PDD Kit brings formal verification to agent-written web apps

PDD Kit checks whether AI-generated code does what a developer intended, addressing the difficulty of reviewing agent-written software line by line. It converts developer intent into a formal Verus specification, then proves that the generated code meets that specification for every input. The approach shifts the developer’s review toward intended behavior rather than inspection of every line of the implementation.

Where formal proofs are not practical, PDD Kit uses tests to check the code against the specifications. Its workflow combines formal verification with testing, using proofs where possible and tests elsewhere. For developers working with agents that generate code at a scale that is difficult to review manually, the focus is on specifying what the software should do and checking that the generated implementation follows it.