← 返回 Skills 市场
okaris

Google Veo

作者 Ömer Karışman · GitHub ↗ · v0.1.5
cross-platform ⚠ suspicious
1295
总下载
2
收藏
2
当前安装
2
版本数
在 OpenClaw 中安装
/install google-veo
功能描述
Generate videos with Google Veo models via inference.sh CLI. Models: Veo 3.1, Veo 3.1 Fast, Veo 3, Veo 3 Fast, Veo 2. Capabilities: text-to-video, cinematic...
使用说明 (SKILL.md)

Google Veo Video Generation

Generate videos with Google Veo models via inference.sh CLI.

Google Veo Video Generation

Quick Start

curl -fsSL https://cli.inference.sh | sh && infsh login

infsh app run google/veo-3-1-fast --input '{"prompt": "drone shot over a mountain lake"}'

Install note: The install script only detects your OS/architecture, downloads the matching binary from dist.inference.sh, and verifies its SHA-256 checksum. No elevated permissions or background processes. Manual install & verification available.

Veo Models

Model App ID Speed Quality
Veo 3.1 google/veo-3-1 Slower Best
Veo 3.1 Fast google/veo-3-1-fast Fast Excellent
Veo 3 google/veo-3 Medium Excellent
Veo 3 Fast google/veo-3-fast Fast Very Good
Veo 2 google/veo-2 Medium Good

Search Veo Apps

infsh app list --search "veo"

Examples

Cinematic Shot

infsh app run google/veo-3-1-fast --input '{
  "prompt": "Cinematic drone shot flying through a misty forest at sunrise, volumetric lighting"
}'

Product Demo

infsh app run google/veo-3 --input '{
  "prompt": "Sleek smartphone rotating on a dark reflective surface, studio lighting"
}'

Nature Scene

infsh app run google/veo-3-1-fast --input '{
  "prompt": "Timelapse of clouds moving over a mountain range, golden hour"
}'

Action Shot

infsh app run google/veo-3 --input '{
  "prompt": "Slow motion water droplet splashing into a pool, macro shot"
}'

Urban Scene

infsh app run google/veo-3-1-fast --input '{
  "prompt": "Busy city street at night with neon signs and rain reflections, Tokyo style"
}'

