# Getting started

> Ask your agent to add one test, then review what changed.

Web page: https://supercov.com/docs/getting-started
All Supercov pages for agents: https://supercov.com/llms.txt · Full text: https://supercov.com/llms-full.txt

Supercov scores your source, checks for security risks and measures your existing
tests. Your coding agent uses the results to find what to refactor and what to
test, makes the change and checks what improved. You keep using the same agent
and test suite.

## Start with a score

Open your project in your coding agent and paste this:

```text supercov-prompt
Measure code quality with npx supercov.
```

You get a score for every source file, and the named properties behind it, so
you can see which files your agent should look at first. Scoring needs a
[TypeSafe AI](https://typesafe.ai) API key, and your agent will usually ask
you for it. [Code quality](https://supercov.com/docs/quality.md) covers the command, and
[How Jev scores](https://supercov.com/docs/jev.md) covers what the number is made of.

## Then add one test

Coverage needs no key and no account. Paste this instead:

```text supercov-prompt
Measure code coverage with npx supercov and write one missing test.
Only change tests. Rerun the full test suite and show me the test you
added and the before-and-after coverage.
```

Your agent can install Supercov if needed. If your project has several test
commands, tell it which suite to use. Approve commands and file edits as you
normally would.

## What you get back

The agent adds or updates a normal test file in your repository. In the
conversation, it should show the change, whether the suite passed and how
coverage changed. Supercov keeps the run reports locally in `.supercov/`.

There is no separate hosted project to open. A commit or pull request is not
automatic; ask your agent for one if you want it.
If your team already reviews coverage in Codecov, Supercov can export LCOV or
Cobertura for it; see [Codecov vs Supercov](https://supercov.com/compare/codecov.md).

Review what the new test checks. Does its expected result make sense? Would
it fail if that behavior broke? A higher percentage alone doesn't answer that.
[Reading the results](https://supercov.com/docs/verification.md) explains what to look for.

To review saved evidence in a browser, run `npx supercov report`. It creates a
single HTML file with coverage, assertion links, quality and security together.
See [Interactive report](https://supercov.com/docs/reports.md) for what each part means.

## Before you start

Your agent needs access to the project and a working test command. Install the
project's dependencies and make sure any test databases or services are available.
Supercov doesn't replace that setup.

Check [Supported languages](https://supercov.com/docs/supported-languages.md) for your runtime and runner.
If you'd prefer to install Supercov yourself, follow [Installation](https://supercov.com/docs/installation.md).
You don't need to configure a new coverage reporter to try Supercov.

## After the first test

Ask the agent to keep working on an area you care about, to inspect what your
existing assertions actually check, or to review what a change introduced
before it lands. See [Working with your agent](https://supercov.com/docs/agent-loop.md),
[Assertion coverage](https://supercov.com/docs/assertions.md) and
[Reviewing a change](https://supercov.com/docs/quality-changes.md). Ask for a [security review](https://supercov.com/docs/security.md)
when you need to inspect risks in the source or a patch.

If you prefer running commands yourself, the [CLI reference](https://supercov.com/docs/cli.md) has them.
For instructions your agent can follow directly, use [Agent instructions](https://supercov.com/docs/agents.md).