据 AlphaMoat 统计,截至 2026-09,TikTok 累计下载 3,761 次、获 4 个星标,在全部 5,078 个 AI 技能中排名第 2,833 位,在「内容创作」分类(362 个)中排名第 196 位。
| 平台 | 累计安装 | 累计下载 | 星标 | 评论 |
|---|---|---|---|---|
| ClawHub ↗ | 91 | 3,728 | 4 | - |
| SkillHub ↗ | 170 | 3,761 | 4 | — |
各平台的计数口径不同,统计时间不一致。
帮助规划账号定位、内容支柱、选题、开场钩子、短视频脚本、分集系列、配文与标签,并分析留存和表现模式。素材可保存在本地;此技能不使用 TikTok API,不会代发内容或自动化平台操作,也不保证播放增长或走红。 可选保存的资料、草稿和手动录入指标会保留在 ~/.openclaw/workspace/memory/tiktok/,应检查保留期限并在不需要时删除。最终内容由人工审核品牌适配、事实和平台规则后自行发布。
资料发布于 2026-10-27
以下是根据技能文档生成的示例,不代表实际运行结果。
2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。
| 月份 | 累计下载 | 环比 | 星标 |
|---|---|---|---|
| 2026-09 | 3,761 | — | 4 |
| 2026-08 | 3,346 | +14.5% | 4 |
| 2026-07 | 2,921 | +20.4% | 4 |
| 2026-06 | 2,427 | +14.4% | 4 |
| 2026-05 | 2,121 | — | 4 |
| 2026-03 | 1,182 | — | 3 |
以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。
--- name: tiktok description: Local-first TikTok Growth OS for strategy, hooks, scripts, retention design, and analytics feedback. Use when the user mentions TikTok, short-form video, hooks, scripts, retention, virality, content pillars, series planning, account positioning, or performance review. Generates execution-ready outputs and stores all assets locally. No API, no posting, no platform automation. --- # TikTok Growth OS A local-first operating system for TikTok creators. Focus on retention, repeatability, and strategic content design rather than random virality.
## What this skill does Use this skill when the user wants help with: - TikTok niche positioning - content pillars - video ideas - hooks - short-form scripts - A/V storyboard formatting - series planning - captions and hashtags - retention diagnosis - performance pattern review This skill should produce execution-ready outputs, not vague inspiration. ## Core operating logic TikTok growth is treated as a system of controllable variables: - **Hook strength**: Does the first 1–3 seconds create tension or relevance? - **Retention design**: Does the script create momentum and payoff? - **Visual pacing**: Is there a pattern interrupt every few seconds? - **Audience fit**: Does this feel native to the target viewer? - **Series potential**: Can this idea create repeated return behavior? - **Data feedback**: What should be repeated, refined, or dropped? Do not present growth as magic. Do not guarantee virality or follower gains. Always frame outputs as strategic guidance. ## Output rules When generating TikTok content, prefer: - spoken-language phrasing - short sentences - immediate tension - specific audience fit - clear payoff - native short-form rhythm Avoid: - essay-style intros - generic motivational filler - slow setup with no payoff - over-explaining before the hook lands ## Default content workflow When the user asks for TikTok help, structure work in this order when relevant: 1. clarify niche or audience if already known from local memory 2. generate 3–10 content angles 3. produce 5 hook variations 4. build one execution-ready script 5. optionally save the result to local memory 6. if analytics exist, use prior performance to refine the output ## Hook generation rules Hooks should usually fall into one of these buckets: - curiosity gap - painful truth - contrarian take - mistake warning - identity-based recognition - specific promise - emotional confession - authority / signal of experience When generating hooks: - make each variation meaningfully different - label the type - explain briefly why it may work - avoid repeating the same sentence shape ## Script generation rules When writing TikTok scripts, default to this format: | Time | Visual | Spoken / Audio | On-Screen Text | |------|--------|----------------|----------------| Guidelines: - first 1–3 seconds must carry the hook - each segment should add tension, clarity, or payoff - include pattern interrupts where useful - optimize for vertical, short-form pacing - include CTA only if it fits naturally ## Performance review rules If the user provides metrics or performance context, diagnose using first principles such as: - weak opening - slow payoff - too much setup - vague topic framing - mismatch between title/hook and delivery - insufficient visual momentum - weak emotional or practical value - unclear audience targeting ## Local memory All files are stored locally only in: `~/.openclaw/workspace/memory/tiktok/` Files: - `profile.json` — niche, audience, goals, pillars - `content_bank.json` — saved ideas, hooks, scripts, captions, notes - `analytics.json` — manually logged video performance - `pattern_report.json` — latest summarized learning report ## Scripts | Script | Purpose | |--------|---------| | `scripts/manage_account.py` | Create or update account profile | | `scripts/save_content.py` | Save ideas, hooks, scripts, captions, or notes | | `scripts/list_content.py` | Browse local content assets | | `scripts/log_performance.py` | Log manual TikTok performance data | | `scripts/analyze_patterns.py` | Summarize local performance patterns | ## References - `references/hooks.md` - `references/retention.md` ## Safety boundaries - Local-only storage - No TikTok API - No account login - No posting - No scraping - No engagement automation - No claims of guaranteed virality The user is responsible for final review, posting, and platform compliance.
识别并去掉文本里的 AI 写作痕迹,让文字读起来更像人写。
把文本里 AI 写作的典型痕迹挑出来并改写得更自然,同时保留原意的写作编辑技能。
识别并去除 AI 生成文本痕迹的中文文字编辑技能。
wechat-search 检索公众号热门文章,并按相关性、热度和时效整理结果。
从公开短视频分享链接提取标题、简介和口播文案,并给出改写版本。
humanizer 为中英文文本提供更自然的表达和风格编辑指导。
写网文章节并按文件落盘,聊天里只回摘要与指针。
把一组营销方法模块收进同一个技能,按请求路由到对应模块取清单与可交付稿。
数据月份:2026-09 · 下载/星标/安装数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。