--- name: ClawHub Skill Growth Engine description: AI-powered ClawHub skill growth optimizer v2.0 — analyzes reviews, tracks 2026 trending topics (AI Agent, MCP, video SEO), rewrites titles/descriptions for maximum downloads and stars. Supports video thumbnail prompts, cross-platform social syndication, and growth metrics tracking. Scope: advisory output only — analysis reports, rewrite suggestions, thumbnail prompts and promotion copy. The skill never edits skill metadata, never publishes anything, and never calls external services on the user’s behalf; every action is performed manually by the user after confirmation. Triggers: clawhub optimization, skill growth, SEO, stars, downloads, review analysis, trending keywords, skill improvement, GitHub stars strategy, video SEO, social media integration, thumbnail generation. version: "2.1.3" allowed-tools: [] capabilities: - educational-reference - advisory-only - requires-human-review - code-examples-reference --- # ClawHub Skill Growth Engine / ClawHub技能热度增长引擎 > **⚠️ SECURITY NOTICE / 安全声明** > - **Type:** Educational reference / analytical framework ONLY > - **Code blocks in this document are illustrative teaching examples** — they show the logic and > structure of an approach; nothing here is executed, and no scripts, binaries, or installers > are bundled with this skill. > - **No persistent storage, background execution, or credential collection.** To be precise about network access: this skill itself makes no network calls. The source endpoints and Python snippets shown later are illustrative material — if a user chooses to run them in their own environment, that traffic originates from their machine and goes to third-party platforms, not from this skill. > - **All outputs are for reference only and require human review before real-world application** > - **This skill does NOT provide financial, legal, or insurance advice** > - **Users must exercise their own judgment and consult qualified professionals** > > **⚠️ 数据安全警告** > - 本技能仅提供ClawHub技能优化策略的参考框架，**不执行任何代码或脚本** > - 文中的Python代码为**教学参考示例**，展示逻辑概念，不会自动执行 > - 文中的API调用（如Google Trends、微博热搜、GitHub Trending）为**外部服务引用**，本技能自身不发起这些请求。用户如实际调用需自行评估数据隐私风险：查询关键词、IP地址、时间戳等信息将被发送至第三方平台 > - 不收集、不存储用户的任何平台数据、技能代码或个人信息 > - 热度分析和SEO建议基于公开信息，实际效果因平台算法变化而异 > - 不保证任何优化策略能带来具体的下载量或Star数增长 > - **用户确认要求**：所有涉及修改技能元数据或发布社交媒体内容的建议，均需用户自行手动操作并确认 - **预览后再落地**：本技能产出的优化报告、改写文案、标签清单在写入任何文件之前，必须先输出预览（条目数、涉及字段、是否含未核实数据），经用户明确确认后再保存；未经预览与确认不落盘 - **不代为修改文件**：不修改任何 SKILL.md、README 或元数据文件；本技能只输出建议与素材文本，编辑动作由用户完成 - **不代为发布与发帖**：不在任何平台发帖、不代为提交技能更新 - **不保证效果**：不承诺任何下载量、Star 数或排名结果；所有预期影响均为经验性判断，实际效果因平台算法而异 > **English:** AI-powered growth engine for ClawHub skills — analyze user reviews, track global trending topics, and rewrite your skill metadata (title, description, tags) to maximize downloads and GitHub-style stars. > > **中文:** ClawHub技能热度增长引擎——分析用户评论、追踪全网热点、优化技能标题与描述，一站式提升下载量与Star数。 ## Trigger Keywords / 触发关键词 **⚠️ 精确触发规则**：仅当用户明确提到ClawHub技能增长/优化需求时激活。日常对话中提及"SEO"、"下载量"、"stars"、"热度"等通用词汇时**不会自动触发**，必须与**ClawHub技能优化**直接关联。 **用户确认规则**：匹配以下关键词时，需先向用户确认后再进入优化模式： - "您需要ClawHub技能优化建议吗？" - 仅在用户明确确认后，才提供具体分析和建议 激活关键词（需用户确认后生效）： - ClawHub 优化 / clawhub 技能增长 / 技能热度提升 - 技能下载量提升 / 技能曝光优化 / 技能SEO - 评论分析 / 用户反馈分析 / review 分析 - 热点追踪 / 热搜分析 / 趋势挖掘 - 标题优化 / description 优化 / 关键词优化 - skill 改进 / 技能改进 / 提升关注度 ## Section 0: Latest ClawHub Platform Updates (2026-10-12) | 更新日期 | 平台/趋势 | 对技能优化的影响 | 推荐动作 | |---------|-----------|----------------|---------| | 2026-09 | 技能数量持续增长，同类竞争加剧 | 仅靠关键词堆砌难以获得持续曝光 | 内容深度 + 差异化定位优先于词藻 | | 2026-08 | 平台对技能内容与声明一致性审查趋严 | 声明与正文矛盾会被标记待审 | 发布前核对能力声明与正文是否一致 | | 2026-07 | 中文技能检索权重继续提升 | 中文 description 前 30 字含核心关键词更有效 | 中文关键词前置，避免先写英文长句 | | 2026-06 | 长上下文与文档处理类需求稳定 | 长文档类技能更看重结构化输出示例 | 补充输入输出示例与边界说明 | | 2026-05 | AI Agent 成为 ClawHub 下载量最大品类 | 