial 2017]`
+
+---
+
+## 2. Citation Standard — Mandatory References
+
+### Accepted Sources (in priority order)
+
+| Priority | Source | Citation Example |
+|----------|--------|-----------------|
+| **P0** | National MoH Guidelines | `[MoH 2023 — Diagnosis and Treatment Guidelines]` |
+| **P1** | ACOG Practice Bulletin | `[ACOG PB #234, 2024]` |
+| **P1** | ASRM Committee Opinion | `[ASRM CO 2023]` |
+| **P1** | ESHRE Guideline | `[ESHRE Guideline 2024]` |
+| **P2** | WHO Recommendation | `[WHO 2023]` |
+| **P2** | Cochrane Systematic Review | `[Cochrane 2023]` |
+| **P3** | NICE / RCOG | `[NICE NG000, 2024]` |
+| **P3** | UpToDate / Standard Textbook | `[Williams Obstetrics 26th Ed]` |
+
+### Citation Rules
+
+- Each article **MUST** have a `## REFERENCES` section at the end
+- Each pathology article (T2) must cite **≥ 3 different guidelines**
+- Each technique article (T3) must cite **≥ 2 guidelines**
+- Each foundation article (T1) must cite **≥ 2 guidelines**
+- Cite **inline** within text, e.g.: `According to ACOG PB #234 (2024), IVF success rate in women > 40 is 15-20%`
+- **DO NOT** accept vague citations: ~~"According to research"~~, ~~"Experts suggest"~~
+
+---
+
+## 3. Anti-Hallucination Rules — No Fabrication
+
+> 🔴 **CRITICAL:** The most important rules for medical content.
+
+| Rule | Description | Violation Example |
+|------|-------------|-------------------|
+| **No fabricated statistics** | Do not invent percentages or statistical figures | ❌ "Success rate 87.3%" (no source) |
+| **No fabricated drugs** | Only list real medications with correct dosages | ❌ Inventing non-existent brand names |
+| **No fabricated guidelines** | Only cite real, existing guidelines | ❌ "ACOG PB #999" (doesn't exist) |
+| **No fabricated studies** | Do not create fake author names/years | ❌ "According to Smith et al. (2024)..." (fabricated) |
+| **When uncertain → state clearly** | Write "Data varies..." or use ranges | ✅ "Rate ~10-20% (varies by study)" |
+
+### Safe Phrases When Exact Data Is Missing
+
+```
+✅ "Rate ranges from X-Y% depending on the study"
+✅ "Based on available literature, estimated at approximately..."
+✅ "Local data is limited; international reports indicate..."
+✅ "Further research needed to confirm"
+❌ "Exactly XX.X%" (when no specific source exists)
+```
+
+---
+
+## 4. Clinical Safety Markers — Medical Markup
+
+Each pathology article (T2) **MUST** include all 4 markers:
+
+| Marker | Purpose | Format |
+|--------|---------|--------|
+| `> 💊 CLINICAL NOTE` | Most important practice consideration | Blockquote |
+| `> ⚠️ RED FLAG` | When urgent intervention / referral is needed | Blockquote |
+| `> 📚 REFERENCE` | Primary guideline for this topic | Blockquote |
+| `> ⚕️ DISCLAIMER` | Specialist consultation required | Blockquote at end |
+
+### Mandatory Disclaimer (end of each article)
+
+```markdown
+> ⚕️ **DISCLAIMER:** This article is for medical reference purposes only.
+> All diagnostic and treatment decisions **require specialist physician guidance**.
+> Do not self-apply protocols without specific clinical evaluation.
