OpenClaw

cineprompt

Build CinePrompt video prompts and share links without a browser. Converts natural language shot descriptions into structured CinePrompt state, generates prompt text, and creates shareable links (cineprompt.io/p/...). Use when asked to create video prompts, build CinePrompt links, describe shots for AI video generation, or batch-create prompts for a sequence of shots.

belafontelabs0 来源安装数☆ 0 仓库 Stars

查看来源 ↗安装与使用 SKILL.md ↗

安装与使用

请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求，再添加到工作流。

SkillHub CLI

skillhub install belafontelabs/cineprompt

打开来源仓库，查看说明和兼容环境后，按来源提供的命令或步骤安装。

将来源链接提供给你的 Agent，让它先读取说明、核对依赖，并向你说明将执行的安装步骤。

查看来源 ↗使用指南

技能说明

CinePrompt Skill

Build CinePrompt prompts and share links directly — no browser needed.

How It Works

User describes a shot in natural language
You translate it into a CinePrompt state JSON object
The create-share-link.js script inserts it into Supabase and returns a /p/ share link
User gets the link + prompt text to copy into any AI video tool

Authentication

Requires a CinePrompt API key (Pro subscribers only). Set via:

--api-key cp_xxx flag
CINEPROMPT_API_KEY=cp_xxx env var

Internal/owner use: set CINEPROMPT_SERVICE_KEY env var for direct insert (bypasses Pro check).

Creating a Share Link

echo '<STATE_JSON>' | node <skill>/scripts/create-share-link.js --api-key cp_xxx

Output is JSON: {"url":"https://cineprompt.io/p/abc123","shortCode":"abc123","promptText":"...","mode":"single"}

You can also pass the prompt text explicitly if you've crafted it by hand:

echo '{"state":<STATE_JSON>,"prompt":"Your custom prompt text"}' | node <skill>/scripts/create-share-link.js --api-key cp_xxx

State JSON Structure

{
  "mode": "single",
  "complexity": "complex",
  "subjectType": "character|object|vehicle|creature|landscape|abstract",
  "fields": { ... },
  "shots": [],
  "characters": []
}

Required Top-Level Fields

mode: Always "single" for now (multishot/fm support later)
complexity: "simple" or "complex" — complex unlocks format, camera body, lens, film stock, color science
subjectType: Which subject panel to show
fields: Object mapping field names to values (strings or arrays)

Field Reference

All valid field names and values are in <skill>/field-values.json. Key fields:

Style:

media_type (array): ["cinematic"], ["cinematic","documentary"], etc.
documentary_style: only when documentary is in media_type
genre: only when cinematic is in media_type (array OK)
commercial_type: only when commercial is in media_type
tone (array): `["...

来源信息

Source: github
Repository: openclaw/skills
Path: skills/belafontelabs/cineprompt
Version: —
Commit: 7f496f52208edc109392aa3485ed414a8cfbdc08
分类来源: rules
状态: active
可见性: public

这些信息来自收录时的来源分析，可能滞后；使用前请核对最新来源文件。

来源内容

SKILL.md

# CinePrompt Skill

Build CinePrompt prompts and share links directly — no browser needed.

## How It Works

1. User describes a shot in natural language
2. You translate it into a CinePrompt state JSON object
3. The `create-share-link.js` script inserts it into Supabase and returns a `/p/` share link
4. User gets the link + prompt text to copy into any AI video tool

## Authentication

Requires a CinePrompt API key (Pro subscribers only). Set via:
- `--api-key cp_xxx` flag
- `CINEPROMPT_API_KEY=cp_xxx` env var

Internal/owner use: set `CINEPROMPT_SERVICE_KEY` env var for direct insert (bypasses Pro check).

## Creating a Share Link

```bash
echo '<STATE_JSON>' | node <skill>/scripts/create-share-link.js --api-key cp_xxx
```

Output is JSON: `{"url":"https://cineprompt.io/p/abc123","shortCode":"abc123","promptText":"...","mode":"single"}`

You can also pass the prompt text explicitly if you've crafted it by hand:

```bash
echo '{"state":<STATE_JSON>,"prompt":"Your custom prompt text"}' | node <skill>/scripts/create-share-link.js --api-key cp_xxx
```

## State JSON Structure

```json
{
  "mode": "single",
  "complexity": "complex",
  "subjectType": "character|object|vehicle|creature|landscape|abstract",
  "fields": { ... },
  "shots": [],
  "characters": []
}
```

### Required Top-Level Fields

- `mode`: Always `"single"` for now (multishot/fm support later)
- `complexity`: `"simple"` or `"complex"` — complex unlocks format, camera body, lens, film stock, color science
- `subjectType`: Which subject panel to show
- `fields`: Object mapping field names to values (strings or arrays)

### Field Reference

All valid field names and values are in `<skill>/field-values.json`. Key fields:

**Style:**
- `media_type` (array): `["cinematic"]`, `["cinematic","documentary"]`, etc.
- `documentary_style`: only when documentary is in media_type
- `genre`: only when cinematic is in media_type (array OK)
- `commercial_type`: only when commercial is in media_type
- `tone` (array): `["...

搜索