SeeCut — security grade SAFE, quality 57/100

Security audit verdict: SAFE · quality 57/100

No red flags found in any of the 11 categories — no credential harvesting, no data exfiltration, no curl-pipe-shell installer. Scanned against the SlowMist agent-security taxonomy, refreshed every 8 hours. Full audit →

by YeJe-cpu · Agent Tool · ★ 75

Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h

🔒 Is SeeCut safe to install? View the security audit →

About SeeCut

SeeCut · 网感口播精剪 An AI editor that watches its own cut — see, cut, loop. 中文 · English · 日本語 数字人 / 真人口播 → AI 自动精剪成"网感"动效短视频 → 可选推成剪映分层草稿微调。 一个给 coding agent(Claude Code 等)用的 skill:它自己理解素材、自己去找真证据、自己编排和渲染,再用 AI 看视频给自己质检、自己迭代,最后交你肉眼终审。 两条都有声音,点开即可播放。右边的画面由 agent 全程自动完成(真证据截图、卡片、手绘圈注、人像在全屏和小窗之间调度),最后只在剪映里调了音量。原始文件在 Releases · demo-v1。 "精剪"不是去气口的粗剪;"口播"是有真人或数字人出镜;"网感"是画面有料、动效跟着每句话走。 亮点 See:让 AI 真正"看"视频。 Claude、GPT 这类模型通常要先把视频抽成几张图片再看,看不出动效和节奏。SeeCut 接入 Gemini(通过 Antigravity CLI)直接看视频,素材理解和质检都靠它。 看得准:评委是校准过的。 同一条视频换个提示词,AI 能从"标准"说成"土味"。我们用答案已知的片子测评委,发现"打绝对分"不可靠、"两条比哪条好"才可靠,所以质检用两两对比,交换位置各跑一次都赢才算。 Loop:自己改、自己挑。 每版先过硬伤检查,再和上一版两两对比,赢了才采纳;最多 3 版,每版必须有肉眼可见的进步。眼睛准了,循环才会越改越好,而不是越改越偏。 找真证据。 口播提到什么,就去截真的网页、界面、案例当画面证据,禁止用假界面凑满画面。没有额外 B-roll 也能做得有料。 有章法,不是堆特效。 一套成体系的风格规范:设计令牌 → 组件库 → 编排决策表("这句话是什么功能 → 画面怎么处理")→ 逐镜施工单,层层有据。 数字人也给了实打实的做法。 HeyGen AV5 数字分身出人像,配音不用 HeyGen 内置音色(偏假),改用豆包语音合成,demo 里的声音就是这么来的。脚本和步骤都在 digital-human/。 能在剪映里微调。 交成片的同时交分层工程包;装了剪映草稿引擎还能直接推成剪映里分轨可改的草稿,音效也能自动配。 最快跑通 准备一个文件夹:一段 20-30 秒竖屏半身口播(自己拍的就行)+ 3-6 张口播里提到的东西的截图(可选)。 装好下面

agent-skillai-avatarai-videoclaude-codegeminiheygenhyperframestalking-headvideo-editing

Quick Facts

Stars75
Forks14
LanguageJavaScript
CategoryAgent Tool
Quality Score57.1023133533595/100
Last Updated2026-09-27
Created2026-09-25
Platformsclaude-code, gemini, node
Est. Tokens~14k

SeeCut alternative? Top 6 similar tools

Looking for a SeeCut alternative? If you're comparing SeeCut with other agent tool tools, these 6 projects are the closest alternatives on Agent Skills Hub — ranked by topic overlap, star count, and community traction.

  • kinocut by KyaniteLabs · ⭐ 174

    Guardrailed video editing MCP server for AI agents. FFmpeg, Hyperframes, repurposing tools, Python client, and

  • mcp-video by KyaniteLabs · ⭐ 72

    Guardrailed video editing MCP server for AI agents. FFmpeg, Hyperframes, repurposing tools, Python client, and

  • fullstack-mkt-skills by minhnv0807 · ⭐ 329

    60 production-ready AI marketing skills (29 VN + 30 Global) for Claude Code, ChatGPT, Gemini & Copilot. 4 regi

  • super-video-maker-skill by Bomx · ⭐ 211

    AI video production skill for agents: HeyGen avatars, Seedance b-roll, OpenAI images, Remotion, HyperFrames, s

  • cutcli-cookbook by xuliang2024 · ⭐ 190

    Cookbook, JSON templates, AI prompts and docs for cutcli — the CapCut / Jianying (剪映) draft CLI. Generate edit

  • erduo-broll-loop-engineering by erduo1998-cell · ⭐ 185

    SRT 驱动的双后端 B-roll Agent Skill:自动路由 HyperFrames / Remotion,集成 152 张 Shotcraft 镜头卡

More Agent Tool Tools

Explore other popular agent tool tools:

View all Agent Tool tools →

Popular JavaScript Agent Tools

Frequently Asked Questions

What is SeeCut?

SeeCut is An AI editor that watches its own cut: talking-head / AI-avatar video → auto-edited short video. 网感口播精剪:数字人/真人口播 → AI 自动精剪成网感动效短视频 → 可选剪映分层草稿. It is categorized as a Agent Tool with 75 GitHub stars.

What programming language is SeeCut written in?

SeeCut is primarily written in JavaScript. It covers topics such as agent-skill, ai-avatar, ai-video.

How do I install or use SeeCut?

You can find installation instructions and usage details in the SeeCut GitHub repository at github.com/YeJe-cpu/SeeCut. The project has 75 stars and 14 forks, indicating an active community.

What are the best alternatives to SeeCut?

The top alternatives to SeeCut on Agent Skills Hub include kinocut, mcp-video, fullstack-mkt-skills. Each offers a different approach to the same problem space — compare them side-by-side by stars, quality score, and community activity.

How this security grade is produced

Grades come from a rule-based scan built on the SlowMist agent-security taxonomy, covering 11 red-flag categories including credential harvesting, data exfiltration, and curl | sh installers. It is a first-layer scan, not a manual audit — we say so rather than overstate it.

The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.

Sources & who's responsible:

View on GitHub → Browse Agent Tool tools