Reference architectures

Where Vericto sits in your stack, which mode to use, what it protects, what it costs in latency and how to start. One page per architecture, with its diagram, its failure behaviour and what each plan includes.

The modes, at a glance

They all evaluate with the same engine and your workspace's rules. They combine: most teams use the proxy at runtime and the CLI in CI.

ModeWhere it evaluatesWhat happens to a blocked statement
TCP proxyIn your infrastructure, between the application and PostgreSQL or MySQLIt never reaches the database: the driver gets an error
Runtime APIIn Vericto Cloud, before your code runs itYour code does not run it (a 403 answer)
CLI and GitHub ActionIn Vericto Cloud, from your CI pipelineThe job fails; nothing runs
MCPIn Vericto Cloud, while your coding agent writesThe agent gets the verdict and the fix
Python and Node SDKs (coming soon)In your code, through the Runtime APIComing soon

Your architecture is not here?

Tell us about your stack and we will tell you where Vericto goes. Or start free with the proxy, the API, the CLI or MCP.