标题含 AI Agent 可提升曝光 | 含 AI Agent 关键词的技能优先更新 | | 2026-05 | MCP (Model Context Protocol) 生态活跃 | MCP 相关技能搜索量显著增长 | 技能 description 加入 MCP 关键词 | | 2026-05 | OpenClaw 版本迭代，技能生态繁荣 | 竞争加剧 | 差异化描述 + 社媒推广成刚需 | | 2026-04 | 视频内容（小红书/抖音/B站）成推广主战场 | 带视频演示的技能互动更高 | 技能增加视频脚本 + 缩略图提示词 | | 2026-04 | 中国市场（DeepSeek/通义/Kimi）热度高 | 中文技能 SEO 权重提升 | 中文 description 前 30 字含核心关键词 | | 2026-03 | Long Context RAG（100K–2M token）成热点 | 长文档分析类技能需求爆发 | 精算/财报/法律类技能 description 强化 RAG | | 2026-03 | 银行保险监管合规类技能需求稳定增长 | 合规类技能长尾流量稳定 | 合规类技能持续更新法规版本 | | 新增·2026-09 | 安全扫描对「触发词过宽」的判定更常见 | 泛化触发词会被要求收窄 | 触发词收窄到具体任务并附不触发清单 | | 新增·2026-09 | 反例文本会被当作正文一并扫描 | 为说明「不要这样做」写出的命令会被判为真实指令 | 反例不写出可直接复制的完整命令 | | 新增·2026-09 | 「只改版本号」型更新被明确识别 | 无实质内容增长的更新难通过 | 每次更新至少补一组示例与一个表格维度 | | 新增·2026-10 | 表格结构完整性影响可读性评分 | 缺列的表格行会被判为格式缺陷 | 提交前检查每个表块各行分隔符数量一致 | > 说明：以上为平台与生态层面的观察与经验总结，具体平台规则与算法**以官方最新发布为准**。 ### Section 0b: 优化前的一致性自检（发布前必做） | 自检项 | 检查方式 | 不一致的后果 | 修复优先级 | |-------|---------|-------------|----------| | 能力声明 vs 正文 | 声明无代码时正文是否真的没有代码块 | 被判定声明与内容矛盾，进入待审 | 最高 | | 触发词范围 | 触发词是否只覆盖本技能职责范围 | 触发过宽会被要求收窄 | 高 | | 版本号一致 | frontmatter 版本与线上最新版本是否衔接 | 发布被拒或版本回退 | 最高 | | 篇幅完整度 | 是否只改版本号就提交 | 被判定无实质更新，审核不通过 | 最高 | | 链接有效性 | 正文引用的文件路径是否真实存在 | 断链，影响使用体验 | 中 | | 术语时效 | 机构名、法规名是否为现行表述 | 专业性受损，可能触发合规审查 | 中 | | 表格结构 | 每个表块各行分隔符数量是否与表头一致 | 缺列行渲染错乱 | 中 | | 反例写法 | 安全类反例是否写出可直接复制的完整命令 | 被扫描器当作正文指令 | 高 | | 反例与禁令自洽 | 示例文本是否与本文自身的禁止清单冲突 | 使用者照抄示例绕过筛查 | 高 | --- ## Core Capabilities / 核心能力 ### 1. User Review & Feedback Analysis Engine / 用户评论与反馈分析引擎 Analyze user reviews, feedback, and usage data to extract actionable improvement suggestions. **Analysis Dimensions:** | Dimension | What It Detects | Action | 误判提示 | |-----------|----------------|--------|----------| | **Feature Requests** | Users asking for capabilities the skill lacks | Add missing modules to SKILL.md | 个别请求不等于普遍需求，看出现次数 | | **Pain Points** | Frustration or confusion signals in reviews | Simplify instructions, add examples | 区分「没看懂」与「不好用」 | | **Competitor Mentions** | Users comparing to other tools | Add differentiation points | 只提名字不算对比，看是否指出差异 | | **Localization Gaps** | Non-Chinese users struggling (language barriers) | Add English README + bilingual docs | 英文优先是本地化取舍，不是安全能力 | | **Pricing/Access Issues** | Access friction, download barriers | Optimize onboarding flow | 部分问题在平台侧，技能改不了 | | **Emotional Signals** | Excitement/disappointment in wording | Prioritize highly-praised features | 情绪强度不等于需求广度 | | **Video/Social Requests** | 能不能出个视频教程 / 想要小红书推广 | Add video script + thumbnail prompts | 先确认内容已定稿再推广 | | **MCP/Integration Gaps** | 能否对接XX工具 / 支持MCP吗 | Add MCP integration section | 确认是否真有对接能力再写 | | **Long Context Needs** | 处理长文档时卡住 / 支持XX万字吗 | Add context window optimization | 不承诺超出实际能力的长度 | | **Review Gaps** | 评论数过少或集中在单一渠道 | 扩大反馈收集面 | 样本过小时结论不可靠 | **评论分析示例（三条真实形态的评论）：** | 评论原文（已脱敏） | 判定维度 | 动作 | 优先级 | |------------------|---------|------|--------| | 功能挺全的，就是刚开始不知道怎么用 | 上手门槛（Pain Point） | 在开头补一段最短上手路径 | 高 | | 能不能支持导出成表格 | Feature Request | 评估后决定是否加入能力清单 | 中 | | 和另外那个 XX 技能比，这个好在 XX | Competitor Mention | 把该差异点写进 description | 中 | **一条容易被忽略的规则：** 评论分析结论要标注样本量。三条评论里两条提到同一个问题，与三百条评论里二十条提到同一问题，结论强度完全不同。不标注样本量的分析容易把个别意见当成普遍需求。 **Review Analysis Code:** ```python import re from collections import Counter def analyze_reviews(reviews: list[str]) -> dict: """ Analyze user reviews and extract actionable insights. reviews: list of review texts Returns: dict with categorized insights """ positive_keywords = [ "great", "amazing", "love", "perfect", "useful", "helpful", "强大", "好用", "实用", "完美", "赞", "棒", "优秀" ] negative_keywords = [ "confusing", "broken", "bug", "missing", "wrong", "复杂", "难用", "没用", "问题", "错误", "缺东西" ] feature_request_patterns = [ r"wish.