ワンクリックで
ワンクリックで
三层内容验证系统,确保内容质量符合平台要求
四层去AI化处理,让内容更自然、有人情味,降低AI痕迹评分
Generate professional technical diagrams (Mermaid, ASCII, flowcharts, architecture diagrams) for AI/ML articles. Supports neural network architectures, data pipelines, algorithm flows, and system architectures.
When the user wants to edit, review, or improve existing marketing copy. Also use when the user mentions 'edit this copy,' 'review my copy,' 'copy feedback,' 'proofread,' 'polish this,' 'make this better,' or 'copy sweep.' This skill provides a systematic approach to editing marketing copy through multiple focused passes.
When the user wants to write, rewrite, or improve marketing copy for any page — including homepage, landing pages, pricing pages, feature pages, about pages, or product pages. Also use when the user says "write copy for," "improve this copy," "rewrite this page," "marketing copy," "headline help," or "CTA copy." For email copy, see email-sequence. For popup copy, see popup-cro.
When the user wants to create or optimize an email sequence, drip campaign, automated email flow, or lifecycle email program. Also use when the user mentions "email sequence," "drip campaign," "nurture sequence," "onboarding emails," "welcome sequence," "re-engagement emails," "email automation," or "lifecycle emails." For in-app onboarding, see onboarding-cro.
| name | quality-scorer |
| description | 5维度质量评分和爆款标题生成,评估内容整体质量 |
| triggers | ["质量评分","内容评分","爆款标题","标题生成","质量评估"] |
5维度质量评分系统 + 8大爆款标题公式,全面评估和优化内容质量。
from src.utils.quality import QualityScorer, TitleGenerator, score_content, generate_titles
# ===== 质量评分 =====
scorer = QualityScorer()
report = scorer.score(
content,
title="文章标题",
fact_report={"confidence": 0.9, "issues": []}
)
print(f"总分: {report.total_score}")
print(f"等级: {report.grade.value}") # excellent/good/needs_improvement
print(f"优势: {report.strengths}")
print(f"劣势: {report.weaknesses}")
# ===== 标题生成 =====
generator = TitleGenerator()
titles = generator.generate(
topic="AI写作技巧",
content=content,
count=5,
platform="weixin"
)
for t in titles:
print(f"[{t.formula_name}] {t.title} (推荐指数: {t.score})")
| 维度 | 权重 | 说明 | 评分标准 |
|---|---|---|---|
| 字数符合度 | 20% | 是否符合3500-4000字标准 | 3500-4000字满分,<3000字扣分 |
| 去AI化程度 | 20% | AI味评分是否低于15% | ≤15%满分,每超1%扣2分 |
| 事实准确性 | 30% | 事实核查报告置信度 | 基于事实核查报告评分 |
| 吸引力指数 | 15% | 标题和内容的吸引力 | 标题+开头+章节结构 |
| 格式规范度 | 15% | Markdown格式、章节结构 | 标题层级、代码块、列表 |
class QualityGrade(Enum):
EXCELLENT = "excellent" # ≥85分 - 优秀
GOOD = "good" # 70-84分 - 良好
NEEDS_IMPROVEMENT = "needs_improvement" # <70分 - 需改进
from src.utils.quality import QualityScorer
scorer = QualityScorer()
report = scorer.score(content, title="AI技术深度解析")
# 详细维度评分
for name, dim in report.dimensions.items():
print(f"{dim.name}: {dim.raw_score}分 × {dim.weight} = {dim.weighted_score}分")
# 输出示例:
# 字数符合度: 95分 × 0.2 = 19.0分
# 去AI化程度: 85分 × 0.2 = 17.0分
# 事实准确性: 80分 × 0.3 = 24.0分
# 吸引力指数: 70分 × 0.15 = 10.5分
# 格式规范度: 90分 × 0.15 = 13.5分
# ─────────────────────────────
# 总分: 84.0分 (good)
| 公式 | 模板 | 示例 |
|---|---|---|
| 数字+利益型 | [数字]+[动词]+[结果] | 7天涨粉3000,我做对了这3件事 |
| 提问好奇型 | [为什么/如何]+[痛点]+[暗示] | 为什么你的文章没人看?试试这个方法 |
| 对比反差型 | [A vs B]+[意想不到的结果] | 月薪3千和3万的新媒体人,差距就在这 |
| 警告紧迫型 | [警告]+[痛点]+[紧迫感] | 别再用AI写文章了!平台开始大规模封号 |
| 故事情感型 | [人物]+[困境]+[转折] | 失业3个月后,我靠写作月入2万 |
| 金句观点型 | [短句]+[情感/价值观] | 写作改变命运,这句话不是鸡汤 |
