Documentation Index
Fetch the complete documentation index at: https://docs.testsprite.com/llms.txt
Use this file to discover all available pages before exploring further.
What Agent Actions Is
A project-level archive of every UI test recording — each tile is the video TestSprite captured while walking through one user journey. The recording itself is the action log: click X, type Y, navigate to Z, all captured as video.This is UI-only. API projects don’t have it (HTTP requests don’t need to be filmed; their request/response bodies are the equivalent artifact).
- Where it lives
- When you'll use it
The Agent Actions tab in the project sidebar, under the Result group. It sits alongside the Test Report tab — Test Report is the aggregated outcome view, Agent Actions is the per-recording browse view.
How the Gallery Is Organized
Tiles are bucketed by status into three sections, in this order:| Bucket | What’s in it |
|---|---|
| Passed | Tests whose run finished and assertions held |
| Failed | Tests whose run finished and assertions didn’t hold |
| Blocked | Tests that couldn’t complete — running, idle, or anything that didn’t resolve to Pass / Fail |
Tiles need a recording to appear. Tests that ran but didn’t produce a video (rare — typically a vendor-side recording timeout) are excluded from the gallery. They’re still visible in the Integration Tests list.
Search and Filter
The toolbar above the gallery has:- Search — matches against the test title and description (case-insensitive)
- Status filter — Passed / Failed / Blocked
- Priority filter — High / Medium / Low
What You’ll See When the Project Is Mid-Run
Two transitional states:- No tiles yet
- Some tiles, more coming
First run hasn’t produced any recordings. You’ll see the in-progress placeholder with stages: Running agents → Recording sessions → Processing videos. Tiles trickle in as agents finish.
Agent Actions vs. Step-by-Step
Both are visual replay tools, but at different scopes:| Agent Actions | Step-by-Step Walkthrough | |
|---|---|---|
| Scope | Project — every UI recording in one gallery | Per-test — the run of a single test |
| Primary artifact | Video tile | Per-step screenshots + recorded video |
| Surface area | Browse / filter / triage at scale | Drill into one test’s flow + assertions |
| When to use | ”Show me everything that failed" | "What exactly happened on step 4 of this test?” |
Step-by-Step Walkthrough
Per-step screenshots, video, and behavior trace for a single test
Edge Cases & Troubleshooting
A failed test has no recording in the gallery
A failed test has no recording in the gallery
The recording wasn’t captured for this run — rare, typically a transient hiccup. Rerun the test; subsequent recordings land in the gallery normally. The original failure is still in the Test Report; the gallery just hides tiles with no playable video.
Tile opens but the video won't play
Tile opens but the video won't play
Usually a stale playback link. Refresh the page; the link is regenerated on every load.
Recordings from multiple projects in the same gallery
Recordings from multiple projects in the same gallery
Agent Actions is project-scoped. If your dashboard URL has multiple frontend projects under one parent (rare — typically a multi-URL setup), tiles show a project-name subtitle so you can tell them apart.
Playing a recording counts against credits?
Playing a recording counts against credits?
No. Recordings are static artifacts; playback doesn’t trigger any agent activity. Only re-running the test consumes credits.
Where to Go Next
Step-by-Step Walkthrough
Drill into a single test’s run
Test Detail
All project-level tabs, including this one
Rerun
Trigger a fresh run when a recording shows the wrong behavior
Refining Tests
Iterate on a test in natural language after watching its recording