用自然语言自动管理 Cloudinary 媒体资源、变换配置与用量监控。
复制安装指令,让 AI 自动完成配置 · 推荐新手
请帮我安装 askskill 上的 "Cloudinary Automation" 技能: 1. 下载 https://raw.githubusercontent.com/ComposioHQ/awesome-claude-skills/master/composio-skills/cloudinary-automation/SKILL.md 2. 保存为 ~/.claude/skills/cloudinary-automation/SKILL.md 3. 装好后重载技能,告诉我可以用了
Automate Cloudinary media management workflows -- create folders, configure upload presets, look up assets, manage transformations, search folders, and monitor usage -- all through natural language.
Toolkit docs: composio.dev/toolkits/cloudinary
https://rube.app/mcpCreate folder structures for organizing hosted images, videos, and raw files.
Tool: CLOUDINARY_CREATE_FOLDER
Key parameters:
folder -- full path of the new asset folder (required), e.g., images/events/2023Supporting tools:
CLOUDINARY_SEARCH_FOLDERS -- search folders by name, path, or creation date using Lucene-like expressions
expression -- search filter (e.g., name:sample AND path:events)max_results -- 1-500 results (default 50)sort_by -- list of sort objects (e.g., [{"created_at": "desc"}])next_cursor -- pagination cursorCLOUDINARY_GET_RESOURCES_BY_ASSET_FOLDER -- list assets within a specific folderExample prompt:
"Create a folder called 'marketing/campaigns/spring-2026' in Cloudinary"
Define centralized upload behavior including target folder, allowed formats, transformations, tags, and overwrite rules.
Tool: CLOUDINARY_CREATE_UPLOAD_PRESET
Key parameters:
name -- preset name (auto-generated if omitted)folder -- target folder path for uploads (e.g., samples/)allowed_formats -- comma-separated list (e.g., jpg,png,webp)tags -- comma-separated tags to apply (e.g., marketing,thumbnail)transformation -- incoming transformation (e.g., c_limit,w_500)eager -- eager transformations to generate on upload (e.g., c_fill,g_face,h_150,w_150)unsigned -- allow unsigned uploads (true/false)overwrite -- overwrite existing assets with same public_id (cannot be true when unsigned=true)resource_type -- image, video, or raw (default image)unique_filename -- append random suffix to avoid collisions (default true)use_filename -- use original filename (default false)moderation -- moderation type: manual, webpurify, aws_rek, etc.auto_tagging -- confidence threshold 0.0-1.0 for AI auto-taggingnotification_url -- webhook URL for upload notificationsExample prompt:
"Create an upload preset called 'product-images' that only allows JPG and PNG, stores in 'products/' folder, and auto-tags with 0.7 confidence"
Retrieve full details for a specific asset by its public ID, including metadata, derived assets, and related resources.
Tool: CLOUDINARY_GET_RESOURCE_BY_PUBLIC_ID
Key parameters:
public_id -- the asset's public ID (required)resource_type -- image, video, or raw (required)type -- delivery type: upload, private, authenticated, fetch, etc. (required)colors -- include color histogram and predominant colorsfaces -- include detected face coordinatesmedia_metadata -- include IPTC, XMP, and detailed metadataquality_analysis -- include quality analysis scoresphash -- include perceptual hash for similarity detectionversions -- include backed-up versionsrelated -- include related assetsmax_results -- max derived/related assets to return (1-500)Example prompt:
"Get full details for the image 'products/hero-banner' including color analysis and quality scores"
List existing transformations, apply eager transformations to uploaded assets, and clean up derived resources.
Tools:
CLOUDINARY_GET_TRANSFORMATIONS -- list all named and unnamed transformations…