Cloudinary
StorageManage a Cloudinary media library from your flows. Agents upload images, videos, and raw files with signed requests, tag, search, and organize assets, generate transformation and delivery URLs, and monitor account usage.
What This Integration Enables
A media library is infrastructure, and the test of infrastructure is what happens at the edges: the upload nobody labeled, the thumbnail serving a 4 MB original, the folder deletion that took a landing page with it. Cloudinary gives media real infrastructure semantics, public IDs, transformations, delivery URLs, a searchable index, and FlowRunner agents operate it that way: ingestion that labels as it stores, delivery that optimizes by default, and destruction that never outruns review. - Upload images, video, and raw files from URLs or FlowRunner storage, with tags, folders, and metadata applied on the way in - Generate resized, cropped, format-converted delivery URLs to feed optimized media into emails, sheets, and downstream services - Pre-generate derived versions so campaign launches hit warm CDN copies instead of first-request transforms - Organize at scale: search with Lucene-style queries, tag in bulk, rename, move, and manage folders - Download transformed variants into file storage for the steps that need a file, not a URL - Monitor credits, storage, and bandwidth from inside the flows that consume them
Without FlowRunner
With FlowRunner
Use Case Scenarios
The library that labels itself
Every image landing from [Dropbox](/integrations/dropbox-service) passes through [AI Vision](/integrations/ai-vision) first. The generated description becomes contextual metadata and the detected subjects become tags on Upload Asset, so the asset enters the library already searchable. Six months later, "find the warehouse exterior shots from spring" is a Search Assets query, not a scroll through nine hundred thumbnails.
Optimized media everywhere, without a designer in the loop
A product image needs to appear in an email, a spreadsheet report, and a partner feed, each at different dimensions. The agent calls Generate Delivery URL with the right transformation string per surface, w_400,h_300,c_fill for the card, q_auto,f_auto everywhere, without any API call or re-upload. Where a downstream system needs an actual file, Download Asset pulls the transformed variant into FlowRunner storage and hands it over. One original, every rendition, zero manual exports.
Reclaiming storage without breaking the site
Quarterly, the agent runs Search Assets for the heavy and the stale: bytes over a threshold, uploaded before the cutoff, tags that mark retired campaigns. The candidate list goes to the media owner with sizes and last-known usage, and only the approved set flows to Delete Assets, with empty folders swept afterward by Delete Folder. Get Usage documents the reclaimed storage in the summary. The library gets smaller; the live site does not notice.
Human-in-Loop Highlight
Delete Assets in prefix mode removes up to 1000 assets in a single call, an entire folder's worth, and Destroy Asset with CDN invalidation stops a deleted asset from being served anywhere its URL is embedded. That is the defining danger of a CDN-backed library: a delivery URL does not know it is about to die, and neither does the landing page, the email template, or the partner integration built on it. A prefix one character too broad is a site-wide incident. So FlowRunner cleanup flows never let an agent's query be the last word. The agent assembles the candidate list from Search Assets, shows the media owner exactly what matches, how many, how large, which folders, and the literal prefix or ID list about to execute. The person approves the blast radius; only then do Delete Assets and Delete Folder run, and the summary records what left the library. Storage is cheap to reclaim; a dead image on a live page is not.
Agent Capabilities
20 actionsUpload
2- Upload Asset Uploads an image, video, or raw file from a remote URL or FlowRunner file, with public ID, folder, tags, incoming and eager transformations, and contextual metadata. Returns the stored asset with its public_id and secure_url.
- Upload from URL Uploads directly from a public URL with Cloudinary fetching server-side, so nothing transits FlowRunner. Supports public ID, folder, tags, incoming transformation, and overwrite control.
Asset Management
5- Rename Asset Changes an asset's public ID, which also moves it when the new ID carries a different folder path, with optional overwrite of an existing target.
- Destroy Asset Permanently deletes a single asset by public ID, optionally invalidating CDN caches so it stops serving. Runs inside reviewed cleanup flows only.
- Update Asset Replaces an asset's tags, sets contextual metadata, or moves it to a different folder through the Admin API, changing only what is provided.
- Delete Assets Deletes up to 100 assets by explicit public IDs or up to 1000 by prefix in one call. The bulk operation this page's human gate exists for.
- Get Asset Details Retrieves an asset's full record: size, format, dimensions, tags, context, derived versions, and optionally color analysis, face coordinates, and EXIF metadata.
Search and Browse
4- List Resources Lists assets of a resource type, newest or oldest first, filterable by public ID prefix including folder paths, with cursor pagination.
- Search Assets Searches the whole library with Lucene-style expressions combining folders, tags, sizes, and dates, such as folder:products AND tags:sale or bytes>1000000. The query engine behind audits and cleanup sweeps.
- List Tags Lists the tags in use for a resource type, filterable by prefix, with cursor pagination.
- Manage Tags Adds, removes, replaces, or clears tags across one or more assets in a single call. The bulk labeling step behind library hygiene.
Transformation and Delivery
3- Apply Transformation Eagerly generates derived versions of an uploaded asset so resized, cropped, and optimized variants serve instantly from the CDN.
- Generate Delivery URL Builds a CDN delivery URL with on-the-fly transformations and optional signing, without an API call. The zero-cost way to feed optimized media to any other service.
- Download Asset Downloads an asset, optionally transformed on the fly, into FlowRunner file storage and returns the stored file's URL for downstream steps.
Folders
4- List Root Folders Lists top-level folders with names and paths, cursor-paginated.
- List Subfolders Lists the direct subfolders of a folder path.
- Create Folder Creates a folder at a path, including missing intermediate levels.
- Delete Folder Deletes an empty folder at a path. Runs after asset cleanup, never before review.
Account
2- Get Usage Retrieves the environment's usage report: plan, credits, storage, bandwidth, transformations, and object counts with limits and percentages. The consumption check inside media-heavy flows.
- Ping Verifies connectivity and credentials against the Admin API. The connection test at the top of a flow.
Frequently Asked Questions
What can FlowRunner do with Cloudinary?
FlowRunner agents can run Upload Asset, Upload from URL, and Rename Asset in Cloudinary, plus 17 more actions.
Does connecting Cloudinary to FlowRunner require OAuth?
Cloudinary uses a custom authentication method to connect to FlowRunner.
Can Cloudinary trigger a FlowRunner workflow automatically?
Cloudinary doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Cloudinary
$100 in credits. No card required. Connect in minutes.