Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Fangyuan025/Chaty --skill tiktok-videogit clone --depth 1 https://github.com/Fangyuan025/ChatyWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/fangyuan025/chaty/tiktok-video)<a href="https://agentmods.dev/skills/fangyuan025/chaty/tiktok-video"><img src="https://agentmods.dev/badge/skills/fangyuan025/chaty/tiktok-video/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/fangyuan025/chaty/tiktok-video"><img src="https://agentmods.dev/badge/skills/fangyuan025/chaty/tiktok-video.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00059 | $0.03073 |
| Opus 5 | $0.00030 | $0.01537 |
| Sonnet 5 | $0.00012 | $0.00615 |
| Haiku 4.5 | $0.00006 | $0.00307 |
Grade A, and why
tiktok-video scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.
TikTok / 抖音短视频生成
你做创意部分——文案、分镜、搜索关键词、质量审查;{SKILL_ROOT}/scripts/ 里的脚本做机械部分——TTS 逐字时间戳、素材搜索下载、字幕渲染、ffmpeg 合成、响度标准化。脚本已随 Chaty 就位,勿重写。
前置:ffmpeg + python3 在 PATH(macOS: brew install ffmpeg;Windows 需 Git Bash,且下文的 .venv/bin/python 换成 .venv/Scripts/python)。零 API key 即可跑;环境变量 PEXELS_API_KEY / PIXABAY_API_KEY(免费申请)可解锁实拍视频片段,有则优先。
0. 一次性安装(若 {SKILL_ROOT}/.venv 已存在则跳过)
bash {SKILL_ROOT}/scripts/setup.sh
1. 先核实,再写文案 + 分镜
语言规则(最先决定):视频语言 = 用户需求的语言(或用户明确指定的目标受众),永远不是本文档或示例的语言。英文需求 → lang: "en",文案、hook、角标("No.1"/"TOP 1")、CTA 全英文,en-US 音色,TikTok 习惯;中文需求 → lang: "zh",抖音习惯("第1名")。用户点名目标市场("给美国观众")则以市场为准。lang 缺省管线会直接报错,没有默认值。
**模型的内部知识可能过期或有错——未经核实的论断禁止进视频。**动笔前用你的联网搜索工具核实文案里每一个具体论断:数字、统计、纪录、价格、日期、排名、一切"第一/最大/唯一/最快",以及任何时效性内容(新闻/产品/版本/"今年/最新"),无论你多有把握。规则:按来源改写文案(不是反过来);核实不了就换成能核实的或直接删;把依据的 URL 记入 storyboard 的 "sources": [...](会出现在 report.txt 供用户复核);确实无联网能力时只用教科书级常识、避免具体数字、并向用户声明未核实。
然后读 {SKILL_ROOT}/references/writing-guide.md(hook 公式、节奏、场景结构)。然后在工作区建 <slug>/storyboard.json:
{
"title": "深海里最诡异的3种生物",
"lang": "zh",
"aspect": "9:16",
"voice": "zh-CN-YunjianNeural",
"rate": "+10%",
"caption_style": "karaoke",
"bgm": {"mood": "mystery", "gain_db": -16},
"hook": {"text": "深海禁区", "seconds": 2.3},
"scenes": [
{
"text": "你知道吗?在阳光永远照不到的深海,藏着比科幻电影更诡异的生物。",
"keywords": ["deep sea NOAA ocean exploration", "submarine dark ocean"],
"providers": ["openverse", "wikimedia"],
"effect": "kb_in"
},
{
"text": "第一名,鮟鱇鱼。头顶挂着一盏会发光的钓鱼灯,守在黑暗里等猎物送上门。",
"keywords": ["humpback anglerfish", "anglerfish museum"],
"badge": "第1名"
}
]
}
字段速查:
| 字段 | 取值 | 说明 |
|---|---|---|
lang |
zh | en(必填) |
= 用户需求/受众的语言(见上方语言规则);决定字幕分组和默认音色 |
aspect |
9:16(默认) | 16:9 | 1:1 |
|
voice |
任意 edge-tts 音色 | zh: zh-CN-YunjianNeural(磁性男) zh-CN-XiaoxiaoNeural(女) zh-CN-YunxiNeural(阳光男);en: en-US-ChristopherNeural en-US-AriaNeural |
rate |
如 +10% |
营销号节奏:zh +8%+15%,en +5%+10% |
caption_style |
karaoke | pop | none |
karaoke = 逐字高亮(推荐) |
bgm |
{"query":"风格词"} | {"mood":…} | {"file":"路径"} | {"mood":"none"} |
默认选 mood:曲表(upbeat funny inspiring chill tech mystery epic sad horror)是无数营销号在用的 MacLeod 熟脸配乐,按项目随机换曲;表内没有贴合风格的才用 query 搜(ccMixter/Jamendo CC 曲,节拍器自动筛掉没鼓点的):盘点/悬念 "trap"、种草 "lofi chill"、励志 "epic cinematic"、搞笑 "quirky";两者可同设(query 先试、mood 兜底);热门歌用 file(版权自负) |
beat_sync |
