Security for the AI-coding era

Security for the AI-coding era.

Kelvex is a set of open-source tools that finds the holes in code written by AI — and by humans — before someone else does.

The problem

AI writes code faster than anyone can review it.

AI assistants now write a large share of new code. That code ships quickly — and so do its mistakes.

Most security tools were built for a slower era. They drown teams in noise, or they get skipped entirely.

Kelvex tools fit how AI-era teams actually work: fast, open, and part of the pipeline from day one.

Open by defaultEvery Kelvex tool is open source under the MIT license. No black boxes — read the code, audit it, extend it.

Built for AI-written codeOur checks target the mistakes AI assistants repeat: leaked secrets, injected prompts, unsafe generated dependencies.

Signal, not noiseEvery finding comes with a plain explanation and a fix. If we can't tell you what to do about it, we don't flag it.

Kelvex in numbers

GitHub stars

Contributors

Security checks shipped

MIT-licensed tools

Our builds

Static analysis for AI-generated code. Scans every diff and flags repeated vulnerability patterns before merge.

Dependency and secret watcher. Catches poisoned packages and leaked keys in CI, with a plain-language reason for every block.

Prompt-injection detector for agent pipelines. Reviews the tool calls an AI agent wants to make — before it makes them.

Latest update

The new confidence model cuts noise without hiding real findings. Here's what changed and how to upgrade.

Kelvex lives on GitHub. Star the project, open an issue, or ship your first pull request — every contribution makes the tools sharper.