CRESCO

Financial independence should not happen all at once. Most family investing products force a binary choice: either a guardian approves individual actions, or the young person receives broad access. CRESCO introduces a third model: progressive financial agency through standing, enforceable boundaries. A guardian creates a Key — technically a versioned Mandate — defining what the young person can do. Inside that Key, they can learn, practice and act independently without asking for permission every time. At the boundary, CRESCO refuses the action, explains why, and offers clear next steps. Learn in context. Act freely inside bounds. Ask for more freedom only at the boundary. How it works The canonical demo starts with a $10 per-action boundary.

1. Maya attempts $5 of AAPL → ALLOW. No guardian approval is required.
2. Maya attempts $12 of AAPL → REFUSE because it exceeds her standing Key.
3. She asks for more room.
4. Her guardian can decline, widen the standing Key, or Allow once for that exact $12 request.
5. If Maya changes the approved request from $12 to $11 → REFUSE.
6. Restoring the exact $12 request → ALLOW.
7. The one-time authorization becomes used.
8. The standing Key remains unchanged.
9. Replaying the authorization → REFUSE. The exception moved. The boundary did not. Why Solana CRESCO does not rely on the frontend to enforce family authority. The Solana program places the authority boundary in the same execution path that moves demo SPL-token capital. It proves:

• program-controlled demo-token execution;
• in-bounds execution without guardian approval;
• out-of-bounds refusal;
• versioned Mandates and nonce lineage;
• stale-authorization refusal;
• guardian widening;
• exact single-use exceptions;
• changed-action refusal;
• replay refusal;
• signed Pyth verification;
• Pyth-derived USD/notional enforcement. The UI is not the guard. The capital path is. Learning without gamifying authority CRESCO includes learning and practice because young people entering financial markets need context, not just controls. But learning progress never silently converts into financial authority. XP, P&L, badges, lessons and AI-generated guidance cannot widen the Key. Human authority remains explicit. Market truth Pyth provides live market evidence for the proven AAPL Money lane and broader Learn/Practice discovery. Market data can restrict or stop an action when evidence is stale or insufficient. Market data can never grant additional human authority. Fail closed, not fake success CRESCO treats negative outcomes as part of the product: ALLOW · REFUSE · PENDING · UNKNOWN If execution cannot be proven, CRESCO does not manufacture a success state. During development, real failures — including blocked Devnet funding, invalid execution input, changed one-time requests and replay attempts — were retained as evidence and converted into product safeguards. Real failure > fake success. What is proven today CRESCO currently demonstrates:
• a deployed Solana Devnet program;
• program-controlled demo SPL-token execution;
• live Pyth evidence in the AAPL execution path;
• guardian and child roles;
• persistent family state;
• boundary requests and guardian decisions;
• exact one-time permissions;
• replay and changed-action protection;
• confirmed Devnet receipts;
• source-backed Learn and Practice experiences;
• mobile/WebKit coverage. CRESCO deliberately does not claim brokerage, custody, mainnet execution, real AAPL ownership, or real minor securities execution. The idea CRESCO is not parental approval software. It is an attempt to create a new primitive for family finance: standing authority that gives a young person meaningful independence without pretending that independence has to arrive all at once.