| 教程承诺型 | [动词]+[对象]+[时间承诺] | 10分钟学会爆款标题,新手也能上手 |
| 悬念省略型 | [陈述]+[暗示/省略] | 研究了100篇10w+,发现一个规律... |
from src.utils.quality import TitleGenerator, TitleFormula
generator = TitleGenerator()
# 生成标题
titles = generator.generate(
topic="RAG技术",
content=content,
count=5,
platform="weixin"
)
for t in titles:
print(f"[{t.formula_name}] {t.title}")
print(f" 模板: {t.template}")
print(f" 推荐指数: {t.score}")
# 分析标题特征
analysis = generator.analyze_title("7天学会AI写作,效率翻倍")
# 输出:
{
"title": "7天学会AI写作,效率翻倍",
"length": 13,
"matched_formulas": [
{"formula": "number_benefit", "name": "数字+利益型"}
],
"score": 90
}
# 获取特定公式示例
examples = generator.get_formula_examples(TitleFormula.NUMBER_BENEFIT)
# {"name": "数字+利益型", "template": "...", "examples": [...]}
# 获取所有公式
all_formulas = generator.get_formula_examples()
| 平台 | 最小长度 | 最大长度 |
|---|---|---|
| 微信公众号 | 5 | 64 |
| 小红书 | 8 | 20 |
| 知乎 | 5 | 50 |
| 5 | 70 |
# 生成符合平台限制的标题
titles = generator.generate(
topic="AI写作",
platform="xiaohongshu" # 限制8-20字
)
from src.utils.deai import DeAIAgent
from src.utils.validation import ValidationSystem
from src.utils.quality import QualityScorer, TitleGenerator
# 1. 原始内容
content = open("article.md").read()
# 2. 去AI化处理
deai_agent = DeAIAgent()
content, deai_report = deai_agent.humanize(content, platform="weixin")
print(f"去AI化: {deai_report.original_ai_score}% → {deai_report.final_ai_score}%")
# 3. 内容验证
validator = ValidationSystem()
passed, validation_results = validator.validate_all(content, platform="weixin")
print(f"验证: {'通过' if passed else '未通过'}")
# 4. 质量评分
scorer = QualityScorer()
quality_report = scorer.score(content, title="文章标题")
print(f"质量评分: {quality_report.total_score}分 ({quality_report.grade.value})")
# 5. 生成标题建议
title_gen = TitleGenerator()
titles = title_gen.generate(topic="主题", content=content, count=5)
print("\n推荐标题:")
for i, t in enumerate(titles, 1):
print(f"{i}. [{t.formula_name}] {t.title}")
# 6. 判断是否可以发布
if passed and quality_report.is_good:
print("\n✅ 内容已达到发布标准")
else:
print("\n❌ 内容需要改进:")
print(f" - 验证问题: {validation_results['layer1'].issues}")
print(f" - 质量建议: {quality_report.suggestions}")
# 评分规则
if 3500 <= word_count <= 4000:
raw_score = 100 # 满分
elif word_count < 3000:
raw_score = (word_count / 3000) * 80 # 按比例扣分
elif word_count > 8000:
excess = word_count - 8000
raw_score = 100 - min(20, excess / 100) # 每100字扣1分,最多扣20分
else:
raw_score = 90 # 3500以下但3000以上
# AI味评分越低越好
if ai_score <= 15:
raw_score = 100 # 目标达成
else:
raw_score = max(0, 100 - (ai_score - 15) * 2) # 每超1%扣2分
# 基于事实核查报告
if fact_report:
confidence = fact_report.get("confidence", 0.8)
issues = fact_report.get("issues", [])
raw_score = confidence * 100 - len(issues) * 5
else:
# 无报告时使用启发式评估
raw_score = heuristic_accuracy_check(content) # 基础80分
score = 70 # 基础分
# 标题检查
if 10 <= len(title) <= 30: score += 10
if has_number(title): score += 5
if has_question(title): score += 5
# 开头检查
if has_story_hook(first_para): score += 5
if has_question_hook(first_para): score += 5
# 章节结构
if subheadings >= 5: score += 5
score = 80 # 基础分
# 标题层级
if h2_count >= 5: score += 5
# 代码块
if has_code_blocks: score += 5
# 列表结构
if list_items >= 3: score += 3
# 段落过长
if long_paragraphs > 30%: score -= 10
# 空行过多
if has_excess_blank_lines: score -= 5
custom_weights 参数自定义权重Created: 2026-02-14
Version: 1.0
Dependencies: src/utils/quality/, src/utils/deai/