Tokenized funds and treasuries
Reconcile on-chain supply against a fund's filed NAV, or name the auditor when a regulator is not in the loop.
OPEN SOURCE / MCP / CLI / HTTP
Every tokenized yield makes the claim. On‑chain, all you see is a token and a number. This reads the filings, checks the arithmetic, and shows you exactly where the proof stops.
Implies a proof that, most of the time, was never actually published anywhere you can read.
Punishes an asset for being private, not for being fraudulent. Unproven is not the same as unsafe.
Most tools skip backing altogether and just show you a big number: the yield.
None of them tell you the one thing that matters: where the proof actually stops.
A tokenized asset is a stack of trust: token, legal wrapper, underlying reserve.
On-chain reads only verify Layer 1. SEC filings and auditor attestations reconcile Layer 3 reserves against Layer 1 supply.
No model votes on the answer. We read the on‑chain supply, pull the SEC filing, quote the net asset value word for word, and multiply. If the two numbers don’t meet, there is no green.
SCROLL TO RUN THE CHECK
Both are tokenized Treasuries. Only one has a public filing to reconcile against. One resolves into proof. The other stays static — and we say so out loud.
Holdings are filed with the SEC (N-MFP). On-chain supply reconciles against that filing, so the green here is arithmetic, not a vibe.
Inspect BENJIReserves sit with a custodian, not published on-chain or in a public filing we can read. A higher APY doesn't change that — the answer stays unknown.
Inspect OUSGThe same AgentVerdict everywhere. Gate on backing.tier, surface trust_boundary and caveats. No API key.
Add the stdio MCP server to Cursor, Claude Desktop, or any MCP host. Two tools: check_asset_backing and list_verified_assets. Returns structured AgentVerdict JSON - no boolean safe flag.
{
"mcpServers": {
"rwa-backing-verifier": {
"command": "npx",
"args": ["tsx", "mcp/server.ts"],
"cwd": "/absolute/path/to/rwa-analyzer",
"env": { "RWA_API_BASE": "https://rwa-analyzer.vercel.app" }
}
}
}Cursor → Settings → MCP → paste the JSON above (or clone the repo and point cwd at your checkout).
Or clone and run locally: https://github.com/Archdiner/rwa-analyzer
git clone https://github.com/Archdiner/rwa-analyzer
cd rwa-analyzer && npm install
npm run mcp # stdio MCP server
npm run verify -- BENJISet your location and size. Assets are sorted by how far the backing proof goes — verified first, higher APY with weaker proof further down.
Verified backing sorts above higher APY you would have to take on trust.
Confirms a redeemable share of a regulator-verified fund. Does not reconcile on-chain float against the fund share register; that linkage is the transfer agent's record.
Minimum investment not confirmed.
Minimum investment not confirmed.
Minimum investment not confirmed.
OR LOOK ONE UP
Backing was the hardest case, so it came first. Coverage grows where proof is possible.
Reconcile on-chain supply against a fund's filed NAV, or name the auditor when a regulator is not in the loop.
Read the reserve on-chain. Split organic borrow interest from reward emissions, then grade utilization and oracle risk.
Pooled ETH against token supply is arithmetic. Forward risk is slashing and validator exposure - named, not hidden.
The part of a headline APY that ends when the program does. Separated from real yield, stamped with an end date.
This is open. Every suggestion gets read, embedded, and clustered by meaning — and the clusters become the directions the product actually moves. Nothing is too big.
DROP A SUGGESTION
EMERGENT DIRECTIONS
Ranked by backing count · synthesized from clustered suggestions
§ THE CLOSE
Open source. Deterministic reconciliation. One call before you route a deposit, and an honest answer when the proof runs out.