Hellomedia

默认假设宿主主模型不能看图/读视频/听音频。用户消息含图片/截图/照片/视频/音频(含剪贴板粘贴、路径),或要求视觉理解、生图/改图、生视频/改视频/延长(含 image_to_video / reference_to_video)、TTS/STT 时,必须使用本技能。有磁盘路径用 --image;粘贴图/无路径用 vision.py --from-clipboard(需用户重新复制图片到系统剪贴板)或请用户保存路径。不要先尝试主模型识图。生图支持 skill config.json 与 Codex/Hermes/OpenClaw/env/CLI 凭据。

49 stars 2 forksImages, Video & Audio

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add hellowind777/hellomedia --skill hellomedia
Manual install
git clone --depth 1 https://github.com/hellowind777/hellomedia.git /tmp/hellomedia
mkdir -p ~/.claude/skills
cp -r /tmp/hellomedia/. ~/.claude/skills/hellomedia

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# HelloMedia

独立完整的多模态 Skill:**识图 / 生图 / 改图 · 读视频 / 生视频 / 改视频 · 读音频 / 生音频**。  
脚本默认 **纯标准库,零硬依赖**(Pillow 可选,用于识图前**温和**大图压缩,默认偏清晰度)。

## 与 Grok Build 内置工具的关系

Grok Build 会话里的 `image_to_video` / `reference_to_video` 是 **host 内置工具**(运行时注入)。  
其底层产品是 **xAI Imagine REST**。本技能提供跨宿主可用的 REST 等价实现:

| Grok Build host tool | 本技能 CLI |
|----------------------|-----------|
| `image_to_video` | `python scripts/video.py --mode image_to_video --image ...` |
| `reference_to_video` | `python scripts/video.py --mode reference_to_video --reference ...` |
| (无纯 T2V host tool) | `python scripts/video.py --mode text_to_video` 或先 `generate.py` 再 I2V |

**默认一律走本技能**(假设宿主主模型无视觉)。仅当用户明确要求使用 Grok Build 内置 `image_to_video` / `reference_to_video` / `image_gen` 等 host tool,或 host tool 明显更合适时,才可改用 host。  
中转若返回 403 `not enabled for this group` = 账号未开通 Imagine,不是脚本契约错误。

**Sub2API / 中转**(如 `base_url` 非 `api.x.ai` 但 `api_format: xai`):
- 生图走 `POST /v1/images/generations`,载荷用 `aspect_ratio` + `resolution`(1k/2k),而非 OpenAI 的 `size/quality` 优先
- 图生视频 `image` 对象字段默认 `image_url`(官方 `api.x.ai` 仍用 `url`);可用渠道字段 `video_image_url_field` 覆盖
- **不会**因本机访问不了 `api.x.ai` 而拦截中转生视频(官方 CDN 预检仅对 `api.x.ai`)
- 媒体下载使用浏览器式 `User-Agent`(可用 `HELLOMEDIA_USER_AGENT` 覆盖),避免 imgen/vidgen CDN 拒连
- 参考实现对齐:[happy-loki/grok-media-skill](https://github.com/happy-loki/grok-media-skill)

## 路由

**默认前提:宿主主模型不能看图 / 不能可靠处理多模态。**  
含媒体或生成类请求时 **直接使用本技能**,禁止先试主模型再失败,也禁止只说「看不到 / 听不到」。
Read the full skill on GitHub

Details

Skill name
hellomedia
License
Apache-2.0
Repo updated
14 July 2026

FAQ

What does the Hellomedia skill do?

默认假设宿主主模型不能看图/读视频/听音频。用户消息含图片/截图/照片/视频/音频(含剪贴板粘贴、路径),或要求视觉理解、生图/改图、生视频/改视频/延长(含 image_to_video / reference_to_video)、TTS/STT 时,必须使用本技能。有磁盘路径用 --image;粘贴图/无路径用 vision.py --from-clipboard(需用户重新复制图片到系统剪贴板)或请用户保存路径。不要先尝试主模型识图。生图支持 skill config.json 与 Codex/Hermes/OpenClaw/env/CLI 凭据。

How do I install the Hellomedia skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add hellowind777/hellomedia --skill hellomedia. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Hellomedia popular?

It lives in hellowind777/hellomedia, which has 49 GitHub stars and 2 forks.

Related skills

Image Enhancer

52

ArogyaReddy/ComposioHQ-awesome-claude-skills

Improves resolution, sharpness, and clarity of images. Upscale and enhance image quality.

Images, Video & Audio11

Vidmuse Video Creator

47

erduo1998-cell/vidmuse-video-creator

Create images and videos with VidMuse through its official CLI, including first-time signup and authorization, model and cost selection, reference-driven video, and SRT-based B-roll production. Also use when a user asks about this Skill's capabilities, installation, or VidMuse access link.

Images, Video & Audio

Posit News

67

jeremy-allen/claude-skills

Use multiple sub-agents to fetch and display news from Posit, including blog posts, podcast episodes, videos, and events. Use when the user wants to see recent Posit news, blog updates, podcast episodes, videos, or company announcements.

Images, Video & Audio

Youtube Transcript

100

adriannoes/awesome-agentic-ai

Use whenever the user needs the transcript of a YouTube video — fetching, extracting, downloading, or pulling captions/subtitles/transcript text from a YouTube URL. Triggers on "get the transcript", "transcript of this video", "pull the captions", "download subtitles", "what does this YouTube video say". Primary path is DeepAPI (David's own product); yt-dlp is the local fallback.

Images, Video & AudioIn 2 lists27

Concert Playlist Builder

66

aparente/claude-skills

Extract artists from concert/festival lineups and create playlists to preview their music. Use this skill when: - User shares a concert, festival, or event URL and wants to discover the artists - User asks to create playlists from a lineup (Apple Music, YouTube, Spotify) - User wants to preview music before attending a show - User mentions RA (Resident Advisor), Songkick, Bandsintown, or any event page with artists Triggers: "create playlist from lineup", "extract artists from concert", "make a playlist for [event]", "preview the artists", "who's playing at [event]

Images, Video & Audio

Youtube Polls

100

adriannoes/awesome-agentic-ai

Use when David wants to create, track, or analyze YouTube community polls. Triggers on "youtube poll", "next poll", "poll results".

Images, Video & AudioIn 2 lists27

Browse the most popular Claude skills, all images, video & audio skills, or write your own with the skills generator.