TLDR: from risk overlays to onchain treasuries
DeFi spent the last cycle trying to professionalise risk through “curators” and managed vaults. On the surface, that looked like progress. In practice, the major blow ups of the past months have exposed the same underlying pattern: the main risk engine sits outside the rail that actually holds user assets.

---

The essay breaks this down into two architectures:

• the curator overlay model, where risk lives in PDFs, forums, and dashboards and execution depends on DAO votes and multisigs
• a full-stack onchain treasury, where vault infra, risk policy, and execution are part of one system, and curators become signals rather than single points of failure

Using recent incidents as examples, I argue that most failures are not just “bad actors” or isolated mistakes, but structural gaps:

• no single, machine-enforced source of truth for risk limits
• execution authority misaligned with the speed at which risk moves
• vault contracts treated as implementation details rather than primary, composable objects

The piece then sketches a simple design framework for what I call a full-stack, agentic onchain treasury layer: risk rules in code, bounded execution rights that match the speed of risk, vault tokens designed as neutral money legos, and curator work feeding in as upstream signal instead of operational control.

If you run a treasury, a fund, or a serious personal DeFi book, the core question I end with is:

What is our full-stack onchain asset management layer, and does it actually deserve the risk and capital we are placing on it?

Full article on X:
🔗 beyond risk curators - why DeFi needs full-stack onchain treasuries