safe-programs

Brainstorm, write, or audit Solana programs with security baked in. Triggers on: exploring design ("thinking about building X", "should I use Anchor or Pinocchio for Y") → brainstorm mode; writing or scaffolding a new program from scratch → build mode; reviewing existing Solana program code for security issues → audit mode. Covers Anchor, Native Rust, and Pinocchio. Enforces rules derived from real Solana audit findings: account validation, PDA safety, CPI trust boundaries, arithmetic, Token-2022, reward accounting, lifecycle integrity, slippage/fee ordering, bonding-curve safety, admin key rotation, BPF runtime limits, and metadata hygiene. Always use this skill when Solana program code is the deliverable or when designing one.