Skip to main content
Blog

AI SRE Artifacts: A permanent, shareable link per report

Reports, design documents, and pages from your AI SRE sessions can now be published as artifacts. An artifact does two things: it stays, with its own address in the team gallery, and it goes out, over a public link that needs no sign-in.

Flashduty Product

AI SRE Artifacts: A permanent, shareable link per report

You can now publish anything AI SRE makes in a session as an artifact: a report, an analysis, a design document. The artifact gets its own address in your team's gallery. Publish again and it updates in place. When you need to, you can create a public link that opens without sign-in, and revoke it at any time.

Here's the problem it solves. Three weeks later, you can't remember which session produced that weekly breakdown. To show a colleague, you point them at your session history. To show a customer, you send screenshots or a PDF that goes stale as soon as anything changes.

Open a real one first: AI SRE — Deep Dive is an artifact we published. It has its own address and opens without sign-in.

What you can publish

An artifact is a file the agent made in a session and then published. Most are self-contained web pages: an investigation report, a design document, a data analysis. Typing /insight in a session, for example, produces an operational insight report. You can also publish Markdown and plain text, CSV, JSON, and YAML files, images, PDFs, source code in common languages, and archives.

The gallery has no upload button. Every artifact comes from a session. When the agent shows you a file it made, the publish action sits right next to it.

It stays

A published artifact lands in the gallery with its own address. You can search by title and filter by personal or team scope. Three behaviors are worth knowing:

  • Publishing again updates the artifact in place. The address doesn't change, so the link you already sent shows the new version.
  • Deleting the session doesn't delete the artifact. Once published, it stands on its own.
  • The artifact page shows which session last updated it. If you still have access, one click takes you back to that conversation, so you can see what the agent looked at and how it got there.

It goes out

There are two ways to share. Pick one based on who will read it.

Account onlyPublic link
Who can open itMembers signed in to the same accountAnyone with the link, no sign-in required
What they seeAlways the latest versionA snapshot of the content taken when the link was generated
Good forColleaguesCustomers, outside collaborators, people without an account

With a public link, the reader needs no account and no invite to your workspace. That's why we built three safeguards:

  • Artifacts are private by default. They are visible only in your account and never go public on their own. You create a public link yourself, and before you do, the panel shows you the content and warns you about the risk.
  • You can revoke a link at any time. It stops working immediately.
  • Readers of a public link see a fixed notice at the top. It says the content is user-generated and unverified, and offers a way to report inappropriate content to us.

Two limits. A public link is a snapshot. When the artifact changes, the link doesn't follow; you refresh it yourself, and the address stays the same. Artifacts over 16 MiB can't get a public link, but account-only sharing still works.

Public links require an On-call Pro or higher subscription and open with the AI SRE public beta (see the artifacts documentation).

Teams and scope

An artifact takes the scope of the session it came from. A personal session makes a personal artifact. A session bound to a team makes a team artifact.

Everyone on a team can open and manage that team's artifacts. The first publish sets the scope. If you have edit rights, you can change it later on the artifact page. Account owners and admins can manage every artifact in the account.

The gallery filters by the same two levels. If you're on many teams, narrow it to the ones you care about.

What to make

Reports are one use. Each example below has a prompt you can paste into a session. In every case the agent gives you something to keep and send, not a block of chat text.

A design document for review. Turn a redesign you just discussed into a document for the team. Or turn an investigation into an end-state fix plan, so whoever picks it up doesn't need it explained again.

Turn the caching redesign we just discussed into a design document: the background, the options we compared, the recommendation and its reasoning, built as a web page and published to the team.

Based on the investigation we just finished, write a fix plan: what the problem is, what the end state should look like, how many steps to get there, and how each step gets verified, built as a web page and published to the team.

When review comments come back, go back to the session, have it revise, and publish again. The address stays the same, so your colleague's link opens the latest version.

A document the team writes together. Postmortems, on-call handbooks, and weekly reports all work this way. Draft with the agent in a session and publish to the team when it's ready. There's one artifact at one address, so you stop passing five copies of v3_final around the group chat. When someone updates it later, they publish again, and the artifact page records which session the update came from.

An investigation timeline. List what an investigation looked at, in order.

Turn the investigation we just did into a timeline, noting what was examined at each step and what it established, and build it as a web report and publish it.

One link for outsiders. A customer wants progress, or a partner wants the conclusion. Skip the PDF and the email. Create a public link, drop it in the chat, and they open it without signing in.

Write an update for an outside reader based on the analysis above, leave out internal hostnames and internal system names, build it as a web page and publish it, then generate a public link.

Results from automations, such as operational insight, alert noise cleanup, and post-deploy checks, can be published as artifacts too. The New artifact button in the product works differently. It opens a session with a draft prompt filled in, and the agent asks about audience, content, and style before it starts.

Getting started

In the console, open AI SRE → Artifacts in the left navigation.

Related posts: From MCP to a CLI, five ways to use AI SRE Automations, and from a real user error to a fix PR in 8 minutes.

Related articles