All checks pass. Both required outputs are written and verified.

What I delivered

artifacts/research-report.md (61 KB) and artifacts/prototype-specification.md (28 KB), plus an artifacts/README.md stating the question and limits.

The core finding

The mandate comparison turned out to be a second-order question. At $100,000 with the 3-month bill at ~4.1%, the entire risk-free income is ~$4,100/year — so the treasury cannot fund development from income under any mandate. The real variable is burn rate, not yield. Three consequences, each quantified:

• The income mandate isn't paid for right now. Aave v3 USDC (3.57%, 2026-09-15) and the Sky Savings Rate (3.60%) both sit below the 1-month (3.99%) and 3-month (~4.10%) Treasury par yields. The DAO would take Euler-class risk for a negative ~45 bp spread, buying about eight days of extra runway over four years.
• Growth's downside is measured; its upside is assumed. BTC −74.4% and ETH −81.8% (Nov 2021→Jun 2022) are observations; ETH is still 44.0% below its peak. I included a deliberately self-undermining table showing the same assets return either ~T-bill-equivalent or +45%/yr depending only on start date.
• Eligibility binds, not portfolio theory. BUIDL's $5M minimum and qualified-purchaser gate categorically exclude this treasury. Gas is negligible ($0.004/transfer); the cost that matters is ~60 bp/yr of product fee drag — a 25× ranking error waiting for any agent that optimizes gas.

Verification performed

• All 33 cited URLs checked programmatically: 28 × HTTP 200, 5 × 403 bot-blocks. I disclosed this in §11 rather than hiding it, and specifically flagged that congress.gov could not be read directly — so I corroborated the GENIUS Act claims against Latham & Watkins, which I did read.
• Downloaded and text-extracted the 46-page Wyoming SF0050 PDF rather than relying on summaries. This caught a conflict between two secondary sources and let me quote the statute directly. It also surfaced something the summaries missed: dropping below 100 members is a statutory dissolution trigger, which I added as a human-review event in both documents.
• Arithmetic independently recomputed, with the code in Appendix A so a reader can rerun it.
• Structure validated: heading hierarchy, balanced code fences, non-ragged tables, and all 36 internal § cross-references resolving. No placeholder markers.

Two things I want to flag rather than bury:

A source discrepancy I did not resolve. The Treasury par curve and Fed H.15 agree exactly on five of six maturities for 2026-09-22 but differ by 7 bp on the 3-month (4.09% vs 4.16%). Most likely an extraction error on one page, but I couldn't confirm it. It changes no conclusion.

The question that blocks everything (§9.1). The brief specifies a "$100,000 treasury" in dollars, but the only IMD treasury mechanism I could actually observe sweeps unclaimed launch tokens to a treasury address after a 365-day sweepDelay. If the treasury is native-token-denominated, every recommendation changes and the first action is diversification, not allocation. I wrote §4–§7 for the literal dollar reading and flagged the dependency explicitly rather than assuming it away — it's the first thing the founders should answer.

Scope note: research only. No capital, no transactions, no keys — and the prototype spec's Phase 1 is deliberately keyless, with live-custody work marked as requiring separate authorization.