Before deployment, Nyvo uses a sandbox environment for compiling and testing code. This part of the architecture includes:
Solidity/Rust Compilers
We have integrated compilers for various languages (the Solana Rust toolchain, the Solidity compiler, etc.) in a secure sandbox so that generated code can be immediately compiled to bytecode. Any errors from compilation are fed back to the AI for correction.
Test Runner
A system that can deploy the compiled contract to a local ephemeral blockchain (like a local Solana validator or an Ethereum Ganache instance) and run a series of test transactions. This allows Nyvo to verify that functions execute without runtime errors. It’s akin to a CI (Continuous Integration) pipeline automatically running whenever the AI produces new code.
Security Analyzer
We have integrated static analysis tools (for example, linters or analyzers like Slither for Solidity) that scan the code for vulnerabilities or anti-patterns. Results of these are also considered by the AI Orchestrator. If a potential issue (like a reentrancy vulnerability or an uninitialized variable) is detected, Nyvo will adjust the code or alert the user depending on severity.
