marketingjuliancongdanh79-pixel/skill-generator
skill-creator-ultra
VerifiedTạo AI Skill mới từ ý tưởng hoặc quy trình công việc. Kết hợp phỏng vấn thông minh, pattern detection, và quantitative evaluation để tạo skill chất lượng production. Dùng khi user nói "tạo skill", "biến quy trình thành skill", "make a new skill", "tự động hóa công việc này", hoặc mô tả quy trình lặp lại mà họ muốn AI tự làm. Cũng dùng khi user muốn cải thiện, test, hoặc optimize skill hiện có.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 14.4 KB |
| agents/analyzer.md | markdown | 10.1 KB |
| agents/comparator.md | markdown | 7.1 KB |
| agents/grader.md | markdown | 8.8 KB |
| assets/eval_review.html | other | 6.9 KB |
| examples/example_anthropic_frontend.md | markdown | 4.9 KB |
| examples/example_anthropic_pdf.md | markdown | 4.5 KB |
| examples/example_api_docs.md | markdown | 5.3 KB |
| examples/example_db_migration.md | markdown | 8.8 KB |
| examples/example_git_commit.md | markdown | 4.1 KB |
| scripts/ci_eval.py | script | 6.4 KB |
| scripts/package_skill.py | script | 4.9 KB |
| scripts/simulate_skill.py | script | 14.2 KB |
| scripts/skill_audit.py | script | 22.7 KB |
| scripts/skill_compare.py | script | 9.3 KB |
| scripts/skill_export.py | script | 9.6 KB |
| scripts/skill_scaffold.py | script | 11.7 KB |
| scripts/skill_stats.py | script | 11.6 KB |
| scripts/validate_skill.py | script | 15.0 KB |
scripts/skill_compare.py
View on GitHub#!/usr/bin/env python3
"""
skill_compare.py — So sánh 2 phiên bản Skill.md
Sử dụng:
python skill_compare.py /path/to/old/SKILL.md /path/to/new/SKILL.md
python skill_compare.py /path/to/old-skill/ /path/to/new-skill/
python skill_compare.py /path/to/old/ /path/to/new/ --json
Hiển thị:
- Diff tổng quan (dòng thêm/xóa/thay đổi)
- So sánh sections (thêm/xóa/thay đổi kích thước)
- So sánh metrics (steps, examples, constraints)
- Đánh giá: cải thiện hay tệ hơn?
Developed by Thân Công Hải — Skill Generator v3.2 Expert
"""
import sys
import os
import re
import json
class Colors:
PASS = '\033[92m'
FAIL = '\033[91m'
WARN = '\033[93m'
INFO = '\033[96m'
BOLD = '\033[1m'
DIM = '\033[2m'
RESET = '\033[0m'
@staticmethod
def enable_windows():
if os.name == 'nt':
try:
import ctypes
kernel32 = ctypes.windll.kernel32
kernel32.SetConsoleMode(kernel32.GetStdHandle(-11), 7)
except Exception:
pass
Colors.enable_windows()
def parse_frontmatter(content):
match = re.match(r'^---\s*\n(.*?)\n---\s*\n', content, re.DOTALL)
if not match:
return {}, content
fm_text = match.group(1)
body = content[match.end():]
frontmatter = {}
current_key = None
current_value = []
for line in fm_text.split('\n'):
key_match = re.match(r'^(\w+):\s*(.*)', line)
if key_match:
if current_key:
frontmatter[current_key] = '\n'.join(current_value).strip()
current_key = key_match.group(1)
current_value = [key_match.group(2)] if key_match.group(2) and key_match.group(2) != '|' else []
elif current_key and line.startswith(' '):
current_value.append(line.strip())
if current_key:
frontmatter[current_key] = '\n'.join(current_value).strip()
return frontmatter, body
def get_metrics(content, body):
"""Trích xuất metrics từ skill."""
