Adversarial pre-ship interrogation of a Solana program's design and security. Walks the design decision-tree branch-by-branch, one question at a time, with a recommended answer for each, grounding every question in the user's actual Anchor / native code when it exists. Use BEFORE an audit — when a user says "review my program design", "roast my program", "is my Solana program safe", "design review", "threat model my program", "pre-audit", "am I missing signer checks", "check my PDA design", "before I deploy to mainnet", or pastes an Anchor program / `#[derive(Accounts)]` struct and asks whether the design is sound. Outputs design-spec.md, threat-model.md, and a pre-audit checklist that downstream audit/scanner skills consume.
2026-06-30