GitTested

How we test.

We put open-source tools through the jobs they promise to do, record what happens, and explain what the results mean for you.

The three steps

  1. A pigeon leans over three task cards, nudging an orange marker onto one with its beak.

    Choose the task

    We choose a task that reflects the tool’s purpose, with realistic inputs and a clear way to check the result.

  2. A pigeon stretches a wing for balance while pressing a pedal and watching a test machine produce a paper strip.

    Run the tool

    An AI agent tests a specific version on an isolated machine. We exercise the main workflow, then try a different input or a failure case. We keep the commands and outputs.

  3. A pigeon in an ochre waistcoat writes in a notebook with a pencil held in its beak, beside the test output.

    Review the findings

    We check the results alongside the project’s documentation, maintenance, and alternatives. An editor reviews the evidence and approves the final article.

Evidence and limits

Each review identifies the version we tested and explains what worked, what failed, and what we couldn’t verify. The findings apply to that version and the tasks we ran.

Reviews include relevant output and screenshots. Where an evidence download is available, it contains the supporting commands, outputs, and logs, with private details removed.

What our verdicts mean

These are recommendations for the use case in each review, not scores.

Try
It did the job in our test. Try it for the use case the review describes.
Adopt
It works, with a specific cost or limitation to account for first.
Wait
The idea is useful, but the tested commit needs more work.
Avoid
The tested commit does not deliver the job it advertises.
Read the reviews