cldsh
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CLOUDINARY_URL | No | Cloudinary credentials URL. Alternative to setting CLOUDINARY_CLOUD_NAME, CLOUDINARY_API_KEY and CLOUDINARY_API_SECRET individually. | |
| CLDSH_FOLDER_MODE | No | Override the detected Cloudinary folder mode. Valid values are fixed or dynamic. | |
| CLOUDINARY_API_KEY | No | Cloudinary API key when not using CLOUDINARY_URL. | |
| CLOUDINARY_PROFILE | No | Name of a saved cldsh credentials profile to use. | |
| CLOUDINARY_API_SECRET | No | Cloudinary API secret when not using CLOUDINARY_URL. | |
| CLOUDINARY_CLOUD_NAME | No | Cloudinary cloud name when not using CLOUDINARY_URL. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| cloudinary_listA | List the subfolders and assets of a Cloudinary folder (paths start at the root), or the matches of a wildcard such as "cld*" or "shots/*.png". |
| cloudinary_uploadA | Upload a local file or directory (inside the allowed local directories) into a Cloudinary folder. The resource type follows the file extension unless resource_type is set. If the upload would replace existing assets, the first call returns needs_confirmation with a confirm_token instead of uploading. |
| cloudinary_get_urlA | Build CDN delivery URLs for assets, optionally with a transformation (e.g. "f_auto,q_auto", "w_800,c_limit"). |
| cloudinary_infoA | Full metadata of one asset: dimensions, bytes, format, tags, version, created date, URLs. |
| cloudinary_deleteA | Delete an asset, the matches of a wildcard, or (with recursive) a folder and everything in it. The first call returns needs_confirmation with the full plan and a confirm_token; nothing is deleted until the call is repeated with that token. |
| cloudinary_moveA | Move or rename an asset, or move a folder (dynamic folder mode). In dynamic folder mode an asset move keeps its URL unless change_public_id is set. The first call returns needs_confirmation with the plan and a confirm_token. |
| cloudinary_treeB | Folder hierarchy with the assets of each folder, starting at a folder (default: root). |
| cloudinary_downloadA | Download assets (name, public ID or wildcard) into a local directory or file inside the allowed local directories. Existing files are skipped unless overwrite is true. |
| cloudinary_mkdirA | Create a folder (missing parent folders are created too). |
| cloudinary_rename_folderA | Rename or move a folder with everything in it (dynamic folder mode only; asset URLs do not change). The first call returns needs_confirmation with the plan and a confirm_token. |
| cloudinary_accountA | The connected Cloudinary account (cloud name, folder mode, where the credentials come from, remaining Admin API requests) and this server's safety settings. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
Most tools target a clearly distinct operation (get_url, info, upload, download, delete, mkdir, account). Two pairs blur: cloudinary_move explicitly also moves folders, which overlaps cloudinary_rename_folder, and cloudinary_tree overlaps cloudinary_list since the tree includes per-folder asset listings; the descriptions mitigate but do not fully eliminate the misselection risk.
Every tool uses the same cloudinary_ prefix with snake_case, which is highly predictable. Minor deviation: most names are verb-based (get_url, list, upload) while a few are bare nouns (info, tree, account), so the pattern isn't perfectly uniform.
11 tools sits squarely in the well-scoped range and each maps to a real asset-management need (listing, metadata, transfer, folder ops, account state). Nothing appears redundant enough to drop.
The core lifecycle is covered: browse (list/tree/info), transfer (upload/download), organize (mkdir/move/rename_folder), remove (delete), URL generation, and account introspection, with a consistent confirmation flow for destructive ops. Gaps exist around asset metadata updates (tags are reported by info but there is no tool to set them) and explicit asset rename/public_id changes outside move.