*could", r"would be nice", r"should have", r"建议", r"希望有", r"能否加入", r"期待" ] results = { "positive_signals": [], "negative_signals": [], "feature_requests": [], "keywords": Counter() } for review in reviews: text_lower = review.lower() # Detect sentiment signals for kw in positive_keywords: if kw in text_lower: results["positive_signals"].append(review) break for kw in negative_keywords: if kw in text_lower: results["negative_signals"].append(review) break # Detect feature requests for pattern in feature_request_patterns: if re.search(pattern, text_lower): results["feature_requests"].append(review) break # Word frequency (simple tokenizer) words = re.findall(r'\b\w{3,}\b', text_lower) results["keywords"].update(w for w in words if len(w) > 3) return results ``` **Output Format:** ```markdown ## Review Analysis Report ### 🔥 Top 5 Praised Features 1. [Feature] — mentioned X times 2. ... ### 💡 Top 5 Feature Requests 1. [Request] — mentioned X times → Priority: HIGH/MEDIUM/LOW 2. ... ### ⚠️ Top 5 Pain Points 1. [Pain point] — urgency: CRITICAL/HIGH/MEDIUM 2. ... ### 📊 Keyword Frequency (Top 20) | Keyword | Count | Sentiment | |---------|-------|-----------| | XXX | 123 | Positive | | ... | ... | ... | ### 🎯 Recommended Actions 1. **[HIGH]** Add [missing feature] to address [request] 2. **[MEDIUM]** Simplify [confusing part] based on [pain point] 3. **[LOW]** Add [example/tutorial] to reduce confusion ``` --- ### 2. Trending Topic Tracker / 全网热点追踪引擎 Monitor trending topics across 40+ platforms to identify hot keywords that can boost skill visibility. **Supported Data Sources:** | Source | API/Endpoint | Data | Use Case | 调用时的隐私提示 | |--------|-------------|------|----------|----------------| | Weibo Hot Search | `uapis.cn` | Real-time热搜 | China trending | 请求含查询词与来源 IP | | Zhihu Hot | `uapis.cn` | 知乎热榜 | Tech discussions | 同上 | | Bilibili Trending | `uapis.cn` | B站热搜 | Youth/tech audience | 同上 | | GitHub Trending | `github.com/trending` | GitHub热门 | Developer tools | 未鉴权请求受速率限制 | | WeChat Index | Tencent API | 微信指数 | China ecosystem | 通常需要平台侧授权 | | Baidu Index | `index.baidu.com` | 百度指数 | Search trends | 请求含查询词与来源 IP | | Google Trends | `trends.google.com` | Global trends | International | 跨境传输，需评估合规性 | | Product Hunt | `producthunt.com` | PH热榜 | Global startup tools | 跨境传输，需评估合规性 | > **外部调用提示（E1）**：上表中的地址仅用于说明「可以从哪些方向取趋势数据」，本技能不会替用户发起这些请求。若你在自己的环境中运行后面的示例代码，查询词、来源 IP 与时间戳会发送到对应第三方平台，请事先确认所在组织的数据出境与第三方调用规定允许这样做。 **示例（仅示意取数逻辑，不代用户执行）：** | 步骤 | 做什么 | 注意 | |-----|-------|------| | 1 | 确定要追踪的关键词集合 | 关键词本身也可能含敏感信息 | | 2 | 选定 2–3 个来源交叉验证 | 单一来源的趋势容易误判 | | 3 | 记录取数时间与口径 | 便于下次对比，也便于复盘 | | 4 | 只用聚合结果，不保存原始明细 | 减少不必要的数据留存 | | 5 | 趋势词与技能实际能力对照后再写入 | 无对应能力的热点词会带来负面评价 | **Trending Data Fetching Code:** ```python import requests import json def fetch_weibo_trending(limit: int = 20) -> list[dict]: """Fetch real-time Weibo hot search topics.""" url = "https://uapis.cn/api/hotboard" params = {"type": "weibo", "limit": limit} try: resp = requests.get(url, params=params, timeout=10) data = resp.json() return [ {"rank": i+1, "title": item.get("title", ""), "hot": item.get("hot", ""), "url": item.get("url", "")} for i, item in enumerate(data.get("data", [])[:limit]) ] except Exception as e: return [{"error": str(e)}] def fetch_github_trending(lang: str = "python", limit: int = 10) -> list[dict]: """Fetch GitHub trending repositories.""" url = f"https://api.github.com/search/repositories" params = { "q": f"language:{lang}+created:>2025-01-01", "sort": "stars", "order": "desc", "per_page": limit } headers = {"Accept": "application/vnd.github.v3+json"} try: resp = requests.get(url, params=params, headers=headers, timeout=10) data = resp.json() return [ {"name": item["name"], "stars": item["stargazers_count"], "description": item["description"], "url": item["html_url"]} for item in data.get("items", [])[:limit] ] except Exception as e: return [{"error": str(e)}] def google_trends_suggestions(keyword: str) -> list[str]: """Get related queries from Google Trends.""" # Using pytrends library from pytrends.request import TrendReq pytrends = TrendReq(hl='en-US', tz=360) pytrends.build_payload([keyword], cat=0, timeframe='today 3-m', geo='') related = pytrends.related_queries() suggestions = [] for kw_list in related.values(): for item in kw_list.get('top', []) if kw_list else []: suggestions.append(item['query']) return suggestions[:10] ``` **Trending Keyword Mapping for Skills:** ```python TRENDING_MAPPING = { # 2026 AI/LLM trends → skill keyword suggestions "DeepSeek": ["DeepSeek", "LLM", "AI agent", "Chinese AI", "open-source LLM"], "AI Agent": ["AI Agent", "workflow automation", "autonomous AI", "MCP"], "Claude": ["Claude", "Anthropic", "context window", "reasoning", "long context"], "MCP": ["MCP", "Model Context Protocol", "tool integration", "AI agent tools"], "Stock Market": ["A-share", "quantitative trading", "technical analysis", "缠论", "量化"], "Insurance": ["insurance tech", "insurtech", "risk management", "C-ROSS", "NFRA"], "Content Creation": ["AI video", "short video", "social media AI", "video SEO", "thumbnail"], "Productivity": ["workflow automation", "efficiency", "productivity tools", "RAG", "long context"], "Compliance": ["bank compliance", "NFRA", "Basel III", "AML", "PIPL", "regulatory"], "Actuarial": ["actuarial pricing", "C-ROSS II", "IFRS 17", "HKFRS 17", "life table 2025"], } def map_trending_to_skill(trending_topics: list[str], skill_tags: list[str]) -> list[dict]: """Map trending topics to skill tags for SEO boost.""" suggestions = [] for topic in trending_topics: for trend, keywords in TRENDING_MAPPING.items(): if trend.lower() in topic.lower(): for kw in keywords: if kw not in skill_tags: suggestions.append({ "trend": topic, "suggested_tag": kw, "priority": "HIGH" if len(suggestions) < 5 else "MEDIUM" }) return suggestions[:10] ``` --- ### 3. SEO Title & Description Optimizer / SEO标题与描述优化器 Rewrite skill titles and descriptions using proven SEO frameworks to maximize search visibility and click-through rate. **Title Optimization Framework:** | Principle | English Example | Chinese Example | 常见误用 | |-----------|----------------|----------------|----------| | **Front-load value** | "AI Insurance Claims Analyzer" | 保险理赔AI专家 | 把品牌名放在最前 | | **Include keyword** | "Stock Technical Analysis" | A股技术分析 | 堆砌不相关热词 | | **Show outcome** | "Increase Downloads 10x" | 提升下载量 | 承诺无法验证的倍数 | | **Use numbers** | "5-Step Process" | 7大核心能力 | 数字与实际内容不符 | | **Be specific** | "China Insurance C-ROSS Actuarial" | 偿二代精算定价 | 生僻到无人搜索 | | **Evoke emotion** | "Stop Losing Money" | 告别选号盲目 | 制造焦虑或夸大 | | **State the boundary** | "Reference Framework, Not Advice" | 参考框架，非投资建议 | 只用情绪词替代定位 | **Description Structure (AIDA Framework):** ``` A - Attention: [Bold hook: "The ONLY ClawHub skill that..."] I - Interest: [Specific problem + your unique solution] D - Desire: [Concrete results: "Used by 500+ analysts"] A - Action: [Clear CTA: "Install now and..."] ``` **Title Rewrite Examples:** | Original (Chinese) | Optimized (English) | Optimized (Chinese) | Stars Impact | |--------------------|--------------------|--------------------|--------------| | 招投标文书助手 | Enterprise Bid Document AI | 企业招投标文书AI助手 | ⭐⭐⭐ | | 保险反欺诈 | Insurance Anti-Fraud Pro | 保险反欺诈分析专家 | ⭐⭐⭐⭐ | | 缠论技术分析 | Chanlun Technical Analysis Engine | 缠论技术分析引擎 | ⭐⭐⭐⭐ | | 彩票预测 | Lottery Data Analysis & Number Generator | 彩票数据分析选号助手 | ⭐⭐ | | 银行合规审查 | Bank Compliance Review Assistant | 银行合规审查助手 | ⭐⭐⭐⭐ | | 知识库管理 | Financial Knowledge Base Manager | 金融行业知识库管家 | ⭐⭐⭐ | **Tag Optimization:** ```python def optimize_tags(current_tags: list[str], trending_keywords: list[str], competitors: list[str]) -> dict: """ Optimize skill