One flat price a month. No seat fees.

Basic starts with a 14-day trial on the personal organization you sign up with, and checkout asks for a card. Go, Pro and Max are billed from day one. During the trial, scans run at the Quick and Focused levels.

Each plan includes usage equal to its monthly price: $20 on Basic, $100 on Go, $300 on Pro and $1,000 on Max. The portal shows how much you have used as a percentage, with the renewal date. Unused usage does not roll over.

Scans beyond the included usage can run on demand from your organization's prepaid balance. You add to it with one-time top-ups of $10, $25, $50, $100 or $250, and there is no metered overage.

Yes: an upgrade charges the prorated difference at once, and a downgrade applies at the next renewal, with no refund. Upgrading during the trial ends it and charges at once. Downgrading during the trial is not available.

No. Every plan is a flat monthly price with no seat fees, and there are no yearly plans. Basic includes 3 members, and Go, Pro and Max include unlimited members.

Pro, Max and Enterprise. Model tiers set how careful the AI review is: Basic and Go include Fast and Balanced, and Pro adds Best, the most careful, along with Adversary scans for attack paths.

Yes, on every plan, including during the trial. MCP calls themselves are not billed, but a re-check requested through MCP counts against your organization's scan usage like any other.

No. Prices are shown before tax, and VAT or sales tax is added at checkout.

Kelrova Studio, the desktop app for AI coding agents, is free and needs no plan: the app runs on macOS, and its server package on Linux. Its agents run on your own provider accounts, and Kelrova does not charge for Studio.

AI models read the repository on the branch you picked, as far as the level you choose: Quick reads the files changed since the last completed scan, and Focused, Standard, Extended and Full widen step by step, up to every entry point, application and config file. Paths you exclude are neither downloaded nor scanned. Each finding names the file and line and shows the code behind it.

Kelrova is an alternative to rule-based SAST, not another rule set: AI models read code across files the way a reviewer does, so a check in one place and a gap in another show up together. Scans report vulnerabilities such as injection, broken access control (IDOR), authentication bypass, SSRF and secrets in code, each with its evidence. A scan is a reviewer, not a guarantee: a clean result means nothing was found at the level you chose.

Each scan downloads the branch you picked, without the paths you excluded, into its own isolated environment, and that copy is deleted when the scan ends; findings keep file paths, line references and short code excerpts. Scans only read your code, Kelrova changes code only through a pull request a member asks Go Fix to open, and on GitLab and Bitbucket linking a repository also adds a webhook so pull request status stays current. Third-party model providers perform the analysis and receive the code a scan reads; under their API terms they do not train on it and keep it at most 30 days to monitor abuse, unless the law requires longer.