return {
'lines': content.count('\n') + 1,
'words': len(content.split()),
'chars': len(content),
'h1_sections': len(re.findall(r'^#\s+', body, re.MULTILINE)),
'h2_sections': len(re.findall(r'^##\s+', body, re.MULTILINE)),
'steps': len(re.findall(r'^\d+\.', body, re.MULTILINE)),
'examples': len(re.findall(r'^##\s+.*[Vv]í dụ|^##\s+.*[Ee]xample|^###\s+.*[Vv]í dụ', body, re.MULTILINE)),
'constraints': len(re.findall(r'(KHÔNG ĐƯỢC|LUÔN LUÔN|🚫|✅)', body)),
'code_blocks': len(re.findall(r'```', body)) // 2,
'tables': len(re.findall(r'^\|.*\|$', body, re.MULTILINE)),
'error_handling': len(re.findall(r'(nếu lỗi|fallback|retry|edge case|VERIFY)', body, re.IGNORECASE)),
}
def get_sections(body):
"""Lấy danh sách sections và kích thước."""
sections = {}
current = None
lines = 0
for line in body.split('\n'):
h1 = re.match(r'^#\s+(.+)', line)
if h1:
if current:
sections[current] = lines
current = h1.group(1).strip()
lines = 0
else:
lines += 1
if current:
sections[current] = lines
return sections
def format_delta(old_val, new_val, higher_is_better=True):
"""Format thay đổi với màu sắc."""
delta = new_val - old_val
if delta == 0:
return f"{Colors.DIM}= (không đổi){Colors.RESET}"
if higher_is_better:
color = Colors.PASS if delta > 0 else Colors.FAIL
else:
color = Colors.FAIL if delta > 0 else Colors.PASS
sign = '+' if delta > 0 else ''
pct = round(delta / old_val * 100) if old_val > 0 else 0
return f"{color}{sign}{delta} ({sign}{pct}%){Colors.RESET}"
def compare_skills(path_old, path_new, json_output=False):
"""So sánh 2 skill."""
# Resolve paths
skill_old = os.path.join(path_old, 'SKILL.md') if os.path.isdir(path_old) else path_old
skill_new = os.path.join(path_new, 'SKILL.md') if os.path.isdir(path_new) else path_new
for path, label in [(skill_old, 'OLD'), (skill_new, 'NEW')]:
if not os.path.exists(path):
print(f"{Colors.FAIL}❌ Không tìm thấy {label}: {path}{Colors.RESET}")
return False
with open(skill_old, 'r', encoding='utf-8') as f:
content_old = f.read()
with open(skill_new, 'r', encoding='utf-8') as f:
content_new = f.read()
fm_old, body_old = parse_frontmatter(content_old)
fm_new, body_new = parse_frontmatter(content_new)
metrics_old = get_metrics(content_old, body_old)
metrics_new = get_metrics(content_new, body_new)
sections_old = get_sections(body_old)
sections_new = get_sections(body_new)
name_old = fm_old.get('name', 'unknown')
name_new = fm_new.get('name', 'unknown')
# JSON output
if json_output:
output = {
'old': {'name': name_old, 'file': skill_old, 'metrics': metrics_old},
'new': {'name': name_new, 'file': skill_new, 'metrics': metrics_new},
'deltas': {k: metrics_new[k] - metrics_old[k] for k in metrics_old},
'sections': {
'added': [s for s in sections_new if s not in sections_old],
'removed': [s for s in sections_old if s not in sections_new],
'changed': {s: {'old': sections_old.get(s, 0), 'new': sections_new.get(s, 0)}
for s in set(sections_old) & set(sections_new)
if sections_old.get(s, 0) != sections_new.get(s, 0)},
},
}
print(json.dumps(output, ensure_ascii=False, indent=2))
return True
# Pretty output
print()