tags for maximum discoverability. """ must_have = ["clawhub", "skill", "ai-agent"] # Always include high_value = ["python", "api", "automation", "analysis", "tool"] trending = [kw for kw in trending_keywords if kw not in current_tags][:5] competitor_tags = [t for t in competitors if t not in current_tags][:3] optimized = must_have + high_value + trending + competitor_tags optimized = list(dict.fromkeys(optimized))[:20] # Dedupe, max 20 return { "current_tags": current_tags, "recommended_tags": optimized, "new_tags_added": [t for t in optimized if t not in current_tags], "tags_removed": [t for t in current_tags if t not in optimized], "seo_score_improvement": f"+{len([t for t in optimized if t not in current_tags]) * 5}%", } ``` --- ### 4. Video Thumbnail & AI Preview Generation / AI视频缩略图生成与预览优化 Generate compelling skill preview visuals and AI video scripts for social media promotion. **Thumbnail Prompt Framework:** | Platform | Thumbnail Style | Prompt Template | |----------|---------------|----------------| | 小红书 | 高对比度大字+真人/产品图 | "Bold Chinese text 'XX', close-up screenshot of [UI], gradient background #hex, 3:4 ratio" | | 抖音 | 强冲击+情绪化画面 | "Explosion effect, bold text 'XX', dramatic lighting, 9:16 vertical, trending color palette" | | B站 | 知识感+人物出镜 | "Clean desk setup, person pointing at screen, code editor visible, warm lighting, 16:9" | | 微信 | 简约商务风 | "Minimal flat design, skill icon centered, subtle gradient, 2:1 ratio, Chinese + English title" | **AI Thumbnail Generation Code:** ```python from openai import OpenAI import json def generate_thumbnail_prompt(skill_name: str, target_platform: str, highlight: str, style: str = "modern") -> str: """Generate optimized thumbnail prompt for skill promotion.""" platform_configs = { "小红书": {"ratio": "3:4", "color": "vibrant coral + white", "text_pos": "top-center"}, "抖音": {"ratio": "9:16", "color": "neon purple + cyan", "text_pos": "center"}, "B站": {"ratio": "16:9", "color": "dark blue + gold", "text_pos": "bottom-left"}, "微信": {"ratio": "2:1", "color": "minimal white + blue", "text_pos": "center-bottom"} } cfg = platform_configs.get(target_platform, platform_configs["小红书"]) return ( f"Professional skill preview thumbnail for '{skill_name}', " f"highlight: {highlight}. Style: {style}. " f"Aspect ratio {cfg['ratio']}, color scheme {cfg['color']}. " f"Large bold text '{skill_name}' at {cfg['text_pos']}. " f"Clean, modern, high contrast, suitable for {target_platform}." ) def generate_video_script(skill_name: str, skill_slug: str, duration_sec: int = 60) -> dict: """ Generate 1-minute video script with 4 acts (15s each). Returns dict with act breakdown and AI image prompts. """ return { "title": f"【技能推荐】{skill_name} — 3分钟上手指南", "duration": f"{duration_sec}秒", "hook": f"XX秒就能搞定的{skill_name}技能，效率提升10倍！", "acts": [ {"act": 1, "seconds": "0-15s", "scene": "Hook — 痛点场景", "narration": f"还在为XX问题头疼？{skill_name}帮你一键解决！", "visual_prompt": f"Stressed person at desk, messy data, red alerts, warm lighting, cinematic"}, {"act": 2, "seconds": "15-30s", "scene": "Demo — 技能展示", "narration": "看，这是它的核心功能，我只需要输入XX，就能得到XX。", "visual_prompt": f"Screen recording UI of {skill_name}, clean interface, smooth animation, cursor clicking"}, {"act": 3, "seconds": "30-45s", "scene": "Result — 效果对比", "narration": "对比一下：原来要XX分钟，现在只要XX秒，效率提升太明显了！", "visual_prompt": f"Split screen: left messy slow process, right clean fast result, dramatic contrast lighting"}, {"act": 4, "seconds": "45-60s", "scene": "CTA — 行动号召", "narration": f"安装命令：npx clawhub install @yourname/{skill_slug}@<版本号>，马上试试！", "visual_prompt": f"Large text overlay '{skill_name}', install command shown, QR code, clean blue gradient"} ] } ``` **Video Script Template (Markdown):** ```markdown ## 🎬 {Skill Name} 视频推广脚本 ({duration}秒) ### Act 1: 钩子 (0-{d1}s) - **画面**: [visual_prompt] - **配音**: {hook_narration} - **字幕**: [大字突出痛点关键词] ### Act 2: 演示 ({d1}-{d2}s) - **画面**: [screen recording showing skill in action] - **配音**: [step-by-step usage walkthrough] - **字幕**: [关键操作步骤] ### Act 3: 效果 ({d2}-{d3}s) - **画面**: Before/After comparison, data visualization - **配音**: [quantified improvement] - **字幕**: [数字: 效率提升XX倍/节省XX时间] ### Act 4: 行动号召 ({d3}-{duration}s) - **画面**: Install command + QR code - **配音**: [enthusiastic CTA] - **字幕**: npx clawhub install @yourname/{slug}@<版本号> **关于固定版本（RP1）：** 上面示例中的安装命令一律带 `<版本号>` 占位，这是一个有意为之的写法——安装固定版本可以确保拿到经过审查的那一版，避免上游变更在你不知情的情况下进入工作环境。请把占位符替换为你要推荐的具体版本号，不要留空，也不要改用不指定版本的写法。同理，命令行工具本身也应使用你所在机构已登记的固定版本。 ``` --- ### 4b. Cross-Platform Social Syndication Strategy / 跨平台社交媒体推广策略 Design multi-platform promotion campaigns for maximum reach. **Platform-Specific SEO Matrix:** | Platform | Title Style | Description Length | Keyword Density | CTA Format | 发布前确认 | |----------|-------------|-------------------|-----------------|------------|----------| | 小红书 | 中文感叹句，含数字 | 300-500字 | 高 | 评论区置顶安装命令 | 内容已定稿再发 | | 抖音 | 悬念式/对比式 | 视频字幕为主 | 中 | 评论区引导 | 字幕与口播一致 | | B站 | 知识干货型 | 800-2000字 | 低 | 简介区链接 | 链接可访问 | | 知乎 | 深度分析型 | 1000-3000字 | 低 | 专栏文章链接 | 声明利益关系 | | 微信公众号 | 商务正式型 | 500-1000字 | 中 | 文末二维码 | 已过内部审核 | **Social Proof Framework:** ```python SOCIAL_PROOF_TYPES = { "download_count": {"format": "🔥 X万次安装", "impact": "HIGH"}, "star_count": {"format": "⭐ X千Star", "impact": "HIGH"}, "user_testimonial": {"format": "用户说：'...'", "impact": "VERY_HIGH"}, "media_mention": {"format": "被XX媒体报道", "impact": "MEDIUM"}, "award_badge": {"format": "🏆 ClawHub热门技能", "impact": "MEDIUM"}, "update_freshness": {"format": "✅ 今日更新", "impact": "HIGH"}, } def build_social_proof_badge(skill_data: dict) -> str: """Build multi-element social proof string.""" badges = [] if skill_data.get("downloads", 0) > 1000: badges.append(f"🔥 {skill_data['downloads']//1000}万+安装") if skill_data.get("stars", 0) > 100: badges.append(f"⭐ {skill_data['stars']//1000}千Star") if skill_data.get("last_updated_days", 99) < 7: badges.append("✅ 近期更新") if skill_data.get("is_top_rated"): badges.append("🏆 热门推荐") return " | ".join(badges) if badges else "" ``` **A/B Testing Framework for Titles:** ```python def generate_title_variants(original: str, skill_domain: str) -> list[dict]: """Generate 3-5 title variants for A/B testing.""" templates = { "insurance": [ f"【保险人必装】{original} — 效率提升10倍", f"保险{original}专家版：XX分钟搞定XX", f"不想加班？{original}让保险工作自动化", ], "bank": [ f"银行人专属{original}，合规效率双提升", f"【合规必备】{original} — 监管合规利器", ], "trading": [ f"{original}：XX个指标一键分析，炒股不迷茫", f"看盘神器！{original}让技术分析零门槛", ], "default": [ f"【AI工具】{original} — 3分钟上手教程", f"{original}：提升效率XX倍的秘密武器", ] } variants = templates.get(skill_domain, templates["default"]) return [{"variant": v, "approach": t} for v, t in zip(variants, [ "数字+感叹词", "领域专属词", "痛点引导" ][:len(variants)])] ``` --- ### 5. GitHub-Style Stars Growth Strategy / GitHub式Star增长策略 Apply proven open-source project growth tactics to ClawHub skills. **Strategy Framework:** | Strategy | Implementation | Expected Impact | 落地前提 | |----------|---------------|----------------|----------| | **README Quality** | First 5 lines = summary. Clear What/Why/How. Screenshots. | ⭐⭐⭐⭐ | 截图为当前版本 | | **Keyword SEO** | Title + first 2 lines contain main keywords. README H1-H3 structure. | ⭐⭐⭐⭐⭐ | 关键词与内容相符 | | **Demo/Preview** | Short video or GIF showing the skill in action | ⭐⭐⭐⭐ | 演示不夸大能力 | | **Cross-posting** | Share on Zhihu, Weibo, Bilibili with skill link | ⭐⭐⭐ | 各平台规则允许 | | **Community Building** | Create WeChat group / QQ group for skill users | ⭐⭐⭐ | 有持续维护能力 | | **Regular Updates** | Version updates with changelog. Updated recently signal. | ⭐⭐⭐⭐ | 每次更新确有实质内容 | | **Comparison Content** | vs [competitor] articles to attract their users | ⭐⭐⭐ | 对比客观、不贬损 | | **Trending Integration** | Tie skill to current hot topics (AI agents, DeepSeek, etc.) | ⭐⭐⭐⭐⭐ | 热点与能力真实相关 | | **Multi-language** | English README = global audience 10x | ⭐⭐⭐⭐⭐ | 英文部分与中文一致 | | **Boundary clarity** | State what the skill does not do | ⭐⭐⭐⭐ | 声明与正文无矛盾 | > **关于「英文在前」的说明（语言与本地化取舍）：** README 模板把英文部分放在前面，是为了照顾跨语言检索与海外读者，这是一项**内容组织上的本地化取舍**，不是安全能力，也不代表本技能优先服务某一语种的使用者。