How Kerno Works
Understand how Kerno works, from indexing your codebase to validating code changes.
Kerno is runtime code and security review. It builds a deep understanding of your codebase and captures how your code operates. Every time you change code, it checks the change against that baseline to catch regressions, integration issues, and security gaps.
Your agent drives Kerno over MCP. It runs inside your agent's loop, so your agent validates what it just wrote, reads the results, and fixes what broke in the same session.
Index your codebase
Kerno analyzes your code and builds a graph of every function, class, model, and endpoint, and how they connect. The graph is stored locally and gives Kerno the exact blast radius of any code change, so it knows how a change affects the rest of your system. See Codebase indexing.
Connect Kerno to your running app
Connect Kerno to your running app, local or remote, so it can validate your code changes against your repo's real services, dependencies, and framework.. See Environment setup.
Generate baseline tests
Kerno generates tests covering your endpoint's functional behavior, edge cases, and security, then runs them against your real app to capture a baseline, a snapshot of your endpoint's current behavior. See Scenarios and baselines and Security testing.
Review code changes
When you change code, Kerno maps its blast radius and re-runs the affected endpoints' tests against your app, comparing the results to the baseline. It shows you exactly what your change altered, so you can tell whether each difference is intended or a bug. See Change validation.
Update tests
Kerno's memory learns from every interaction with your team, so its testing gets more tailored to your codebase over time. Custom rules capture your team's best practices and apply them consistently across every endpoint Kerno tests. See Custom rules and Memory and learning.
Last updated