Skip to main content
Using Claude Code, Cursor, or another coding agent? Paste this into your prompt to run cross-browser and real-device tests, debug sessions, and wire up CI on the TestMu AI cloud:
Every kane-cli run produces an evidence pack: a single sealed .evidence file containing everything about the run — the test definition, a result summary, per-step screenshots (plus annotated copies highlighting what the agent acted on), browser console and network logs attributed to each step, run logs, and on failures a per-step failure record with the error, page state, and pointers into the logs. A pack is a self-contained zip. You can open it in the hosted viewer, share it with a teammate, archive it in CI, validate it, or merge several packs into one.

​
What’s inside a pack

Opened in the viewer (or unzipped), a pack contains:
  • The test definition — what was asked of the agent.
  • A result summary — status, duration, per-step outcomes, tags, who ran it (user name/email), a share link, and the environment (browser resolution, OS version).
  • Per-step screenshots — the page as the agent saw it, plus an annotated copy highlighting the element the agent acted on.
  • Browser console and network logs — captured for the whole run and attributed to the step that produced them.
  • Run logs — the CLI and runner logs for the execution.
  • Failure records — on failed runs, a per-step record with the error message, the page state at failure, and references into the console/network logs.
For a mobile run, the result summary records the device in the run environment, for example the device model and OS version.
The pack is the only place run artifacts live. There is no separate per-run log directory on disk.

​
Where packs live

Every run except testrun run seals a pack in its session directory:
Runs you keep are additionally copied into the project store in your working directory:
What lands in the project store depends on the surface: An interactive TUI session maps to one pack: it accumulates every run in the session and seals when you /exit or start over with /new. If kane-cli crashes mid-run, the next start automatically sweeps incomplete or orphaned packs. Packs belonging to live concurrent processes are untouched. There is nothing to clean up by hand.

​
The kane-cli evidence commands

​
Publishing to the dashboard

Runs that upload to Test Manager attach their sealed pack automatically. Replayed testmd runs and testrun executions publish their packs to your project as well, so the dashboard shows execution evidence even for cache-replayed runs that never re-author.

​
Next steps