Prompt Tips

Camera movements: drone shot, tracking shot, pan, zoom, dolly, steadicam

Lighting: golden hour, blue hour, studio lighting, volumetric, neon, natural

Style: cinematic, documentary, commercial, artistic, realistic

Timing: slow motion, timelapse, real-time

Sample Workflow

# 1. Generate sample input to see all options
infsh app sample google/veo-3-1-fast --save input.json

# 2. Edit the prompt
# 3. Run
infsh app run google/veo-3-1-fast --input input.json

Related Skills

# Full platform skill (all 150+ apps)
npx skills add inference-sh/skills@inference-sh

# All video generation models
npx skills add inference-sh/skills@ai-video-generation

# AI avatars & lipsync
npx skills add inference-sh/skills@ai-avatar-video

# Image generation (for image-to-video)
npx skills add inference-sh/skills@ai-image-generation

Browse all video apps: infsh app list --category video

Documentation

安全使用建议
This skill appears to do what it says (invoke Google Veo models via inference.sh), but exercise caution before installing or letting an agent run the suggested installer automatically. Specifically: - Do not blindly run 'curl https://cli.inference.sh | sh' — prefer manually downloading the binary from a verified release page and checking the SHA-256 checksums the SKILL.md references. - Understand what 'infsh login' will do: it likely creates API tokens and persists them locally; find the CLI's docs or source to see where credentials are stored and what scopes are granted. - Verify the publisher/domain (inference.sh, dist.inference.sh) and look for an official project repository, release artifacts, and independent reviews. If no reputable source exists, treat the installer as high risk. - If you must test: run the installer in an isolated environment (VM/container), avoid running as root, and inspect the downloaded binary and network activity. - If you need higher assurance, ask the skill author for a canonical, verifiable install URL (e.g., GitHub releases) and explicit documentation of what credentials the CLI requires and how it handles them. Confidence is medium because the skill is internally coherent but the remote-install pattern and undeclared credential handling are legitimate security concerns that should be resolved before wide use.
功能分析
Type: OpenClaw Skill Name: google-veo Version: 0.1.5 The skill is classified as suspicious primarily due to the high-risk installation method presented in `SKILL.md`: `curl -fsSL https://cli.inference.sh | sh`. While this is a common way to install CLIs, it represents a significant supply chain vulnerability, allowing arbitrary code execution if the `inference.sh` domain or its distribution server is compromised. Additionally, the `allowed-tools: Bash(infsh *)` permission grants the agent broad capabilities to execute any `infsh` command via Bash, which, combined with the `curl | sh` risk, elevates the overall risk profile. There is no clear evidence of intentional malicious behavior like data exfiltration or persistence within the provided files, but the potential for exploitation via the installation method is high.
能力评估
Purpose & Capability
Name/description match the runtime instructions: all examples and commands use the inference.sh CLI to run Google Veo apps. Nothing in the SKILL.md asks for unrelated cloud providers, binaries, or services.
Instruction Scope
Instructions are narrowly scoped to installing/running the inference.sh CLI and invoking specific 'infsh app run' commands. The SKILL.md does not instruct reading arbitrary system files or unrelated environment variables.
Install Mechanism
The SKILL.md recommends piping an installer from https://cli.inference.sh (curl -fsSL https://cli.inference.sh | sh). That is a network-downloaded installer from a third-party domain (not e.g., a traced GitHub release URL). Although the doc claims checksum verification is available, directing users/agents to run a remote install script (and to download binaries from dist.inference.sh) is higher-risk and should be validated manually before execution.
Credentials
The skill declares no required env vars, but the instructions call 'infsh login', which implies credential creation/storage and network-transmitted tokens. The registry metadata doesn't document what credentials the CLI will request or how tokens are stored — a gap between declared requirements and runtime behavior that could lead to unexpected credential handling.
Persistence & Privilege
The skill does not request always: true and has no install spec that writes files under the skill system. It does rely on an external CLI which may persist its own credentials/config, but the skill itself does not claim elevated or persistent platform privileges.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install google-veo
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /google-veo 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.1.5
- Initial release of google-veo skill using inference.sh CLI. - Supports Google Veo video models: Veo 3.1, Veo 3.1 Fast, Veo 3, Veo 3 Fast, and Veo 2. - Enables text-to-video generation with options for cinematic, high-quality outputs. - Includes quick start instructions, example prompts, and tips for effective video creation. - Provides model comparison table and sample workflow for customizing inputs. - Lists related video and image generation skills for expanded functionality.
v0.1.0
Initial release of the google-veo skill. - Generate high-quality videos using Google Veo models through the inference.sh CLI. - Supports multiple Veo models: 3.1, 3.1 Fast, 3, 3 Fast, and 2. - Enables cinematic text-to-video generation with detailed prompt guidance. - Provides sample prompts, workflow instructions, and links to related skills for broader AI video capabilities.
元数据
Slug google-veo
版本 0.1.5
许可证
累计安装 2
当前安装数 2
历史版本数 2
常见问题

Google Veo 是什么?

Generate videos with Google Veo models via inference.sh CLI. Models: Veo 3.1, Veo 3.1 Fast, Veo 3, Veo 3 Fast, Veo 2. Capabilities: text-to-video, cinematic... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1295 次。

如何安装 Google Veo?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install google-veo」即可一键安装,无需额外配置。

Google Veo 是免费的吗?

是的,Google Veo 完全免费(开源免费),可自由下载、安装和使用。

Google Veo 支持哪些平台?

Google Veo 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Google Veo?

由 Ömer Karışman(@okaris)开发并维护,当前版本 v0.1.5。

💬 留言讨论