批量处理(目录级)

SkillFiles & storage

Batch processing: apply the same operation uniformly to a batch of images/videos/audio files in a directory — batch compression, watermarking, format conversion, resizing, aspect-ratio conversion, volume normalization, etc. Use when the user says: 批量处理, 批量压缩, 批量加水印, 批量转格式, 一批图片/视频, 给这个文件夹, 全部转成, 批量缩

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the 批量处理(目录级) skill

What this skill tells your AI

The instructions your AI receives, as published by zju-real/easel in skills/openclaw/batch-process/SKILL.md and read by ahel’s review.

对整个目录的图片/视频/音频统一套用同一操作。走 skills/shared/scripts/batch_process.py, 逐个委派给对应确定性脚本(image_ops / video_ops / audio_ops)。

单文件处理见 image-editing / video-editing / audio-editing;本 SKILL 是它们的目录批量版

输入

字段必填说明
目录待处理文件所在目录
类型image / video / audio(决定用哪个 ops 脚本 + 文件筛选)
操作ops 子命令(如 resize/compress/watermark/aspect/convert/normalize)
操作参数视操作写在 -- 之后,原样透传给 ops 脚本

输出(默认 <目录>/batch_out/

  • 处理后的同名文件;报告成功/失败数。

执行

脚本路径(相对项目根):skills/shared/scripts/batch_process.pyrun -h / list -h)。

# 先预览会处理哪些文件
python skills/shared/scripts/batch_process.py list --dir imgs --type image

# 批量压缩图片到 500KB
python skills/shared/scripts/batch_process.py run --dir imgs --type image --op compress \
  --out-dir out -- --max-kb 500

# 批量加水印
python skills/shared/scripts/batch_process.py run --dir imgs --type image --op watermark \
  -- --text "@我的账号" --position bottom-right

# 批量视频转竖版 9:16
python skills/shared/scripts/batch_process.py run --dir clips --type video --op aspect \
  --out-dir out -- --ratio 9:16 --mode pad

# 批量音频转 mp3(改扩展名用 --ext)
python skills/shared/scripts/batch_process.py run --dir raw --type audio --op convert \
  --out-dir out --ext .mp3 -- --bitrate 192k

可用操作(透传给 ops 脚本,<脚本> <op> -h 看参数)

  • image(image_ops):resize / crop / pad / convert / compress / watermark / round / thumbnail
  • video(video_ops):compress / aspect / watermark / speed / gif / frame
  • audio(audio_ops):convert / normalize / denoise / fade / speed / trim

规则

  1. list 预览文件范围,确认无误再 run
  2. -- 之后的参数原样透传给 ops 脚本;不确定参数先跑单文件版(image-editing 等)验证一次。
  3. 改输出格式用 --ext(如 .mp3/.png),否则沿用原扩展名。
  4. 批处理串行执行(避免高负载);量大时耐心等,失败文件会单独列出不中断整体。
  5. 默认输出到 <目录>/batch_out/,不覆盖原文件。

参考来源

复用项目既有确定性脚本(image_ops/video_ops/audio_ops),本 SKILL 只做目录遍历 + 逐文件 委派 + 结果汇总,不重复实现处理逻辑。

Signals

GitHub stars
1k
Forks
143
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
batch-process-zju-real
Source
github.com/zju-real/easel