Skip to content

QA test cases and test runs without a separate tool

A small team does not need a test management product. It needs a list of what must keep working, a way to mark each item pass or fail, and a link from a failure to a bug.

Written by the BugBoard team. 5 minute read. Updated 16 November 2026.

Why keep them together

When test cases live in another tool, they go stale. The bug is in one place, the test that found it in another, and nobody knows which tests to run after a change. Putting both in the tracker gives you one list to look at. It does not give you the depth of a dedicated test product, and later we say where that matters.

How BugBoard models it

  • A test area groups related tests, such as Sign in, Billing or Export. Areas show as columns on the Testing board of a project.
  • A test feature sits in an area and says what should keep working, for example “Log in with email”. It can carry one-line checks, short things to verify. Under the hood it is a task of type test, so it takes comments, an assignee and a priority.
  • A feature has one status: untested, verified, needs review or failing.
  • You can link a feature to the tasks that affect it. When a linked task is closed, a verified or failing feature moves to needs review, so green means “verified since it last changed”.

Testing is not behind a paid plan in the code: it works on Free. Sprints, initiatives and time tracking are on Starter and above, but the Testing board is not on that list.

Write test cases people can run

BugBoard keeps a name, a description and a list of one-line checks for each feature. It has no separate fields for steps and expected result, so put them in the description. Keep each case to one thing that can be true or false.

Area: <Sign in>
Feature: <Log in with email and password>
Priority: <high / medium / low>

Description:
Start: <logged out, on /login>
Steps:
1. <enter a known email and password>
2. <press Log in>
Expected: <you land on the dashboard within 3 seconds>

Checks (one line each):
- Wrong password shows an error, and does not clear the email
- The session survives a page refresh
- Log out returns to /login

Notes: <test account, data needed, known gaps>

Record a result

There are three results. Pass marks the feature verified. Fail marks it failing and files a bug with the area, the feature and its checks in the description, linked to the feature. If you already have a bug for it, name that bug and it is linked instead of a new one being filed. Re-review marks it as needing another look, without filing anything. A coding agent can do the same through MCP: list_test_areas, record_test_result and link_task_to_test_feature.

An example run

Before a release, one person walks the Sign in and Export areas. Anything that is not green is looked at first.

AreaFeatureResultWhat happened
Sign inLog in with emailPassVerified.
Sign inPassword resetFailA bug was filed and linked: the reset link opens a blank page.
Sign inLog outPassVerified.
ExportCSV of a filtered boardRe-reviewA linked task closed last week; check again.
ExportMarkdown exportPassVerified.

After the run: two verified, one failing with a bug, one needing review. When the reset bug is fixed and closed, the Password reset feature goes back to needs review, so someone confirms the fix before it is green again.

A routine

  1. Make an area for each part of the product a customer would name. Start with five to ten features in total, not fifty.
  2. Link each bug and change to the features it touches when you open it.
  3. Before each release, filter for anything that is not verified. Run those first.
  4. Record every result the same day. A result from last week is a guess.
  5. Add a feature when a bug escapes. A bug that got through shows a missing test.

Who runs it

Give each area an owner, even if it is the same person for all of them. The owner decides what belongs in the area, keeps its checks short and notices when a feature has been verified for too long. Rotate the person who walks the run before a release, so more than one person knows how the product behaves. A new starter walking the Sign in area is a good way to learn it, and to find the checks that are out of date.

What this does not do

  • It keeps the current status of each feature, not a history of runs. If you need a record of what passed in release 1.4, write it in a comment or in the release notes.
  • There are no test plans with named runs and no metrics across runs.
  • It does not execute tests. Automated tests belong in your CI, and their results are not read into BugBoard by default.