+```
+
+---
+
+## 5. Cross-Reference Protocol
+
+### Inter-Article Links
+
+- When mentioning a pathology/technique with its own article → **MUST include article code**
+- Format: `→ See details: [VSN-04] Endometriosis`
+- Section `## CROSS-LINKS` must list all related articles
+
+### ICD-10
+
+- Pathology articles (T2) with ICD-10 codes → **MUST include in header**
+- Format: `> **ICD-10:** N97.0 | **Group:** VSN`
+
+---
+
+## 6. Medical Writing Audit — 8-Dimension Evaluation
+
+When reviewing medical articles, score across 8 dimensions (1-10):
+
+| Dimension | Question | Passing Standard |
+|-----------|----------|-----------------|
+| **1. Evidence Quality** | Are LoE markers on key recommendations? | ≥ 3 LoE markers |
+| **2. Citation Depth** | Are specific guidelines cited? | ≥ 3 different sources |
+| **3. Clinical Accuracy** | Are drug doses, metrics correct? | No medical errors |
+| **4. Structure** | Correct template, all sections present? | 100% sections with content |
+| **5. ICD-10 Compliance** | Correct ICD-10 code present? | Header has ICD-10 |
+| **6. Safety Markers** | Red Flag, Disclaimer present? | 4/4 markers |
+| **7. Cross-References** | Links to related articles? | ≥ 2 cross-links |
+| **8. Word Count** | Meets minimum length? | T1≥1500, T2≥2000, T3≥1500 |
+
+**Scoring:**
+- **72-80:** Excellent — publish ready
+- **56-71:** Good — minor revisions needed
+- **40-55:** Average — needs more evidence
+- **< 40:** Fail — rewrite
+
+---
+
+## 7. Template Enhancement Rules
+
+When creating new articles, inject these additional requirements into the prompt:
+
+### For Pathology Articles (T2)
+
+```
+EVIDENCE-BASED REQUIREMENTS:
+1. Each treatment recommendation includes [LoE: I-V] + source
+2. "DIAGNOSIS" section must have a test table with "Source Guideline" column
+3. "TREATMENT" section must cite specific protocols (ACOG/MoH/ASRM)
+4. End of article has "## REFERENCES" listing ≥ 3 guidelines
+5. End of article has medical DISCLAIMER
+6. DO NOT fabricate statistics — use estimated ranges when exact data is unavailable
+```
+
+### For Technique Articles (T3)
+
+```
+EVIDENCE-BASED REQUIREMENTS:
+1. Each procedure step cites source protocol (ASRM/ESHRE/MoH)
+2. Drug tables must have a "Guideline" column
+3. Success rates must cite source + confidence interval
+4. End of article has "## REFERENCES" ≥ 2 guidelines
+5. End of article has medical DISCLAIMER
+```
+
+---
+
+## Verification Script
+
+Run after article generation to check quality:
+
+```bash
+python3 ~/.gemini/antigravity/skills/medical-research/scripts/evidence_checker.py <output_dir>
+```
+
+Script checks: References, LoE markers, ICD-10, Disclaimer, word count, cross-links.
+
+---
+
+## Quick Diagnostic
+
+| Question | If NO | Action |
+|----------|-------|--------|
+| Has REFERENCES section? | Article lacks sources | Add ≥ 3 guidelines |
+| Has LoE for recommendations? | Unclear reliability | Assign LoE I-V |
+| Has ICD-10 in header? | Missing coding | Add ICD-10 |
+| Has RED FLAG? | Missing clinical warning | Add ⚠️ marker |
+| Has DISCLAIMER? | Legal risk | Add disclaimer at end |
+| Do statistics have sources? | Suspected hallucination | Convert to estimated ranges |

package/skills/medical-research/scripts/evidence_checker.py

@@ -0,0 +1,288 @@
+#!/usr/bin/env python3
+"""
+Medical Research Evidence Checker
+Validates generated medical articles against evidence-based writing standards.
+
+Usage:
+  python3 evidence_checker.py <output_dir>           # Check all .md files
+  python3 evidence_checker.py <output_dir> --strict   # Fail on warnings too
+  python3 evidence_checker.py <file.md>               # Check single file
+"""
+import argparse
+import re
+import sys
+from dataclasses import dataclass, field
+from pathlib import Path
+
+
+# === Thresholds ===
+MIN_WORDS = {"T1": 1500, "T2": 2000, "T3": 1500}
+MIN_REFERENCES = {"T1": 2, "T2": 3, "T3": 2}
+MIN_LOE_MARKERS = {"T1": 1, "T2": 3, "T3": 2}
+
+GUIDELINE_ORGS = [
+    "ACOG", "ASRM", "ESHRE", "WHO", "BYT", "NICE", "RCOG",
+    "Cochrane", "FIGO", "ISUOG", "RANZCOG", "SOGC",
+]
+
+TIER_MAP = {
+    "F-": "T1",
+    "VSN-": "T2", "VSM-": "T2", "NTI-": "T2", "BT-": "T2",
+    "TC-": "T2", "VT-": "T2", "CTC-": "T2", "AD-": "T2",
