RWA SonarNew

Methodology · live operational status

How we check what we publish.

A token address, a legal right, a market price and a DeFi listing are different facts. RWA Sonar collects them separately, states what each source can prove, and marks a gap wherever the evidence runs out.

Four headlamp-wearing dolphins survey the submerged part of an iceberg, with two farther away.

Three rules we apply to every conclusion

01

Identity before analysis

We identify the exact Solana token address and issuer programme. A familiar ticker does not prove that a token is authentic, or that two wrappers give the same rights.

02

A technical power says nothing about legal permission

A freeze key proves that freezing is technically possible. It does not prove when the issuer may lawfully use it. Legal terms and on-chain state are recorded as separate layers.

03

Unknown stays unknown

Missing price, custody, redemption or source data is shown as unknown. It is never converted to zero or to “no risk”.

Live status

Collector health by domain

Open token health monitor →

Loading collector status…

Loading…

Checking the database API…

Freshness is measured from the timestamp of each completed observation. File modification times are ignored. Hourly, three-hourly and daily watchers publish their own completion state immediately. “Current, with failures” means the pass finished but one or more upstream reads or downstream writes did not.

What runs, and how often

The monitoring jobs

Each job runs on the server on its own schedule and records when it last completed, so a stalled job shows as stalled without affecting how fresh the others look. Times are UTC.

Hourly · :07

Chain watch

Reads every mint account we track, token metadata and labelled issuer wallets. It stores one row per observed state, never per read. A rotated authority, a toggled extension, a rebase, a supply move of 1% or more, a metadata change or a treasury move becomes a dated event with the Solana slot as evidence. The first sighting of a mint records a baseline and raises nothing.

Hourly

Trade collector

Decodes swaps on the busiest pools within a fixed RPC budget per pass. The trades page reads only our API; your browser never contacts a Solana node.

Hourly · :50

Watch digests

For watches whose owner connected @rwa_sonar_bot on Telegram, a private digest of material changes to that exact target at the hour the owner chose. We store only an encrypted chat id.

Daily · 02:41

Document watcher

Re-reads every URL a dossier cites, diffs it and re-checks each quote word for word. If a page refuses a script or renders nothing, it tries the publisher's own API for the same document, then the newest raw Wayback capture. Each source row says which reader produced the text. An archive.today copy is linked, never read.

Daily · 04:23

Case-law watcher

Searches CourtListener opinions and dockets and the SEC's litigation and administrative-proceeding feeds for every issuer's legal entities and key parties. A new case or docket entry is a lead for review. It never marks a what-if answer litigated; only a cited decision does.

Daily · 06:47

Change judge

A language model reads the ten newest unjudged document changes in one batch and says whether each alters what a holder owns, can do or can have done to them. See the policy below.

Daily · 23:05

Redemption observer

Scans new transactions at the issuers' redemption addresses and programmes since the last checkpoint. See what counts as observed below.

Every 6 h · :17

Public build

Rebuilds the token universe, cards and social preview images, the power map, flows and xStocks float (with a fresh inventory read), tracking, exits, the weekly summary and the review queue. Each published file keeps its own build time.

Weekly

This week in tokenized stocks

One generated page per ISO week, Monday to Sunday UTC, rebuilt with each public build while the week is in progress. Numbers are compared across weeks only when both weeks were observed.

Automation with a label

Model assessments

Shown next to the diff

Every model verdict is shown as a "model assessment" next to the exact change it read. It never decides whether a change is published, included in a digest or counted.

Must quote what it calls material

An answer whose quoted fragments are not word for word in the change, or that calls a change material without quoting it, is stored as invalid and never shown as text.

Bounded and costed

At most ten changes a day, sent as one batch at half price. Tokens and cost are recorded per item. A batch stuck for 12 hours is cancelled and that run judges directly. A legal conclusion still needs an analyst's review.

Redemptions

What counts as an observed redemption

A redemption right in a prospectus is documented. A route the issuer currently offers is operational. A completed redemption we read on chain is observed. Every card and issuer page shows the three separately.

Ondo Global Markets

A burn of the GM token by the Ondo GM programme that pays USDC or USDon to the same signer in the same transaction, directly or through a solver programme. Valued at the stablecoin paid. Creations are its mint-with-stablecoin subscriptions, priced the same way.

xStocks

A holder transfer to the issuer's redemption address, which the issuer sweeps into its treasury: the prospectus calls this de-activation. The USDC payout is linked to the deposit by inference within a 180-second window. xStocks creations are not collected yet.

Superstate

A burn at the published equity burn address: a conversion of the token back to book-entry shares at the transfer agent. No cash is paid.

Not observable

PreStocks and Tessera redemptions do not leave a recognisable on-chain trace; each page records why, and a small recurring check would notice if that changes. Other issuers have no recurring scan and their redemptions stay documented or unknown.

Coverage is shown per day

A day counts only over the hours actually read. An unread day is missing, a partly read day shows its hours, and a failed scan shows "scan failed", never "no redemptions".

Who is redeeming

Primary markets are open only to onboarded, allowlisted wallets, so these flows mostly come from authorised participants and market makers. End holders rarely redeem directly.

Float

How we measure xStocks public float

Redeemed xStocks are not burned; they return to issuer inventory and still count in on-chain supply. We define public float = mint supply − the balances of the issuer-attributed wallets, read in raw base units in one pass. The wallets are attributed from on-chain behaviour recorded, with citations, in the xStocks dossier; the chain cannot prove whom they belong to. Tokens in wallets we cannot attribute count as float, so the float is an upper bound. Since 24 Sep 2026 the issuer wallets include an inventory wallet that the issuer itself excludes from its circulating figure. Dollar values use the token universe's market price, and mints without a price are counted but not valued.

