Pay for runs, not seats

Security tooling is usually priced per seat, per repository, or per finding you are allowed to see. We tried each of those on paper and none of them described what the product actually does, so we priced the thing it does: a run.

Credits

A run costs a whole number of credits, and one credit is one dollar. Every workflow shows its price per profile before you start — the registry lists it, the pre-run screen confirms it for your repository, and the invoice matches it. If you change the profile, the price changes on the screen before you commit to it.

Credits are bought in packs, and packs never expire. A bounty hunter who runs twice a month and a team that runs every day buy the same credits at the same price per run; the team just buys more of them. Any member of a workspace can spend the workspace's credits.

Plans

Plans exist for people who run every week and would rather not think about packs. Pro and Team include a monthly allowance of credits at a discount, and add the things a regular user wants: more concurrent runs, a priority queue, longer retention, and — on Team — a shared workspace with roles, private forks of workflows, and notifications where your team already reads them.

Plan credits reset at the end of each month; pack credits do not. If you are not sure which you are, start with a pack.

Enterprise is a conversation. There is no feature list to compare against, because the things enterprises ask for tend to be specific to them.

No free plan

There is no free tier, and that is deliberate. A free plan for a product that spends real model budget on every run means either a crippled run or a business that cannot pay for its own runs, and we did not want to ship either. Instead, every new account gets a starter grant that covers a first run, and the parts of Midkernel that cost nothing to give away are free to everyone: the registry, the bench, the methodology, and the open-source runner that executes any workflow on your own machine.

Maintainers of public repositories are the exception. Open-source projects run free on a monthly credit grant; apply with a link to the repository.

What you are not charged for

Runs that fail before analysis begins — a repository that cannot be cloned, a sandbox that does not start — are not charged. Runs that stop early because they hit their profile's cap are charged for what ran, and the invoice line says so. Spend caps exist per workspace and per run; a run that would exceed a cap does not start.

There are no seats, no per-repository fees, no annual discount (there is nothing to prepay), and no severity you have to upgrade to see. A finding is a finding.

Why publish all this

Because pricing that has to be explained by a salesperson is pricing that is hiding something. Every number on the pricing page is the number on the invoice, and the pre-run screen exists so that nobody is surprised between the two.