← 返回 Skills 市场

Image Video Generation Workflow

Name: Image Video Generation Workflow
Author: codeman840425-del

作者 codeman840425-del · GitHub ↗ · v2.0.0 · MIT-0

cross-platform ✓ 安全检测通过

171

总下载

当前安装

版本数

在 OpenClaw 中安装

/install image-video-generation-workflow

功能描述

基于 ComfyUI / MCP image server / 浏览器审看回路的图片与视频生成工作流技能卡，纳入 YouTube、Bilibili、Flickr 与 GitHub 的 source pack

使用说明 (SKILL.md)

When to Use

用户要快速搭一条本地可操作的图片/视频生成工作流，需要把生成引擎、浏览器审看、灵感素材收集和可复用提示词整理成一条稳定回路时使用。

What It Does

给出一条更接近当前 OpenClaw 采集现实的最小落地路径：

用本地生成引擎产出第一轮图片或视频结果
用浏览器统一审看生成结果
用 YouTube / Bilibili 补工作流演示，用 Flickr 补视觉灵感，用 GitHub 补可执行入口
只把稳定提示词、稳定工作流、稳定素材来源沉淀成可复用资产

How to Use

先选主生成入口：
- ComfyUI / Stable Diffusion：适合节点式图片和视频工作流
- mcp-image：适合把图片生成与编辑接进 MCP 工具链
- MeiGen-AI-Design-MCP：适合本地多方向并行生成与设计探索
再建立浏览器审看回路：
- 生成后统一在浏览器里看结果
- 标记保留、废弃、待重跑
- 把最稳定的一组提示词和参数回写到工作目录
再按 source pack 补上下文，但不要混用角色：
- GitHub：找真正可接入的仓库与工作流入口
- YouTube / Bilibili：看实际演示、参数思路、回路编排
- Flickr：补视觉灵感与素材方向，不把它当成可执行 workflow 文档
只保留高价值沉淀：
- 稳定提示词模板
- 稳定工作流 JSON
- 稳定素材来源清单
- 明确的人工筛选规则

Source Pack Guidance

当前这张技能卡对应的已验证 source pack 是：

youtube
bilibili
flickr
github-search

使用顺序建议：

先从 GitHub 选可执行入口
再用 YouTube / Bilibili 补实操和审看回路
最后用 Flickr 只补视觉参考，不让灵感源反客为主

Minimal Workflow

选一个主生成入口并固定参数范围
用同一主题跑 3 组图片或视频结果
在浏览器统一审看并标记保留/废弃/待重跑
查一个 GitHub 入口和一个视频演示，补最小执行路径
只把保留下来的提示词、参数、审看结论沉淀成可重复工作流

Guardrails

不把 Flickr 这类灵感源当成实施说明书
不把单条视频教程当成唯一事实来源
先跑最小闭环，再增加模型、风格或后处理复杂度
每次只调整一组提示词或参数，避免失去对比性

References

安全使用建议

This is a benign, high-level workflow card — it doesn't ask for credentials or install code. Before relying on it: (1) treat the SKILL.md as guidance, not an automated script — if you expect automation (browser control, headless runners), ask for explicit tooling and safe install instructions; (2) review any third-party GitHub repos linked before running their code (check licenses, README, and installation steps); (3) when using online source material (YouTube/Bilibili/Flickr), ensure you respect copyright and usage rights; (4) if you plan to automate browsing or remote downloads, require an explicit install spec and vetted release sources first.

功能分析

Type: OpenClaw Skill Name: image-video-generation-workflow Version: 2.0.0 The skill bundle defines a structured workflow for image and video generation using ComfyUI and MCP tools. It provides instructions for the agent to generate content, review results via a browser, and utilize external sources like GitHub, YouTube, and Flickr for inspiration and technical setup. No malicious code, data exfiltration, or harmful prompt injection attempts were found; the instructions are consistent with the stated purpose of managing creative workflows.

能力评估

ℹ Purpose & Capability

The skill's name and description (ComfyUI / MCP image server / browser review loop + source pack) align with the SKILL.md content: a human/agent-guided workflow for producing and curating images/videos. Minor mismatch: the frontmatter lists capability as "浏览器与执行" (browser and execution) but the instructions are high-level and don't declare any automated browser tooling, binaries, or commands; this is plausible for a manual workflow card but worth noting.

✓ Instruction Scope

SKILL.md stays within the stated purpose: selecting a local generator, reviewing outputs in a browser, capturing stable prompts/params, and consulting GitHub/YouTube/Bilibili/Flickr for context. It does not instruct reading unrelated files, exfiltrating secrets, or calling unexpected endpoints.

✓ Install Mechanism

No install spec and no code files are present (instruction-only). This minimizes on-disk risk — nothing will be downloaded or installed by the skill itself.

✓ Credentials

The skill requires no environment variables, credentials, or config paths. That matches the workflow's purpose of guiding local/manual generation and curation.

✓ Persistence & Privilege

always is false and there is no install step that modifies agent config or requests persistent privileges. The skill does not request elevated or persistent presence.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install image-video-generation-workflow
安装完成后，直接呼叫该 Skill 的名称或使用 /image-video-generation-workflow 触发
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

v2.0.0

Version 2.0.0 introduces a major update with enhanced workflow clarity and source pack integration: - Added explicit integration of YouTube, Bilibili, Flickr, and GitHub as recommended source packs for inspiration, tutorials, and workflow assets. - Clarified usage guidance, emphasizing stepwise use of generation engines, browser-based review, and distinct roles for each source type. - Provided detailed minimal workflow steps and new guardrails to avoid source misuse and improve result stability. - Expanded practical instructions and context for asset organization, human review, and template iteration. - Updated references to include a live Flickr sample.

v1.0.0

Image Video Generation Workflow v1.0.0 - Initial release of the Image Video Generation Workflow skill. - Provides a concise, actionable workflow for setting up local image and video generation pipelines using ComfyUI, MCP image server, and browser review loops. - Includes references to key GitHub projects and video tutorials for quick onboarding. - Details step-by-step instructions to connect model generation, scripting, review processes, and prompt iteration in a reproducible workflow. - Recommends readily available tools and content for users building or refining local AI generation pipelines.

元数据

Slug image-video-generation-workflow

版本 2.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 2

常见问题