+    "VU-": "T2", "TK1-": "T2", "TK2-": "T2", "TK3-": "T2",
+    "TNC-": "T2", "SSL-": "T2", "HSS-": "T2", "VNPK-": "T2",
+    "SMTD-": "T2", "SLTT-": "T2", "UPKSS-": "T2", "PLHP-": "T2",
+    "KTHT-": "T3", "PD-": "T3", "PHO-": "T3", "TDP-": "T3",
+    "DT-": "T3",
+}
+
+
+@dataclass
+class CheckResult:
+    file: str
+    tier: str = "T2"
+    word_count: int = 0
+    has_references: bool = False
+    reference_count: int = 0
+    guideline_refs: list = field(default_factory=list)
+    loe_count: int = 0
+    has_icd10: bool = False
+    has_red_flag: bool = False
+    has_disclaimer: bool = False
+    has_cross_links: bool = False
+    errors: list = field(default_factory=list)
+    warnings: list = field(default_factory=list)
+
+    @property
+    def score(self) -> int:
+        s = 0
+        min_words = MIN_WORDS.get(self.tier, 1500)
+        min_refs = MIN_REFERENCES.get(self.tier, 2)
+        min_loe = MIN_LOE_MARKERS.get(self.tier, 1)
+
+        # Word count (0-10)
+        if self.word_count >= min_words:
+            s += 10
+        elif self.word_count >= min_words * 0.7:
+            s += 6
+        elif self.word_count >= min_words * 0.5:
+            s += 3
+
+        # References (0-10)
+        if self.reference_count >= min_refs:
+            s += 10
+        elif self.reference_count >= 1:
+            s += 5
+
+        # Guideline orgs (0-10)
+        if len(self.guideline_refs) >= min_refs:
+            s += 10
+        elif len(self.guideline_refs) >= 1:
+            s += 5
+
+        # LoE markers (0-10)
+        if self.loe_count >= min_loe:
+            s += 10
+        elif self.loe_count >= 1:
+            s += 5
+
+        # ICD-10 (0-10) — only for T2
+        if self.tier == "T2":
+            s += 10 if self.has_icd10 else 0
+        else:
+            s += 10  # N/A for T1/T3
+
+        # Red flag (0-10) — only for T2
+        if self.tier == "T2":
+            s += 10 if self.has_red_flag else 0
+        else:
+            s += 10
+
+        # Disclaimer (0-10)
+        s += 10 if self.has_disclaimer else 0
+
+        # Cross-links (0-10)
+        s += 10 if self.has_cross_links else 0
+
+        return s
+
+    @property
+    def grade(self) -> str:
+        s = self.score
+        if s >= 72:
+            return "✅ XUẤT SẮC"
+        if s >= 56:
+            return "🟡 TỐT"
+        if s >= 40:
+            return "🟠 TRUNG BÌNH"
+        return "❌ FAIL"
+
+
+def detect_tier(filename: str) -> str:
+    for prefix, tier in TIER_MAP.items():
+        if filename.startswith(prefix):
+            return tier
+    return "T2"
+
+
+def check_file(filepath: Path) -> CheckResult:
+    content = filepath.read_text(encoding="utf-8")
+    filename = filepath.stem
+    result = CheckResult(file=str(filepath), tier=detect_tier(filename))
+
+    # Word count
+    result.word_count = len(content.split())
+    min_words = MIN_WORDS.get(result.tier, 1500)
+    if result.word_count < min_words:
+        result.errors.append(f"Word count {result.word_count} < {min_words} minimum")
+
+    # References section
+    ref_patterns = [
+        r"##\s*\d*\.?\s*TÀI LIỆU THAM KHẢO",
+        r"##\s*\d*\.?\s*REFERENCES",
+        r"##\s*\d*\.?\s*THAM KHẢO",
+    ]
+    result.has_references = any(re.search(p, content, re.IGNORECASE) for p in ref_patterns)
+    if not result.has_references:
+        result.errors.append("Missing '## TÀI LIỆU THAM KHẢO' section")
+
+    # Count reference lines (lines starting with - or * or number in ref section)
+    ref_section = ""
+    for p in ref_patterns:
+        match = re.search(p, content, re.IGNORECASE)
+        if match:
+            ref_section = content[match.start():]
+            # Find next ## to limit scope
+            next_section = re.search(r"\n##\s", ref_section[5:])
+            if next_section:
+                ref_section = ref_section[:next_section.start() + 5]
+            break
+    ref_lines = re.findall(r"^\s*[-*\d]+\.?\s+.+", ref_section, re.MULTILINE)
+    result.reference_count = len(ref_lines)
+
+    # Guideline references (anywhere in text)
+    found_orgs = set()
+    for org in GUIDELINE_ORGS:
+        if re.search(rf"\b{org}\b", content, re.IGNORECASE):
+            found_orgs.add(org)
+    result.guideline_refs = sorted(found_orgs)
+