Discrepancies

Kinds of discrepancy and change

Issuer claim versus evidence

A current, source-backed conflict between what an issuer says and what another governing document or the chain shows: governing-document conflicts, implementation conflicts, terminology qualifications and asset-specific evidence gaps. Both sides keep their citations, and the conflict is shown near the top of affected pages.

Protocol docs versus chain

A lending or trading protocol whose documentation describes its keys or governance differently from what we decode on chain, for example Loopscale's upgrade threshold and its refinance admin. Each carries the condition under which it would be resolved.

Source changes

A document diff touching a claim or keyword, a document that disappeared, or a quote no longer found in its source. The claim resting on it is shown as provisional until reviewed.

On-chain changes

Authority rotations, extension toggles, rebases, supply moves, metadata and treasury moves, each with the slot and account read.

Litigation leads

New court or SEC matters naming an issuer's entities, queued for review.

Our own corrections

Corrections to RWA Sonar's own research are kept as internal audit history. The public pages show the best current analysis.

Process

The evidence pipeline

  1. DiscoverSearch issuer lists, token registries and known tickers for candidate addresses.
  2. AttributeMatch an address to an issuer using published addresses, metadata, programme APIs and corroborating chain evidence. Inferred attribution is labelled.
  3. ObserveRead token extensions, authorities, supply, holders, pools, trades, protocol accounts and source documents.
  4. NormalizeKeep market, control, legal/evidence and DeFi composability as four separate results, never combined into a weighted score. The headline splits the eleven checks by who they describe: the programme (legal evidence, authority keys, DeFi enforceability — the same for every token of an issuer) and the token itself (the other eight), with a plain count of the token checks it passes among those that could be judged. An unknown is never a pass or a fail, and a token is only called good on its own checks if at least one market check could be run. Organic flow cautions only when a few wallets make most trades (more than 25 per trader); a low organic share spread across many wallets is ordinary arbitrage. An issuer's scheduled session closure (Ondo's unavailable_in_session) is not a pause.
  5. CompareSnapshot observations, diff consecutive days and preserve the before/after evidence.
  6. ReviewMaterial document and control changes remain reviewable findings. Automation detects a change; a person draws the legal conclusion.

Health dimensions

What each health dimension means

Market

Price tracking, pool liquidity, organic activity, failed swaps, holder concentration and cross-venue disagreement.

Does not prove: legal ownership, backing, or that quoted liquidity can absorb a large liquidation.

Control

Mint, freeze, pause, clawback, transfer-fee, allowlist, permanent-delegate, transfer-hook and rebase powers, plus how keys are held.

Does not prove: that a power will be exercised lawfully, predictably or in the holder’s interest.

Legal & evidence

What the holder owns, who owes the obligation, governing law, insolvency position, custody, redemption, eligibility and document authority.

Does not prove: a court outcome or successful redemption where neither has been observed.

DeFi composability

Exact-mint protocol support, permitted operation, smart-contract custody, collateral parameters and whether seized collateral has a credible exit.

Does not prove: that generic Token-2022 compatibility equals a live listing or that protocol possession equals legal title.

Evidence

Evidence labels and precedence

Binding or official terms

The operative legal source for the contractual claim, subject to mandatory law and later product-specific documents.

On-chain observation

What the deployed token or protocol account can do at the observed slot. Strong evidence of what is technically possible. It says nothing about legal effect.

Observed transaction

A transfer, trade, liquidation or redemption that actually happened. It proves that path worked once under those conditions.

Issuer or operator statement

A first-party operational claim. Useful, but not independently verified.

Third-party claim

Reporting or an aggregator result. We use it to find leads. Primary evidence overrides it.

Inference / unknown

Our reasoned reading is labelled as inference. If the documents do not answer the question, the result is unknown and records where we looked.

Limits

Known blind spots

The universe is built from searches

It is a union of search results, issuer lists and manual research; no official registry exists. An undiscovered issuer or address is absent, and a newly seen address may have been issued long before.

Holder counts are token accounts

Holder counts are non-zero token accounts. One person can control many accounts, and exchanges or custodians can aggregate many people into one.

A quoted price may not be what you get on exit

A pool quote, oracle price, issuer NAV and underlying equity price answer different questions. None alone states what a forced sale of meaningful size would realize.

Documents can be inaccessible or incomplete

Bot walls, JavaScript-only pages and unpublished agreements create explicit evidence gaps. Archived copies and hashes preserve what was reviewed where possible.

Protocol support can disappear

Only exact-mint listings and reviewed live products count. A protocol's UI, its registry and its on-chain accounts can disagree for a while.

A documented right may be untested

Redemption terms in a document do not show that a redemption has worked. Our bankruptcy analysis reads the structure of the documents. It does not predict how a court would rule in a future insolvency.

Flows are partly unobserved

xStocks creations and Superstate conversions into tokens are not collected, so their net flow is not computed. Only Ondo, xStocks and Superstate have a recurring redemption scan, and each day shows the hours it actually covers.

Float is an upper bound

Issuer inventory in a wallet we have not attributed counts as public float, and most xStocks have no market price to value it with.

Models can misread a change

A model assessment is a reading of a diff and carries no legal weight. It can call a change immaterial when a lawyer would disagree, so it never filters what is shown.

Litigation search is by name

The case-law watcher finds matters that name an issuer's entities or parties. A case under another name, outside the searched courts, or not yet public is missed.

External change

When sources change

We watch sources and tie every conclusion to dated evidence. When an issuer changes terms, an authority rotates or a protocol removes support, the previous observation remains part of the external history. Improvements to RWA Sonar's own research are not listed as market events; the public page shows the best current analysis. Detection is automated; material legal interpretation is reviewed before it is presented as settled.

Inspect watched evidence →