这段 README 实际上写的是 inference.sh skills，一套给 AI agent 用的模型技能，底层通过 inference.sh CLI 调用。它能生成图片、生成视频、调用 LLM、搜索网页等。可用技能里，图像生成有 50 多个模型，比如 FLUX、Gemini、Reve；视频生成有 40 多个模型，比如 Veo、Seedance、Wan；LLM 包括 Claude、Gemini、Kimi、GLM；搜索包括 Tavily、Exa，还有 X/Twitter 自动化。安装方式包括 Claude Code 插件、Skills 和手动，具体命令原文用占位符，前提是先装 inference.sh CLI（belt），可用 belt app list 浏览应用。装好后技能以 /inference sh:flux image、/inference sh:google veo 等形式出现。适合在 agent 工作流里做图像、视频、LLM 和搜索调用。
查看英文原文
inference.sh skills AI agent skills for models via inference.sh CLI. Generate images, videos, call LLMs, search the web, and more. Contents Install as Claude Code Plugin Install as Skills CLI Setup Available Skills Documentation Claude Code Plugin Install all skills as a Claude Code plugin: [安装/使用命令见下方] Or add from GitHub: [安装/使用命令见下方] After install, skills are available as /inference sh:flux image , /inference sh:google veo , etc. Install as Skills All Skills [安装/使用命令见下方] Specific Skills [安装/使用命令见下方] Manual [安装/使用命令见下方] CLI Setup Requires inference.sh CLI ( belt ). Install instructions [安装/使用命令见下方] Browse apps: belt app list Available Skills Tools Skill Description ai image generation 50+ image models (FLUX, Gemini, Reve, etc.) ai video generation 40+ video models (Veo, Seedance, Wan, etc.) llm models Claude, Gemini, Kimi, GLM web search Tavily, Exa search twitter automation X/Twitter API SDKs Skill Description javascript sdk JS/TS SDK with streaming, tools, React python sdk Python SDK with async, streaming UI Components Skill Description agent ui Full agent interface chat ui Chat components tools ui Tool call/result rendering Guides Category Topics prompting Prompt engineering, video prompts design Landing pages, thumbnails, logos video Storyboards, explainers, ads writing Blogs, case studies, newsletters social LinkedIn, Twitter threads, carousels product Competitor analysis, personas, launches agent Agent self regulation when stuck or looping Documentation Getting Started Running Apps CLI Setup API & SDK Links: Website App Store Docs Blog