19 surfaces, one standard of evidence

Checks the promises the chain can settle — authority revocations, metadata immutability, program authority renouncement — against today's chain state and each deadline, and records the rest as not checkable, each with its reason.

Takes. A mint address, and the commitments you recorded — kind, deadline, where you saw them.

What it does not establish

The promise text is the reader's record and is never verified. A chain read shows today's state, never when it changed, so done-in-time is only provable while the deadline is still ahead.

A persistent list of mints bound to your address rather than a browser tab, with your own note beside each entry. Open a dossier from the list when you want fresh evidence.

Takes. A wallet sign-in, and the mint addresses you choose to keep.

What it does not establish

The list stores addresses, not opinions: nothing is scanned by being kept, and the note is the reader's record, never checked. Keeping a mint reaches nobody on its own — it decides which incidents a delivery destination receives, and an account without one is notified about nothing.

A library of documented failure mechanisms, each traceable to a checkable public incident and admitted under a written sourcing standard. A mint's recorded shape is mapped against each entry's markers — matched, unmatched and not establishable, separately, with counter-signals on every entry.

Takes. A token mint address — or nothing, to read the library itself.

What it does not establish

Matching an episode's markers is resemblance of structure, not destiny: no probability is stated, no outcome is predicted, and episodes document mechanisms, never operators. The registry grows one signed entry at a time and is deliberately small.

For each structural property recorded at a captured mint's enrolment — authorities, holder concentration, liquidity, pair age — reads every launch in this product's own corpus that shared the same band, and what fraction still had a tradeable market at day 1, 7 and 30, as a population rate with a 95% interval.

Takes. A token mint address the collector has captured.

What it does not establish

A population rate is never this token's probability: similarity is not outcome probability. Cells are single properties, never crossed, and a cell below its publication floor answers with why it is withheld and how far along it is rather than a number. Only mints the collector captured can be read.

Reads the token's saved observation sequence — nothing live — and names the transitions it can evidence: distribution concentrating or broadening, control tightening or loosening, liquidity thinning, dormancy and resurrection. Each is a stated rule over recorded values, with the evidencing observations and the window named.

Takes. A token mint address.

What it does not establish

A transition is an observation, not a warning: no base rate is attached until the reference corpus can back one, movement below a stated threshold is not named, and a quiet window is a reading of the record rather than a clean bill. Depth follows the viewer's timeline window.

A research desk's shared keeping: seats by wallet address, one shared list, an audit trail of who changed what, investigation rooms per token (hypotheses with their counter-evidence, notes, decisions, frozen pointers into the observation ledger), and a private overlay of the team's own labels with confidence, evidence and expiry.

Takes. A wallet sign-in on a plan with seats, and the wallet addresses you seat.

What it does not establish

A team reads nothing: no scan runs because an address is seated or a mint is listed, no score moves, and a quiet list is not a clean bill. Rooms and the overlay are the team's own words — unverified by Evidesa, visible only to seats, in no export, and never in the public graph, a score, a cohort or an alert. Alerting the whole team from its list is not built.

Sends the evidence watch's incidents to your own endpoint — signed, so you can prove the message came from here — or to a Discord channel. A destination receives incidents only for the mints on that account's watchlist.

Takes. A wallet sign-in, and an https endpoint or Discord channel you control.

What it does not establish

An https URL is not yet a destination: nothing is sent until the endpoint proves it expects deliveries. Silence has three causes that look identical from outside — nothing changed, nothing was re-observed, or the change was not notifiable — and no delivery is a recommendation to act.

Reads where each wallet's first lamports came from and reports the links it can evidence: a shared earliest funder, funding that landed in the same slot. Every link carries its confidence and the alternative reading.

Takes. Two to eight Solana wallet addresses — the set someone called a consensus.

What it does not establish

This read can challenge independence, never establish it: a link found is evidence, and no link found is the limit of one edge type over a bounded history. It ranks nobody, keeps no list of who is smart, and never says two wallets are one person.

Compares each statement against the evidence read for that mint and returns a document where every line carries how it was established — verified against chain, contradicted by it, or recorded as the project's own statement, unchecked.

Takes. What the project says about its own token.

What it does not establish

Preparing a document here establishes nothing about who prepared it: anyone can fill it in about any token. A signature narrows that to the holder of one key — never to a person, and never to authority over the project — and an unsigned statement stays labelled as the project's own, unchecked.

Mints the collector enrolled and captured, ordered by how many documented thresholds their captured evidence crossed. Every row prints what did not cross and what was never established, and links to the live dossier.

Takes. Nothing — the page reads this product's own captured corpus.

What it does not establish

A count of crossings, never a score, and never a recommendation: the order is a claim about this corpus rather than about any project. It lists only what was enrolled and captured, so absence from it is absence of observation — and no position on it is for sale.

Measures how much of a project's published wording, domain shape and on-chain metadata is shared with another token you name — or with the reference entries configured in this deployment.

Takes. One mint address, or two to compare directly.

What it does not establish

Shared copy is evidence of shared copy — not of shared ownership, coordination, or wrongdoing by anyone. The reference list ships empty by design, so a run against it alone reports that nothing was compared.