+    min_refs = MIN_REFERENCES.get(result.tier, 2)
+    if len(found_orgs) < min_refs:
+        result.warnings.append(
+            f"Only {len(found_orgs)} guideline org(s) cited, need ≥ {min_refs}"
+        )
+
+    # Level of Evidence markers
+    loe_matches = re.findall(r"\[LoE:\s*[IViv]+", content)
+    result.loe_count = len(loe_matches)
+    min_loe = MIN_LOE_MARKERS.get(result.tier, 1)
+    if result.loe_count < min_loe:
+        result.warnings.append(f"Only {result.loe_count} LoE marker(s), need ≥ {min_loe}")
+
+    # ICD-10 (for T2)
+    icd_pattern = r"ICD-10\s*:\s*[A-Z]\d"
+    result.has_icd10 = bool(re.search(icd_pattern, content))
+    if result.tier == "T2" and not result.has_icd10:
+        result.warnings.append("No ICD-10 code found (expected for pathology article)")
+
+    # Red flag marker
+    red_flag_patterns = [r"CỜ ĐỎ", r"⚠️.*CỜ ĐỎ", r"RED FLAG"]
+    result.has_red_flag = any(re.search(p, content) for p in red_flag_patterns)
+    if result.tier == "T2" and not result.has_red_flag:
+        result.warnings.append("Missing '⚠️ CỜ ĐỎ' section")
+
+    # Disclaimer
+    disclaimer_patterns = [
+        r"DISCLAIMER",
+        r"cần chỉ định của bác sĩ",
+        r"bác sĩ chuyên khoa",
+        r"⚕️",
+    ]
+    result.has_disclaimer = any(re.search(p, content, re.IGNORECASE) for p in disclaimer_patterns)
+    if not result.has_disclaimer:
+        result.errors.append("Missing medical disclaimer")
+
+    # Cross-links
+    cross_patterns = [r"CROSS.?LINK", r"Xem chi tiết", r"→\s*Xem", r"Xem thêm"]
+    result.has_cross_links = any(re.search(p, content, re.IGNORECASE) for p in cross_patterns)
+    if not result.has_cross_links:
+        result.warnings.append("No cross-links to related articles")
+
+    return result
+
+
+def main():
+    parser = argparse.ArgumentParser(description="🏥 Medical Research Evidence Checker")
+    parser.add_argument("path", help="File or directory to check")
+    parser.add_argument("--strict", action="store_true", help="Treat warnings as errors")
+    args = parser.parse_args()
+
+    target = Path(args.path)
+    if target.is_file():
+        files = [target]
+    elif target.is_dir():
+        files = sorted(target.rglob("*.md"))
+    else:
+        print(f"❌ Path not found: {target}")
+        sys.exit(1)
+
+    if not files:
+        print(f"⚠️ No .md files found in {target}")
+        sys.exit(0)
+
+    results = [check_file(f) for f in files]
+
+    # === Summary Table ===
+    print("\n" + "=" * 80)
+    print("🏥 MEDICAL RESEARCH EVIDENCE CHECK")
+    print("=" * 80)
+    print(f"\n📁 Checked: {len(results)} file(s)\n")
+
+    print(f"{'File':<30} {'Tier':<5} {'Words':<7} {'Refs':<5} {'LoE':<5} {'Score':<7} {'Grade'}")
+    print("-" * 80)
+
+    total_errors = 0
+    total_warnings = 0
+
+    for r in results:
+        fname = Path(r.file).stem[:28]
+        print(f"{fname:<30} {r.tier:<5} {r.word_count:<7} {r.reference_count:<5} {r.loe_count:<5} {r.score}/80  {r.grade}")
+        total_errors += len(r.errors)
+        total_warnings += len(r.warnings)
+
+    # === Detail Errors ===
+    print(f"\n{'=' * 80}")
+    print(f"📊 Total: {total_errors} error(s), {total_warnings} warning(s)")
+    print(f"{'=' * 80}")
+
+    if total_errors > 0 or total_warnings > 0:
+        for r in results:
+            if r.errors or r.warnings:
+                print(f"\n📄 {Path(r.file).stem}")
+                for e in r.errors:
+                    print(f"  ❌ {e}")
+                for w in r.warnings:
+                    print(f"  ⚠️  {w}")
+
+    # === Score Summary ===
+    if results:
+        avg_score = sum(r.score for r in results) / len(results)
+        print(f"\n📈 Average Score: {avg_score:.1f}/80")
+
+        excellent = sum(1 for r in results if r.score >= 72)
+        good = sum(1 for r in results if 56 <= r.score < 72)
+        avg = sum(1 for r in results if 40 <= r.score < 56)
+        fail = sum(1 for r in results if r.score < 40)
+        print(f"   ✅ Xuất sắc: {excellent} | 🟡 Tốt: {good} | 🟠 TB: {avg} | ❌ Fail: {fail}")
+
+    # Exit code
+    if args.strict and (total_errors > 0 or total_warnings > 0):
+        sys.exit(1)
+    elif total_errors > 0:
+        sys.exit(1)
+    sys.exit(0)
+
+
+if __name__ == "__main__":
+    main()