Wasabi
StorageLet AI agents manage hot cloud storage on Wasabi. Agents create and delete buckets, upload, download, copy, and delete objects, check metadata and existence, and generate presigned URLs through Wasabi's S3-compatible API.
What This Integration Enables
Object storage is where workflow evidence should end up: the order payload as it arrived, the report as it was generated, the document as it was signed. Wasabi's S3-compatible hot storage makes that archive cheap enough to keep everything, which means the real problem shifts from storing data to governing it. FlowRunner agents run the whole object lifecycle, ingest, organize, verify, share, and retire, while the operations that permanently destroy data wait for a person. - Archive payloads, reports, and documents from any workflow with Put Object and Upload Object from URL - Share private objects through presigned URLs instead of credentials or public buckets - Verify existence and inspect metadata before downstream steps depend on a file - Copy objects across buckets for backup, migration, and replication patterns - Run retention and cleanup as an approved, auditable process rather than a console session
Without FlowRunner
With FlowRunner
Use Case Scenarios
Every order leaves a permanent record
A new order fires in [Shopify](/integrations/shopify). The agent serializes the full payload and calls Put Object to write it under an orders/2026/07/ prefix, then Get Object Metadata to confirm the write and capture the ETag. [Gmail](/integrations/gmail) sends the fulfillment team a confirmation with the object key. Months later, when a chargeback dispute needs the order exactly as it arrived, the evidence is one Get Object call away instead of a database archaeology project.
Report distribution without an attachment in sight
A nightly workflow assembles the finance close package and stores it with Put Object. The agent then calls Generate Presigned URL with a 72-hour expiry and posts the link to the controller's [Slack](/integrations/slack) channel. Nobody's inbox holds a stale copy, the bucket stays private, and when the link lapses the document is simply no longer reachable. The same pattern hands auditors access to a specific folder of evidence for exactly the duration of the audit.
Migrating a media library one row at a time
A [Google Sheets](/integrations/google-sheets) sheet holds 4,000 rows of source file URLs from a legacy DAM. The agent reads each row, calls Upload Object from URL to pull the file straight into the target bucket, verifies it with Check Object Exists, and writes the new object key back to the row. Failures stay flagged in the sheet for a human pass. The migration runs for days without anyone babysitting a transfer window.
Human-in-Loop Highlight
Delete Multiple Objects removes up to 1000 objects in a single call, and on a non-versioned bucket there is no undo. A retention agent with a slightly wrong prefix filter does not delete one file incorrectly, it deletes a thousand. So FlowRunner's cleanup flows never purge on their own judgment. The agent assembles the manifest, verifies archive copies with Check Object Exists, and presents the summary: "912 objects under invoices/2023/, 38.2 GB, all verified in the archive bucket. Purge?" The storage owner approves, and only then does the batch delete run. The [human-in-the-loop](/concepts/human-in-the-loop) gate stands between a filter expression and an empty bucket.
Agent Capabilities
13 actionsBuckets
3- List Buckets Lists all buckets in the configured account with names and creation dates. The discovery step before any storage workflow.
- Create Bucket Creates a new bucket in the configured region. Names must be globally unique, lowercase, and 3 to 63 characters.
- Delete Bucket Deletes an empty bucket. Every object and version must be removed first, which makes this the final step of a decommission flow, never the first.
Objects
5- List Objects Lists objects with prefix filtering and pagination, returning keys, sizes, modified dates, and storage classes. Up to 1000 objects per request with a continuation token.
- Get Object Downloads an object's content inline, as UTF-8 text or Base64 for binary files. For large files or sharing, a presigned URL is the better tool.
- Put Object Uploads text, JSON, or Base64-encoded binary content as an object with a custom content type. The workhorse write for archiving workflow payloads.
- Upload Object from URL Downloads a file from a URL and uploads it into a bucket in one step. The ingestion path for external sources and migrations.
- Copy Object Copies an object within a bucket or across buckets in the same account, leaving the source unchanged. The backup and replication primitive.
Deletion
2- Delete Object Permanently deletes a single object. On non-versioned buckets there is no recovery, so agents pair it with a verification step.
- Delete Multiple Objects Deletes up to 1000 objects in one call with per-key success and failure results. The operation this page's human gate exists for.
Inspection and Sharing
3- Get Object Metadata Retrieves content type, length, modified date, ETag, and custom metadata without downloading the object. The inspection step before decisions.
- Check Object Exists Confirms an object exists and returns its last modified timestamp. Lighter than full metadata when existence is the only question.
- Generate Presigned URL Creates a time-limited URL for downloading or uploading a private object without credentials. The sharing mechanism that keeps buckets private.
Frequently Asked Questions
What can FlowRunner do with Wasabi?
FlowRunner agents can run List Buckets, Create Bucket, and Delete Bucket in Wasabi, plus 10 more actions.
Does connecting Wasabi to FlowRunner require OAuth?
Wasabi uses a custom authentication method to connect to FlowRunner.
Can Wasabi trigger a FlowRunner workflow automatically?
Wasabi doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Wasabi
$100 in credits. No card required. Connect in minutes.