print(f"{Colors.BOLD}{'=' * 60}{Colors.RESET}")
print(f"{Colors.BOLD} 🔄 SKILL COMPARE — So sánh 2 phiên bản{Colors.RESET}")
print(f"{Colors.BOLD}{'=' * 60}{Colors.RESET}")
print(f" {Colors.DIM}OLD: {skill_old}{Colors.RESET}")
print(f" {Colors.INFO}NEW: {skill_new}{Colors.RESET}")
print()
# Metrics comparison
metric_labels = {
'lines': ('📏 Dòng', True),
'words': ('📝 Từ', True),
'steps': ('📋 Steps', True),
'examples': ('🎯 Examples', True),
'constraints': ('⚠️ Constraints', True),
'code_blocks': ('💻 Code blocks', True),
'error_handling': ('🛡️ Error handling', True),
}
print(f" {Colors.BOLD}📊 So sánh metrics:{Colors.RESET}")
print(f" {'Metric':<22} {'OLD':>8} {'NEW':>8} {'Thay đổi'}")
print(f" {'─' * 55}")
improvements = 0
regressions = 0
for key, (label, higher_better) in metric_labels.items():
old_v = metrics_old[key]
new_v = metrics_new[key]
delta_str = format_delta(old_v, new_v, higher_better)
print(f" {label:<22} {old_v:>8} {new_v:>8} {delta_str}")
delta = new_v - old_v
if (higher_better and delta > 0) or (not higher_better and delta < 0):
improvements += 1
elif delta != 0:
regressions += 1
print()
# Sections comparison
added = [s for s in sections_new if s not in sections_old]
removed = [s for s in sections_old if s not in sections_new]
if added or removed:
print(f" {Colors.BOLD}📋 Sections thay đổi:{Colors.RESET}")
for s in added:
print(f" {Colors.PASS}+ {s} ({sections_new[s]} dòng){Colors.RESET}")
for s in removed:
print(f" {Colors.FAIL}- {s} ({sections_old[s]} dòng){Colors.RESET}")
print()
# Changed sections
changed = [(s, sections_old[s], sections_new[s])
for s in set(sections_old) & set(sections_new)
if sections_old[s] != sections_new[s]]
if changed:
print(f" {Colors.BOLD}📐 Sections thay đổi kích thước:{Colors.RESET}")
for name, old_lines, new_lines in sorted(changed, key=lambda x: x[2] - x[1], reverse=True):
delta = new_lines - old_lines
color = Colors.PASS if delta > 0 else Colors.FAIL
sign = '+' if delta > 0 else ''
print(f" {name[:30]:<30} {old_lines:>5} → {new_lines:>5} ({color}{sign}{delta}{Colors.RESET})")
print()
# Overall verdict
print(f"{Colors.BOLD}{'=' * 60}{Colors.RESET}")
if improvements > regressions:
print(f" {Colors.PASS}🎉 VERDICT: Phiên bản mới TỐT HƠN ({improvements} cải thiện, {regressions} giảm){Colors.RESET}")
elif regressions > improvements:
print(f" {Colors.FAIL}⚠️ VERDICT: Phiên bản mới CÓ VẤN ĐỀ ({regressions} giảm, {improvements} cải thiện){Colors.RESET}")
else:
print(f" {Colors.INFO}ℹ️ VERDICT: Hai phiên bản TƯƠNG ĐƯƠNG{Colors.RESET}")
print(f"{Colors.BOLD}{'=' * 60}{Colors.RESET}")
print()
return True
if __name__ == '__main__':
if len(sys.argv) < 3:
print("Sử dụng: python skill_compare.py <old-path> <new-path>")
print("Options:")
print(" --json Output dạng JSON")
print()
print("Ví dụ:")
print(" python skill_compare.py ./my-skill-v1/ ./my-skill-v2/")
print(" python skill_compare.py old-SKILL.md new-SKILL.md")
sys.exit(1)
target_old = sys.argv[1]
target_new = sys.argv[2]
json_mode = '--json' in sys.argv
compare_skills(target_old, target_new, json_output=json_mode)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |