Focused testing for AI-written code

Test every change.
Only where it matters.

When Claude changes a function, CodeMD follows its call graph to find the existing tests that matter—or creates the missing tests—then runs that focused set before and after commit.

Start with the CodeMD VS Code extension.

1

Claude changes code

CodeMD detects each modified or newly created function as you work.

2

CodeMD maps the impact

The call graph reveals callers, dependencies, and the behavior that could be affected.

3

Relevant tests are selected

Existing tests are found first. When coverage is missing, CodeMD creates an appropriate test.

4

The focused tests run

Run them pre-commit in VS Code and post-commit locally or with supported online coding-agent APIs.

Why focused testing

Confidence without running everything.

CodeMD uses the actual change and its call-graph relationships to keep the test scope relevant. You get fast feedback while protecting the behavior around the changed function.

✓

Coverage follows the change

Every modified and new function gets attention as soon as Claude touches it, helping prevent coverage gaps from becoming future defects.

◎

Reuse before creating

CodeMD looks for the most relevant existing test first. It creates a new test only when the change needs coverage that is not already there.

↗

Test the interaction, too

Testing does not stop at one function. Call-graph context helps verify how the change interacts with callers and connected behavior.

The right test for the risk

More than unit tests.

CodeMD can create the form of test that best matches the changed behavior and its place in the system.

Unit

Checks the function's behavior in isolation.

Contract

Protects expectations between components.

Integration

Verifies connected functions work together.

Regression

Preserves behavior a change could break.

Risk-based

Targets the highest-impact call paths first.

Before and after commit

Feedback where you already work.

Use the CodeMD VS Code extension to run relevant tests before the code is committed. After commit, run the same focused approach locally or through supported online coding-agent APIs.

This gives developers immediate feedback during implementation and fresh verification after the change enters repository history.

Install the free extension
Change verificationComplete
Changed functioncalculateRisk()
Relevant existing tests4 found
Missing coverage1 test created
Call paths checked7
Result5 tests passed

Let every Claude change arrive with evidence.

Install CodeMD for VS Code and try focused, call-graph-aware testing for free.