Blog

How to share a Claude artifact with your team so it keeps what people enter

Claude made you a tracker or a request form, and your team has started using it. The two ways to share it, what each one keeps, and exactly what comes across when you bring it into Spryloom as a page that saves data.

An HR lead asks Claude for a small tool: a page where managers log equipment requests for new starters, and IT marks each one done. Claude makes it as an artifact. It saves what people enter, it looks right, and within a week three managers and two people in IT are using it every day.

Then the questions start. Can someone without a Claude account use it? Who changed that request? What happens to the data if the HR lead leaves? Can it email IT when a new request comes in?

This article is about sharing an artifact so the team can keep using it, and keep what they've entered. We make Spryloom, which is the second option below, so read that part knowing who wrote it.

Option 1: share it from Claude

Claude lets you share an artifact with other people, and an artifact can keep data that everyone using it sees. For a small team that already works in Claude, this is often enough, and it is the least work: nothing moves, and it stays where it was made.

Its limits come from where it lives:

  • Adding to it needs a Claude account, usually on your company's plan. Claude's help center says people outside your organization (on a personal plan, anyone but you) can read its saved data once signed in to Claude, but can't change it. A contractor without a seat can't add to it.
  • It belongs to one person's account. The tool and what the team entered sit with whoever made it.
  • It runs only while someone has it open. A morning reminder, or an email to IT when a request arrives, needs a schedule.

If none of those matter to your team, keep sharing it from Claude.

Option 2: bring it into Spryloom as a page that saves data

A page that saves data is a Spryloom page with lists of records. It opens behind sign-in for the people you invite, by email, whether or not they have a Claude account. Every record is stamped with who saved it and when, from their sign-in. It needs no server or database, and when it needs a schedule or email, it grows into an app at the same address.

There is no import button. Claude does the move, following the Spryloom skill, in Claude Code.

The artifact the page saved records data in each browser attached files stay behind Spryloom page index.html /~data lists behind sign-in, for people you invite
The page and its saved records come across. Data kept in each person's browser, and files people attached, don't.

What you need

  • Claude Code signed in to claude.ai, so Claude can read the artifact and list its saved data.
  • An artifact you can edit. If you can only view it, ask its owner first.
  • The Spryloom plugin, which gives Claude the tool that adds records to a page. Without it, Claude can bring the page but not the records, and it will say so and stop before publishing a replacement that is missing them.

Then give Claude the artifact and ask: "bring this into Spryloom for the managers and IT".

What Claude does

Reads the shapeField names and counts, not the records.
Tells you, and waitsNothing is published until you say yes.
Turns it into a pageOnly saving and reading change.
Publishes, then adds recordsOldest first.

It moves the data without reading it. Claude writes each saved document to a folder outside the one it will publish, and looks only at field names, record counts and which fields hold people. The records were typed by your team, so a script moves them.

It tells you what will come across, and waits. Something like:

It has 2 lists and 46 records: requests (34) and comments (12). I'll publish it as a page that saves data and bring every record. Only you can open it until you invite people. Records will show as saved by you, with each person's name from the artifact kept beside them.

It turns the artifact into a page. It copies the artifact into index.html, so it looks and works as your team knows it. Each collection becomes a list, declared in spryloom.yaml:

app:
  name: Equipment requests
  description: Managers request equipment for new starters, and IT marks it done.
runtime:
  frontend: static
  backend: none
access:
  visibility: invited
  admins: [emily@yourcompany.com]
data:
  lists:
    requests: shared
    comments: shared
  1. What the invitation names.
  2. No server, no database. Plain files.
  3. People you invite. Admins can change any entry.
  4. One list per collection, same names as in the artifact.

It publishes, then adds the records, oldest first, and lists before the lists that point at them. spry data equipment-requests shows how many records each list holds, and the counts should match what Claude told you.

Exactly what carries over

From the artifact On Spryloom
The page, its layout and behaviour Kept, with only saving and reading changed
Saved records Every one, in a list of the same name
Who saved each record Their name, kept beside the record. The record itself shows as saved by you, since you brought it
Record ids Each record gets a new id. The old one is kept as a key, so a comment still points at its request
Live updates while open The page reads again after each change and every 15 seconds
Anyone could edit any entry Only the person who saved an entry, or the page's admins, can change it. Claude asks who should be admins
Data kept in each person's browser Doesn't come across
Files people attached Don't come across
Asking Claude questions as it runs A page can't. It can become an app that calls Anthropic with your own API key, or drop that part
Reading a Claude connector Stays in Claude. An app can read the service directly if Spryloom allows its address and you add a key

Where Claude can't find a name for someone, their records show "Someone", and Claude tells you how many before you agree.

The artifact's link stays private. Claude never sends it to Spryloom or writes it into the page; Spryloom publishes the files Claude wrote.

Switching the team over

Your colleagues may still be using the artifact. Ask them to switch, and stop sharing it from Claude. If anyone adds to the artifact before then, ask Claude to bring across only the new entries: it compares against the keys already in each list.

Then invite people. Each gets an email with an Open button that signs them in, once, for seven days; after that they ask for a sign-in link by email. New records are stamped with the person's email automatically, so the page never asks for a name.

When it outgrows a page

"Email IT when a request comes in" is something a page can't do. Ask Claude to make it an app. It adds a server and a database, and publishes again under the same name, at the same address. The records are copied into tables named page_requests and page_comments, and the app can send email to its own people and run a job once or twice a day.

You can also read and add to the records from a chat. With the Spryloom connector in Claude.ai, Claude can read a page's saved records and add one when you ask it to, as you, with your rights.

What it doesn't do

  • Files can't come across, and a page can't store new ones.
  • Browser-only data stays in each browser. There is nothing to read it from.
  • No Claude inside the page. A page can't call Claude or any outside API. That needs an app.
  • No live sync with the artifact. Once you move, the artifact and the page are separate. Bring later entries across by asking again.
  • Sign-in is an emailed link. No passwords, and no Google sign-in.
  • Limits. A record is up to 64 KB, a list holds up to 10,000 records, and a page has up to 20 lists. Pages don't count toward the 10 apps per workspace.

spry data export equipment-requests saves every record as JSON and CSV whenever you want your own copy.

Start here

In Claude Code, signed in to claude.ai, add the plugin once:

/plugin marketplace add spryloom/claude-plugin
/plugin install spryloom@spryloom

Then give Claude the artifact and say who should use it. To read and add to the page's records from Claude.ai later, add the connector https://app.spryloom.com/mcp under Settings → Connectors.

Spryloom publishes apps and pages built with a coding agent behind sign-in, for the people you invite. This article was written by the people building it.

Get started free