若你的受众主要在国内，完全可以把中文部分前置——本技能不做强制要求。 > > 需要保持一致的是：两种语言描述的是同一个技能、同一组能力。中英文内容出现能力描述不一致，比语种顺序更容易引发问题。 **README Bilingual Template:** ```markdown # [English Title] / [中文标题] > **English Description**: One powerful sentence describing the skill's core value. > Built for [target user]. Solves [specific problem]. ## ✨ Features / Features / 核心功能 - ✅ Feature 1 with specific metric or result - ✅ Feature 2 — [why it matters] - ✅ Feature 3 ## 🚀 Quick Start ```bash # Install npx clawhub install @yourname/your-skill@<版本号> # Use /your-skill [command] ``` ## 📖 Documentation Full docs at [link] or continue reading below. --- > **中文介绍**：一句话描述技能核心价值。针对[目标用户]，解决[具体问题]。 ## 🎯 核心功能 - ✅ 功能1 — [具体效果/数据] - ✅ 功能2 — [为什么有用] - ✅ 功能3 ## ⚡ 快速上手 1. 安装：`npx clawhub install @yourname/your-skill@<版本号>` 2. 使用：`/your-skill [命令]` 3. 查看文档见下方 ## 📚 详细文档 [详细内容...] ``` --- > **⚠️ 操作提示**：以下策略为**人工操作建议**，需用户自行判断并手动执行。本技能**不会自动修改**任何技能元数据、不会自动在社交媒体发帖、不会调用任何外部服务。每项操作前请自行评估合规性和平台规则。 **2026 年补充要点** | Strategy | 2026 Implementation | Expected Impact | 前提 | |----------|-------------------|----------------|------| | **Demo Video** | Short video or GIF showing the skill in action (小红书/抖音/B站) | ⭐⭐⭐⭐⭐ | 内容已定稿 | | **Video Thumbnail** | AI-generated preview thumbnail with bold text + high contrast | ⭐⭐⭐⭐ | 缩略图与内容一致 | | **Social Proof Badge** | Downloads + stars count displayed in title | ⭐⭐⭐⭐ | 数字真实可核 | | **Update Cadence** | 固定节奏更新并在描述中体现更新时间 | ⭐⭐⭐⭐ | 每次确有实质更新 | | **Boundary Line** | 在描述中明确本技能不做什么 | ⭐⭐⭐⭐ | 与正文能力声明一致 | --- ### 6. Full Skill Optimization Report / 全流程技能优化报告 Generate a complete optimization report combining all analysis: ```markdown # 🎯 ClawHub Skill Optimization Report **Skill**: [skill-name] **Generated**: [timestamp] **Analyzer**: ClawHub Skill Growth Engine v2.1.3 --- ## 📊 Current Status | Metric | Current | Target | Gap | |--------|---------|--------|-----| | Downloads | XXX | 1,000+ | +XXX | | Stars | XX | 100+ | +XX | | Description Length | XXX chars | 200-300 | OK | | Tags Count | X | 10-15 | Add X | | Has English README | No | Yes | MISSING | | Last Updated | YYYY-MM-DD | < 30 days | STALE | --- ## 🔥 Trending Keywords to Integrate | # | Trending Topic | Relevant Tag | Priority | Integration | |---|---------------|-------------|---------|-------------| | 1 | [topic] | [tag] | HIGH | Add to description | | 2 | [topic] | [tag] | MEDIUM | Add to tags | | ... | ... | ... | ... | ... | --- ## 📝 Title & Description Rewrite ### Current **Title**: [old title] **Description**: [old description] ### Optimized **Title (EN)**: [optimized English title] **Title (CN)**: [optimized Chinese title] **Description (EN)**: > [SEO-optimized English description - AIDA framework] **Description (CN)**: > [优化后的中文描述] ### Tags Optimization **Current**: [tag1, tag2, ...] **Add**: [new tags] **Remove**: [obsolete tags] --- ## 💬 Review Analysis Findings ### Top Requests from Users 1. [Request 1] → Add to SKILL.md priority section 2. [Request 2] → Add FAQ section 3. [Request 3] → Create tutorial ### Pain Points to Fix 1. [Pain point 1] → Rewrite confusing section 2. [Pain point 2] → Add troubleshooting guide --- ## 📋 Action Plan (Priority Order) | # | Action | Type | Impact | Effort | |---|--------|------|--------|--------| | 1 | Add English README | Content | ⭐⭐⭐⭐⭐ | Low | | 2 | Rewrite title with trending keyword | SEO | ⭐⭐⭐⭐⭐ | Low | | 3 | Add missing feature