Locker

Locker is a signed, tamper-evident archive for automation runs. Add one step at the end of a workflow and every run is kept exactly as it happened, with a receipt anyone can check.

Receipts are kept while you subscribe, run data for fourteen months, and the export checks without us, so keep it as long as your rules require.

A ship’s records cabinet standing open: rolled charts closed with wax seals, ledgers, a ring of keys, and a chart spread on the desk in front of it

What you need

A FlurryPORT account. There is no password: you enter your email and we mail you a sign-in link or a code. You can create the account first, or let the connection step create it for you.

Locker starts free: 250 runs or 14 days, whichever comes first. When the allowance ends the step keeps archiving, so your workflow never breaks and no run is lost. Reading the archive from then on needs a Locker plan. Locker plans and prices are on the pricing page.

A paid plan includes a set number of runs each month. Each run past that is billed at $0.002 a run, with the month’s bill, and nothing stops.

Connect a workflow

Locker takes runs from wherever your workflow lives. Each way in is listed with where it stands today.

Make

By install link

Works today through an install link. Waiting for Make to approve it for the app list.

Set it up in MakeInstall link

Zapier

Awaiting approval

Built and submitted. Waiting for Zapier to approve it; it cannot be found in Zapier until then.

How it will work in Zapier

n8n

Self-hosted

Works today on self-hosted n8n as a community node. Waiting for n8n to verify it for their own list.

Set it up in n8nThe node on npm

A script or any HTTP call

Available

One request at the end of a job archives the run and returns the receipt.

The Locker API

An AI agent

Available

An agent archives its own work through the FlurryPORT MCP server.

The MCP server

The receipt page

Every run has a page at flurryport.io/receipts/{receiptId}. It needs no account. The page recomputes the fingerprints and checks the signature in the visitor’s own browser, so it shows whether the receipt holds without asking anyone to trust us. It shows the proof, not the run’s contents.

How receipts work explains what a receipt proves and what it does not. Once a day the head of every chain is folded into one root, shown on the anchors page, so anyone who keeps a copy can show what the chains looked like that day.

Reading your archive

Sign in to FlurryPORT and open Locker. Each workflow appears as an install with its own chain of runs. From there you can:

  • Search runs by text, date or label, and open any run to see what went in and what came out.
  • Check the whole chain from the first run to the last.
  • Rename an install. Your name wins over the one the workflow sends.
  • Export an install as a bundle. It holds every run, every receipt and a verifier that runs with no network and no FlurryPORT.

Removing a value you must not keep

When someone asks for their data to be removed, or a secret was archived by mistake, you can remove the value from a run without breaking the proof for anything else. Only the owner of the install can do it, signed in to FlurryPORT. It cannot be done from Zapier, Make or n8n, or with an API token.

One field in one run

Open Locker, open the run, and under Redact enter the path of the field, such as /inputs/email. A path is a JSON Pointer: it starts with / and names each step down to the field.

A value in every run that has it

Search the install for the value, such as a name, an email address or an account number. Then choose Redact these. One confirmation names the runs and the fields before anything changes. Search waits for a plan once the free allowance has ended.

What happens

  • It happens at once and cannot be undone. The value is replaced by a fingerprint that cannot be turned back into the value. Every other field stays exactly as it was.
  • The receipt still verifies, because that fingerprint is what the receipt covered from the start.
  • The removal is recorded on the chain as a receipt of its own. The record shows that a field was removed and when, never what it held.
  • A run whose run data has already aged out has nothing left to remove. Its receipt stands.

What is kept, and for how long

Receipts are kept while you subscribe, run data for fourteen months, and the export checks without us, so keep it as long as your rules require.

  • A receipt is the signed entry in the chain. It is what a verify link opens.
  • Run data is what went in and what came out.
  • The export is a bundle of every run, every receipt and the verifier. Take one before run data ages out. We mail a reminder a month ahead.

Locker is a separate product from Pipes, the capture and replay side of FlurryPORT, and has its own plans. They are on the pricing page. The documentation overview says how the two differ.

API reference

Every integration is a thin client over four routes. You can call them yourself from any tool that can send an HTTP request.

https://api.flurryport.io/api/v1/locker

Authentication

OAuth 2.1 with PKCE, as a public client, with one scope: locker:write. A token with that scope can archive runs and read them back and reaches nothing else in the account. The answer that carries the token also names the account by email, so a workflow tool can label the connection with it. Send the token as Authorization: Bearer <token>.

StepRoute
AuthorizeGET https://api.flurryport.io/api/v1/auth/oauth/authorize
Token and refreshPOST https://api.flurryport.io/api/v1/auth/oauth/token

The token route takes a form body, not JSON. Access tokens are short-lived and come with a refresh token.

Routes

MethodPathPurpose
GET/installsThe account’s installs and its allowance. The connection test.
POST/installs/ensureGet or create the install for a workflow by platform and name. Returns its install id.
POST/installs/{installId}/runsArchive one run. Returns the receipt.
GET/runs/{receiptId}Read one archived run with its receipt.

Two more routes are public and need no token:

MethodPathPurpose
GET/api/v1/receipts/{receiptId}The receipt alone, as the receipt page reads it.
GET/api/v1/locker/anchorsThe daily roots, as the anchors page shows them.

Archive a run

Post a run document to /installs/{installId}/runs. Only schema is required. inputs and outputs take any JSON value.

{
  "schema": "locker.run/1",
  "platform": "zapier",
  "workflow": { "id": "386608734", "name": "Order to invoice" },
  "run": {
    "id": "order-1001",
    "status": "success",
    "startedAt": "2026-11-13T18:02:11Z",
    "finishedAt": "2026-11-13T18:02:13Z"
  },
  "inputs": { "orderId": 1001 },
  "outputs": { "invoiceId": "INV-7" },
  "labels": { "client": "acme" }
}

The answer is 201 Created with the receipt:

{
  "ReceiptId": "2BsKb3W3pL6qkbnuH7yJGg",
  "VerifyUrl": "https://flurryport.io/receipts/2BsKb3W3pL6qkbnuH7yJGg",
  "ChainIndex": 12,
  "Hash": "0f4c6d1c9a7a...",
  "PrevHash": "9be2a41f77c0...",
  "InstallId": "1ThmJ7Hil227Rv0ntafrPd",
  "StoredBytes": 212,
  "Duplicate": false,
  "Allowance": { "State": "free", "RunsRemaining": 238, "DaysRemaining": 9 }
}

Send an Idempotency-Key header, or a run.id, to make a retry safe. The same key with the same content returns the first receipt with Duplicate: true and adds nothing to the chain.

Errors

StatusCodeWhen
400locker_run_invalidThe document is not valid: unknown schema, bad timestamp, too many labels.
401The token is missing, invalid or expired. Refresh it and retry.
404The install or receipt does not exist, or is not yours.
409locker_run_conflictThe same key was sent with different content. The first receipt comes back.
410locker_install_revokedThe install was removed and its chain is closed.
413locker_run_too_largeThe document is over 1 MB.
429Too many requests. Wait for the Retry-After seconds, then retry.

Limits

  • A run document is at most 1 MB.
  • 300 requests a minute for each account.

What happens to a run over 1 MB depends on where it comes from, because some tools can shrink it for you and some cannot:

FromA run over the limit
ZapierThe step archives a fingerprint of the oversized part in place of the part. The run gets its receipt.
n8nThe node archives the fingerprints without the contents. The run gets its receipt.
MakeThat run stops with a message. Archive the fields that matter in place of a whole file.
The APIThe answer is 413. Send a fingerprint of the oversized part in place of the part itself.