By default, kane-cli uploads each session to TestmuAI Test Manager as a test case. This page covers what gets uploaded, where it ends up, and the related features that hang off the upload pipeline: project and folder selection, code export, share links, the post-session feedback prompt, and the local session directory.
This is the default behaviour for every session in both the TUI and the CLI. You do not need to opt in.
When a session ends, kane-cli runs an upload pipeline that finalises the test case in Test Manager. Two kinds of artefacts move from your machine to TestmuAI:
- Screenshots — uploaded incrementally as the agent runs. By the time the session ends, most of them are already in place. PNG screenshots are converted to WebP before upload.
- Run artefacts and metadata — packaged at session exit. This includes the per-run action logs, an execution blob describing the run (objective, status, steps, durations, variables in scope), and the zipped contents of the session run directories.
You see a short progress indicator at exit covering these stages: convert (preparing the metadata), zip (packaging the run directories), presign (requesting upload URLs), upload (sending the zip and metadata), and finalize (committing the test case in Test Manager). If code export is enabled, a code_export stage runs after finalize. If it is disabled, that stage is reported as skipped.
If the upload fails, kane-cli prints the error and exits. The local session directory is preserved either way — see Session history on disk.
Each test case is filed under a Test Manager project. Optionally, you can also choose a folder inside that project. You can configure either explicitly, list and create from the command line, or let kane-cli auto-default one for you on first run.
Configure your project once, and every subsequent session uploads under it.
kane-cli config projectWith no value, this opens a search-as-you-type picker. Projects are loaded from your Test Manager account on demand. Type to filter, use the arrow keys to navigate pages, press Enter to select, or Tab to create a new project. If your account has no projects yet, the picker jumps straight to the create flow. The picker works with either OAuth or basic-auth credentials.
You can also set a project non-interactively by ID:
kane-cli config project <project-id>In the TUI, the same picker is available via the /config command.
Folders are optional. Once a project is set, you can pick a folder the same way:
kane-cli config folderThis opens a search-as-you-type folder picker. As with projects, type to filter, Enter to select, Tab to create a new folder. Each folder is shown with its current test case count.
kane-cli config folder <folder-id>If you try to set a folder before choosing a project, kane-cli asks you to pick a project first.
For scripts, CI, and any non-interactive shell where the picker is not appropriate, kane-cli exposes the same data as commands:
kane-cli projects list [--search <q>] [--limit <n>] [--offset <n>]
kane-cli projects create "<name>" [--description "<text>"]
kane-cli folders list [--search <q>] [--limit <n>] [--offset <n>]
kane-cli folders create "<name>" [--description "<text>"]In a TTY, list commands render a paginated table. When stdout is piped or redirected — or when --agent is passed — the output switches to NDJSON, one JSON object per line, terminated by a small pagination summary. Use --search to filter by name on the server, and --limit / --offset to page through results.
folders list and folders create operate inside the currently configured project, so set a project before calling them.
After creating, persist the new id with kane-cli config project <id> or kane-cli config folder <id> so subsequent runs upload there.
If you launch a run without ever configuring a project or folder, kane-cli auto-resolves a sensible default at run start — find-or-create against your account — and tells you which one it picked. The same fallback fires for kane-cli testmd run and kane-cli generate. You don't have to pre-configure anything for the first run to succeed.
The same auto-default kicks in if a previously-configured project or folder has been deleted, renamed, or revoked, or if you set an invalid ID by hand (for example, a typo). kane-cli detects the bad ID, clears it, and resolves a working default rather than letting the run proceed and silently failing the upload at the end. To rebind explicitly, run kane-cli config project (or kane-cli projects list followed by kane-cli config project <id>).
After a successful upload, kane-cli prints two links to the terminal (see Share links at session exit). Both lead into the TestmuAI Test Manager UI for the test case that was just created. From there you can browse the project and folder you configured (or that were auto-defaulted), view the run summary, and dig into individual steps and screenshots.
Code export converts a completed test case into runnable Playwright code that you can check into your repository or extend by hand.
Currently the only supported language is Python with Playwright. A JavaScript backend is wired into the configuration shape but intentionally disabled for now. Setting any other language is rejected.
Code export runs server-side after the test case is finalised. kane-cli polls Test Manager until generation is complete, then downloads the resulting files.
Code export is off by default. Turn it on either in your stored config or per-run on the CLI.
In ~/.testmuai/kaneai/tui-config.json, set the code_export block:
{
"code_export": {
"enabled": true,
"language": "python",
"skip_validation": true
}
}enabled controls whether code export runs at all. language must be python. skip_validation controls whether the worker-side validator is skipped after generation.
For one-off CLI runs, you can override the config with flags:
kane-cli run "Add an item to the cart" \
--code-export \
--code-language python \
--skip-code-validation--no-skip-code-validation forces validation on for that run. --code-language only accepts python.
Generated code is downloaded into a local directory under your session, by default ~/.testmuai/kaneai/sessions/<session-id>/code-export/. At session exit kane-cli prints a CodeExport line in the links box pointing to that directory; the link uses your terminal's hyperlink support to open it.
The same code is also available alongside the test case in Test Manager.
When the upload finishes successfully, kane-cli prints a small links block to the terminal. You see up to three links, depending on what was generated:
- ShareLink — a public-style URL into the test case dashboard that includes a short-lived share ID. Anyone you send this link to can view the run summary in TMS without needing access to the project. Links are issued with a seven-day expiry.
- TestCase — a direct URL to the test case dashboard inside your Test Manager project. Use this when you are signed in to TestmuAI and want to drill into the run.
- CodeExport — a
file://link to the directory on your machine that holds the generated Playwright code. Only shown when code export is enabled and produced files.
The links survive the terminal exit and remain in your scrollback, so you can come back to them after kane-cli has quit.
After a session uploads successfully, kane-cli shows a small thumbs-up / thumbs-down prompt at the bottom of the terminal. Use the left and right arrow keys to switch between the two options, Enter to submit, or Esc to skip. Skipping sends nothing.
If you submit, kane-cli posts your choice (positive or negative) for the test case that was just finalised. The prompt only appears when there is a test case to rate — sessions that had no successful upload skip it. Both the TUI and the CLI render the same prompt; in the CLI it appears on stderr after the links block.
You can also submit feedback after the fact for a known test case:
kane-cli feedback --test-id <test-id> --feedback-type positive
kane-cli feedback --test-id <test-id> --feedback-type negative --details "..."Every session, regardless of upload outcome, leaves a directory on your machine you can inspect later:
~/.testmuai/kaneai/sessions/<session-id>/
├── session.json # Metadata: started_at, ended_at, model, profile,
│ # test_id, testcase_id, upload_status, run summaries
├── tui.log # Append-only event log for the TUI session
├── runs/<n>/ # One directory per run in the session
│ └── run-test/
│ └── actions.ndjson # Step-by-step record of what the agent did
└── code-export/ # (when code export is enabled) Generated code files
<session-id> is a UUID generated when the session starts. The runner writes its step logs and screenshots into runs/<n>/, where n increments per run within the session.
This directory is useful when you want to look back at a past run without going to Test Manager, debug a session that failed to upload, or hand a teammate the raw artefacts.