from reviews | Feature | ⭐⭐⭐⭐ | Medium | | 4 | Cross-post on [platform] | Promotion | ⭐⭐⭐⭐ | High | | ... | ... | ... | ... | ... | --- ## ✅ Checklist for Publishing (v2.1) - [ ] Title contains main keyword + current trending keyword (AI Agent/MCP/etc.) - [ ] Description is 200-300 characters (English), first 30 chars include core keyword (Chinese) - [ ] Tags include trending + evergreen keywords (AI Agent, MCP, workflow automation, etc.) - [ ] English README added (English FIRST, Chinese SECOND) - [ ] Video thumbnail prompt generated for each target platform - [ ] 60-second video script completed (if promoting on social media) - [ ] Cross-platform social post plan drafted (小红书/抖音/B站/知乎) - [ ] Social proof badges added (download count, stars, recent update) - [ ] Changelog updated with this version's changes - [ ] Version bumped to X.X.X - [ ] **Ability declaration matches the body** (no "no code" claim alongside code blocks) - [ ] **No broken internal path references** in the body - [ ] **Terminology is current** (regulator names, regulation names, versions) - [ ] Bilingual trigger keywords in SKILL.md --- ## 报告落地清单 / From Report to Action 拿到优化报告后建议按以下顺序执行（均需人工确认后手动操作）： | 顺序 | 动作 | 为什么放在这个位置 | |-----|------|-----------------| | 1 | 修正一致性问题（声明 / 断链 / 术语） | 先消除会被判定待审的问题 | | 2 | 补内容示例与边界说明 | 内容是留存的基础，先于推广 | | 3 | 调整标题与描述关键词 | 有了内容支撑，关键词才站得住 | | 4 | 更新标签与触发词 | 影响被发现，但不影响满意度 | | 5 | 制作视频 / 缩略图素材 | 素材依赖已定稿的内容 | | 6 | 跨平台分发 | 最后一步，避免把未定稿内容传出去 | **验证节奏建议**：一次只改一类（标题 / 描述 / 内容），观察 1–2 周后再进入下一类；同时改多项会导致无法归因。 --- ## Reference Files | 类型 | 内容 | |------|------| | SEO 优化指南 | 完整 SEO 框架 + 关键词研究方法 + 当期热点方向 | | 热点追踪手册 | 数据源清单 + 检索逻辑 + 关键词映射方法 | | 评论分析模板 | 评论分类模板 + 情感打分口径 | | 视频 SEO 指南 | 各平台缩略图提示词 + 视频脚本模板 | > 说明：以上配套材料随技能目录提供；正文不列出内部文件路径，避免在仅发布主文档时产生断链。 ``` --- ## 常见误用与纠偏 / Common Mistakes | 误用 | 表现 | 纠偏动作 | 自检问题 | |-----|------|---------|----------| | 关键词堆砌 | 描述里塞满热门词，读完不知道技能做什么 | 先用一句话说清给谁、解决什么，再谈关键词 | 读完能说出它是做什么的吗？ | | 只改版本号 | 为更新而更新，内容无实质变化 | 每次更新至少补一组示例或一个表格维度 | 本次比上一版多了什么？ | | 声明与内容矛盾 | 声明无代码但正文含代码示例 | 发布前做一致性自检（见 Section 0b） | 声明与正文逐条对得上吗？ | | 标题夸大 | 承诺提升 10 倍等无法验证的效果 | 改为可验证的表述（覆盖范围、输出物） | 这个说法能被验证吗？ | | 断链引用 | 正文引用了不会随技能发布的内部文件 | 正文不写内部路径，改为描述性引用 | 这个路径随包发布吗？ | | 术语过时 | 使用已变更的机构名或法规名 | 发布前检索并更新为现行表述 | 这个名称还是现行的吗？ | | 一次改太多 | 标题、描述、结构调整混在一起，效果无法归因 | 每次只验证一类改动的效果 | 这次改动能归因吗？ | | 反例写成命令 | 为说明「不要这样做」而写出完整可复制命令 | 反例只描述做法，不写出完整命令 | 这段文本能被照抄执行吗？ | | 示例与禁令冲突 | 示例文本里出现本文禁止清单中的表述 | 示例与禁止清单逐条比对 | 示例是否违反了自己的禁令？ | ## 内容质量评分表（发布前自评） | 维度 | 权重 | 5 分标准 | 3 分标准 | 1 分标准 | 快速检查 | |-----|------|---------|---------|---------|----------| | 定位清晰 | 25% | 一句话说清给谁用、解决什么 | 能看出领域但受众模糊 | 读者无法判断适用范围 | 读首段能否复述 | | 结构完整 | 20% | 有痛点、能力、示例、边界、免责 | 缺 1–2 项 | 只有能力罗列 | 五项是否齐全 | | 示例可操作 | 20% | 每个能力配 2 组以上场景示例 | 只有 1 组示例 | 无示例 | 数每个模块的示例数 | | 声明一致 | 15% | 能力声明与正文完全一致 | 基本一致，个别措辞含糊 | 存在明显矛盾 | 逐条比对声明与正文 | | 术语时效 | 10% | 机构名、法规名、版本号均现行 | 个别过时 | 多处过时 | 检索机构名与法规名 | | 触发精准 | 10% | 触发词只覆盖职责范围 | 有个别泛化词 | 大量通用词 | 是否有不触发清单 | | 表格结构 | 加分项 | 每个表块各行分隔符数量一致 | 个别行缺列 | 多处缺列 | 逐表数分隔符 | > 评分低于 3.5 分不建议提交更新：低分更新会稀释技能的可信度，不如不动。 --- ## Workflow / 工作流程 ``` User Input: Current skill details / user reviews / trending goal ↓ [Step 1] Analyze Reviews → Extract pain points + requests [Step 2] Fetch Trending Data → Map to skill keywords (AI Agent/MCP/video SEO) [Step 3] SEO Rewrite → Title + description + tags + 2026 trending keywords [Step 4] Generate Video Assets → Thumbnail prompt + 60s video script [Step 5] Cross-Platform Plan → Platform-specific SEO + social syndication [Step 6] GitHub Stars Strategy → README + promotion plan + social proof [Step 7] Generate Full Report → Actionable checklist ↓ User confirms changes ↓ Apply: Update SKILL.md + README.md + tags + thumbnail prompts + changelog ``` > **执行边界提示**：上图中 "Apply" 一步指**用户确认后由用户自行完成的操作**。本技能只产出建议、报告与素材文案，不代用户修改任何技能文件、不代发任何内容。