A Chrome side-panel extension for executing Testomat.io manual test runs next to the site being tested. You read a test’s steps, tick them off, set passed / failed / skipped with a comment, and attach the evidence — a screenshot, a console and network log — without leaving the tab under test.
It works against app.testomat.io and against self-hosted instances. It is not
in the Chrome Web Store; you install it yourself (see below).
Runs. Browse the project’s runs and run folders, or paste a run URL. A run opens as a checklist grouped by suite. Statuses, comments, tri-state steps synced to the server, parametrized example rows, custom statuses, assignee, priority, finish run, and the web runner’s keyboard shortcuts. Marking a test never navigates — you move on when you choose to. An open run re-reads itself about every 20 seconds, so a colleague’s change lands on its own; a status write that fails offline is queued locally and replayed when the connection returns.
Evidence. Rec records the console messages and network traffic of the
tab under test into a rolling in-memory window. Errors show inline in the test
view, and setting a test to Failed uploads a readable .txt log and links it on
the result. Attach screenshot captures the tab — viewport or full page —
and opens an annotator over the page itself: arrow, box, pixelate and text,
each selectable afterwards to move or delete. The pixelate tool is a real
mosaic, and the un-pixelated original is dropped on Apply.
Tests. Browse suites and test cases; an existing test opens as a rendered read-only view. New test cases are written in a Markdown editor (with suites and folders), and a step recorder turns your clicks on the page under test into human-readable Markdown steps, masking values it recognises as sensitive.
Graceful degradation. The panel upgrades your API token to a web session automatically. When it cannot, it drops to basic mode: statuses, comments and the run list keep working over the public v2 API, while the session-only features are disabled with a stated reason rather than vanishing.
It is not in the Chrome Web Store, so there is no auto-update either way: a new
version is a zip you unpack, or a git pull, followed by the reload button.
Take a release if you just want to run it, and the clone if you intend to follow
main or send patches.
Either download the newest zip from Releases and unpack it, or clone the repository:
git clone https://github.com/testomatio/browser-extension.git
chrome://extensions and turn on Developer mode (top right).extension/ folder inside it, not the repository root.https:// URL into Instance under
Advanced before saving.Chrome 123 or newer. To update: unpack a newer zip over the same folder, or
git pull, then press the reload (↻) icon on the extension’s card in
chrome://extensions. Your settings and the offline queue survive the reload.
There is no build step: extension/ runs exactly as it is checked in. No npm
install, no bundler, no compiler.
These are the permissions declared in extension/manifest.json. Chrome
summarises the host permission at install as “Read and change all your data on
websites you visit”.
| Permission | Why |
|---|---|
storage |
Keeps the API token, the project choice, your preferences and the offline queue in your local Chrome profile. chrome.storage.sync is never used, so nothing is copied to your Google account |
sidePanel |
Draws the panel in Chrome’s side panel |
scripting |
Injects, on your click, the screenshot annotator, the step recorder and the console/network instrumentation into the tab under test |
webRequest |
Lists the page’s own network traffic (method, URL, status, timing) in the recording. Observational listeners only — no headers requested, never the blocking form, so nothing can be modified, redirected or cancelled |
debugger |
Full-page screenshots only, via the DevTools protocol — the only way to capture a whole scrollable document. Attached for that one shot and detached immediately, which is when Chrome’s “…is debugging this browser” bar appears. Viewport screenshots do not use it |
host_permissions: <all_urls> |
The site under test is different for every user and session, and screenshotting, recording and instrumenting it all require reading that page. Nothing runs on a page without your click. You can narrow this in chrome://extensions → Details → Site access at any time |
The tabs permission is not requested: the extension never enumerates your
open tabs, and cannot read your browsing history.
Full detail — what is stored, what is transmitted, and every off switch — is in PRIVACY.md.
| Using it | docs/user-guide.md — every screen and button, the limits and the quirks |
| Privacy | PRIVACY.md — what is sent where and when, every off switch, what it cannot do |
| Working on the code | docs/architecture.md — module map, data flows, the permission model, storage keys, known traps |
The end-to-end suite is not part of this tree: it drives a live Testomat.io
account with fixtures that only exist there, so it is unrunnable from a clone and
is maintained separately. Two consequences. Pull requests are verified by
maintainers by hand, so expect review to take longer than the diff suggests. And
that suite drives a few test seams in the shipped code (a couple of
chrome.storage.session keys, one message with no production sender) which look
unused from inside this tree — they are documented in
docs/architecture.md §5.3; please leave them alone.
Bugs and feature requests belong here, in this repository’s Issues. A useful report says what you clicked, what happened (with the exact wording of any message), your Chrome version, and whether the Basic mode pill was showing. Please do not paste API tokens or run URLs you would rather keep private.
MIT — see LICENSE.
Third-party code is vendored under extension/vendor/ as committed single files
and keeps its own licence: showdown (MIT)
and OverType (MIT). The icon paths come
from Material Symbols (Apache-2.0), and
JetBrains Mono ships under the SIL Open Font License (extension/shared/fonts/OFL.txt).