true(默认) | false |
BGM 自动节拍分析,所有切镜吸附到节拍上(卡点);人声永不截断,只伸缩场景尾部留白;鼓点弱的曲子会自动跳过 |
bgm.vibe |
"spedup" | "slowed" | 不设 |
抖音标志性音色:spedup ≈1.25×提速升调(卡点/盘点主流),slowed = 减速+混响(情感向);任何 bgm 来源都可加,处理后自动重测节拍 |
hook |
{"text","seconds"} |
开头大字标题卡,≤8 字/词 |
sticky_title |
{"text":…} |
可选的顶部常驻话题条;默认关,用户要才加 |
sfx |
true(默认) | false |
转场 whoosh 音效 |
场景 keywords |
英文、具体名词的列表 | 每条 = 一个镜头画面;视频约每 3s 切一镜,场景超过 ~4s 就给 2–3 条(给少了会循环补拍) |
场景 badge |
如 "第1名" / "TOP 1" |
场景开头的大号盖章标签,榜单类必用 |
场景 providers |
列表 | 免 key 图片:openverse wikimedia nasa;免 key 实拍视频:wikimedia_video(Commons 视频转码)nasa_video(PD 太空/科学)archive_video(Prelinger 历史胶片);有 key:pexels_video pexels_photo pixabay_video |
场景 effect |
auto kb_in kb_out pan_left pan_right static |
第一镜的 Ken Burns 动效,后续镜头自动轮换 |
场景 media |
文件路径 | 跳过搜索,用你自己准备的素材 |
What ships with it
17 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- examples/deep-sea-zh.json 3.0 KB
- examples/nature-video-zh.json 2.1 KB
- examples/peru-chinese-history-zh.json 2.8 KB
- examples/space-facts-en.json 2.5 KB
- examples/tools-zh.json 1.6 KB
- LICENSE 1.1 KB
- references/writing-guide.md 4.7 KB
- scripts/assets.py 27 KB runs code
- scripts/beats.py 4.9 KB runs code
- scripts/bgm.py 16 KB runs code
- scripts/captions.py 11 KB runs code
- scripts/check.py 6.3 KB runs code
- scripts/common.py 11 KB runs code
- scripts/compose.py 18 KB runs code
- scripts/pipeline.py 1.0 KB runs code
- scripts/setup.sh 1.7 KB runs code
- scripts/tts.py 2.7 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 121 lines · 59 tokens per session scan A ade5939cb7d7
tiktok-video is a skill published in the GitHub repository Fangyuan025/Chaty (42 stars, last pushed today), licensed MIT. It adds 59 tokens to every session and 3,073 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
proposal-deck
Build a Wednesday proposal through five approval gates, using approved local references and on-device models.
audio-transcribe
Transcribe speech from audio files (mp3, m4a, wav, ogg, flac, webm) to text using the local whisper CLI — no API key. Use whenever a task hinges on the spoken content of an audio attachment.
ffmpeg
Process audio and video with the ffmpeg / ffprobe CLIs — convert, trim, extract audio, resize, change format, make GIFs, inspect media. Use for any audio/video transformation.
imagemagick
Edit and convert images with the ImageMagick magick CLI — resize, crop, convert format, compress, rotate, montage, annotate. Use for any still-image transformation.
lilbee-mcp
Search and manage the user's local lilbee knowledge base over MCP. Use whenever the user has indexed code, docs, PDFs, or web pages into lilbee and you need cited answers, or whenever they ask you to ingest content, swap models, or tune retrieval against their library. Every fact returned cites file and line.…
lilbee-mcp-wiki
Wiki layer for lilbee. Use only when the user explicitly asks about wiki / concept / entity / synthesis pages, or when lilbeestatus shows a built wiki. Requires the lilbee-mcp skill to be active for the underlying MCP connection.