菜单
Generate a video from text prompts and/or image inputs using Coze video generation. Use when you need text-to-video, image-to-video, first-frame control, last-frame control, or a returned last-frame image URL.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
OpenClaw/龙虾/扣子常见问题解答助手。当用户遇到启动问题、服务中断、模型调用失败、飞书集成问题，或对 OpenClaw、Coze、龙虾有疑问时召回此技能。 通过阅读官方 FAQ 文档帮助用户找到解决方案或指导操作步骤。
Convert speech audio to text using Coze ASR. Use when you need to transcribe spoken content from a remote audio URL or a local audio file into plain text.
Generate one or more images from text prompts using Coze image generation. Use when you need to create images from a natural-language prompt, produce multiple image variants, or generate sequential frames from the same prompt.
Convert text to speech using Coze TTS. Use when you need to synthesize spoken audio from one text input or multiple text segments, optionally with a specific speaker, format, sample rate, or speech settings.
| name | coze-video-gen |
| description | Generate a video from text prompts and/or image inputs using Coze video generation. Use when you need text-to-video, image-to-video, first-frame control, last-frame control, or a returned last-frame image URL. |
| homepage | https://www.coze.com |
| metadata | {"openclaw":{"emoji":"🎬","requires":{"bins":["node"],"config":["plugins.entries.coze-openclaw-plugin.config.apiKey"]}}} |
Generate a video URL from text and image inputs using Coze.
This skill runs as a Node CLI wrapper around the backend SDK. Do not use the SDK from client-side code.
node {skillDir}/scripts/gen.mjs --prompt "A serene mountain landscape with flowing clouds"
node {skillDir}/scripts/gen.mjs --prompt "A cinematic robot walking through neon rain" --duration 6 --ratio 9:16 --resolution 1080p
node {skillDir}/scripts/gen.mjs --prompt "Animate this concept art" --first-frame "https://example.com/first.png"
node {skillDir}/scripts/gen.mjs --prompt "Transition from sunrise to night" --first-frame "https://example.com/start.png" --last-frame "https://example.com/end.png" --return-last-frame true
--prompt <text> text prompt--image <url> reference image URL; repeat the flag to pass multiple images--first-frame <url> first frame image URL--last-frame <url> last frame image URL--model <id> model id, default SDK model is doubao-seedance-1-5-pro-251215--duration <seconds> video duration in seconds--ratio <ratio> one of 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive--resolution <value> one of 480p, 720p, 1080p--watermark <true|false> whether to keep watermark--seed <n> fixed random seed for reproducibility--camera-fixed <true|false> whether to fix the camera--generate-audio <true|false> whether to generate synced audio--return-last-frame <true|false> whether to return the last-frame image URL--max-wait-time <seconds> max synchronous wait time--callback-url <url> async callback URL passed through to the SDK--header <key:value> custom HTTP header; repeatable, alias -H--prompt, --image, --first-frame, or --last-frame is required.Task ID, Status, Video URL, and optionally Last Frame URL.sign must not be dropped, otherwise the asset may be inaccessible.plugins.entries.coze-openclaw-plugin.config.apiKey.{skillDir} means the directory containing this SKILL.md.doubao-seedance-1-5-pro-251215 supports text input plus first_frame and last_frame. Generic reference_image inputs from --image may be rejected by that model.--image when you have selected a model that explicitly supports reference_image.