Sign up

CScale 1. Our own model, built to check tests.

Your tests missed a change. Was it harmless, or did they miss a mistake? CScale 1 helps tell the difference. Our own model answers specific questions from the test check. It uses excerpts and clears them after processing.

Check your next change See how it helps

Harmless editexample

function canPay(sum, cap) {
-  return sum <= cap;
+  return cap >= sum;
}

Tests still pass.

CScale 1: same result.
The comparison is reversed.
It allows the same amounts.

No missing case in this edit.

Missed mistakeexample

function canPay(sum, cap) {
-  return sum <= cap;
+  return sum < cap;
}

Tests still pass.

CScale 1: the result changes.
100 is rejected at its limit.
Tests only cover 90 and 110.

Missing case: amount = limit.

Five questions. Each has a job.

Illustrative examples

Five questions, with a different example for each

Yes
  1. 1
    A missed limit case

    <= becomes <. An amount exactly at the limit is now rejected, but the tests still pass.

    Gap
  2. 2
    A new file, limits_spec.ex

    Its contents define tests. The check treats it as a test file, even when its filename is unfamiliar.

    Test
  3. 3
    An exact check becomes loose

    The expected total was 1005. Now any value greater than zero passes. Incorrect totals can slip through.

    Weaker
  4. 4
    A missing import breaks the build

    The code never reached the test. This failed run does not show that a test caught the mistake.

    Build
  5. 5
    What to run first

    Try the amount exactly at the limit first. The prediction chooses the order; a test run still has to check it.

    First

Run the tests. Make sense of the result.

A test run shows what happened. CScale 1 helps answer what it means, such as whether a missed edit changes behaviour. Both appear in the report.

Actual tests still run

Commitscale runs your tests on your machine or in your CI. Model answers help interpret the checks; they don't prove the whole change is correct.

Uncertainty stays visible

“Does this edit change the result?” → “Uncertain.” The finding stays open. The agent can fix it or accept it with a recorded reason, subject to your settings.

The judgment stays visible

Each answer and its confidence appear in the report and signed result. Confidence is an estimate, so you can inspect it alongside what actually ran.

Send the question. Keep the repository.

Your repository stays with you. Test runs stay on your machine or in your CI. Only excerpts needed for the questions are sent.

Your machine or CI

The repository and test runs stay here. Commitscale prepares only the excerpt needed for a question.

An excerpt goes to CScale 1

Detected secrets are removed. Our hosting partner clears requests within five minutes. See each request with:
--show-requests

The answer comes back

The answer joins your local report. We keep IDs, hashes and scores. Our hosting partner does not log request contents or use them for training. Sharing on Free is optional and off by default.

Find out what your tests miss

Start with one repository and your agent's next change. Sign up and install the CLI. The first 100 checks each month are free.

curl -fsSL https://commitscale.com/cli/install.sh | sh

On Windows, install in WSL with the macOS and Linux command.

curl -fsSL https://commitscale.com/cli/install.sh | sh

Native Windows, not yet validated

irm https://commitscale.com/cli/install.ps1 | iex

Then, in your repository:

cscale init

Ask your agent to check its next committed change. Agents can sign up from the terminal instead, with cscale signup.

Sign up Already have an account? Sign in