The problem with a bare link
When the AI took a screenshot via the browser tool, the result was a plain sentence: "Screenshot captured: https://api.aicodesit.com/storage/.../browser-screenshot-....png". Technically correct, practically useless — you had to manually click or copy the link just to see what the AI was looking at, and the file wasn't tracked anywhere in your project.
Screenshots save to Storage automatically
Every screenshot taken by the browser tool — in chat or inside a pipeline — now writes straight to your project's Storage panel, the same place generated images and videos already land. No extra step, no manual upload. Open Storage and every screenshot the AI ever took for that project is right there.
We also added a short delay before the screenshot is captured whenever a file was edited earlier in the same pipeline run — long enough for your dev server to rebuild and apply the change first, so the screenshot reflects the actual updated page instead of a stale one.
Real images, not markdown text
If the AI writes a screenshot link in its own reply — including the classic markdown image syntax  — the chat now renders it as an actual image inline, not raw text with a clickable link buried inside it.
The copy & open card
Every screenshot, wherever it appears, now shows a small card directly underneath the image with:
- The full URL, truncated and in monospace
- A copy button — one click, with a checkmark confirmation
- An open in new tab button — for viewing it full-size or sharing the direct link
Works everywhere a screenshot shows up
This isn't limited to one spot — the same image-plus-card treatment appears in both places a screenshot can appear:
- The AI message bubble — when the AI includes a screenshot in its written response
- The pipeline logs — the tool-call timeline row for a
<browser action="screenshot" />call, right where you'd otherwise just see a plain result line
Everything at a glance
| Change | Before | After |
|---|---|---|
| Screenshot storage | Not saved anywhere | Saved to the project's Storage panel automatically |
| Screenshot timing | Could capture before a file edit applied | Waits for the dev server to rebuild first |
| Chat display | Raw URL / markdown text | Real inline image |
| Pipeline log display | Plain "Screenshot captured: url" text | Inline image in the tool-result row |
| Getting the link | Manually select/copy from text | One-click copy button + open-in-new-tab |
All live now — no setup required. The next screenshot the AI takes, in chat or a pipeline, will show up the new way automatically.
Related: AI Browser Agent — browse, click, and screenshot real websites · How to automate your dev workflow with the Pipeline Builder