A scope class is the document that says what a caller may attempt against your interface, how often, and what happens when it goes outside. This is a worked one, so you can see the shape before you write yours.
Everything on this page is illustrative. It is not published, it binds nobody, and no member is admitted into it. The real one for your interface is written with you.
This is the part people get wrong first, and it is worth a minute because it explains the shape of everything below.
So a scope class governs what a caller may attempt, not what its payload contains.
Those rules still exist — they live at your own edge. The rulebook lets a receiving member apply any restriction stricter than the class permits, privately, without notifying the scheme or anybody else. You may always be stricter. You may never be looser. A refusal under that right is not a breach by anyone and carries no consequence.
Clauses 12.1 to 12.4, 13.1, 24.1 and 24.2.
| Method kind | Permitted | Meaning |
|---|---|---|
read.record | Yes | Retrieve one identified record |
read.collection | Yes | Retrieve a list or search result |
create | No | Bring a new record into existence |
update | No | Change an existing record |
delete | No | Remove a record |
payment.initiate | No | Move money |
notify.external | No | Cause a message to a third party |
The class names kinds, not methods, because it cannot know what your interface exposes. You map each of your own endpoints to a kind and file that mapping with the scheme. The mapping is of effect, not of signature — if one endpoint does several things behind the scenes, it maps to the most consequential of them. Mapping something to a kind that understates its effect is a breach, attributed to you, because only you know what your methods do.
Clauses 13.1 to 13.3.
| Limit | Ceiling | Window |
|---|---|---|
| All interactions, per counterparty | 600 | 1 hour |
| All interactions, per counterparty | 5,000 | 24 hours |
read.collection onlyList and search operations are heavier and are capped separately | 200 | 1 hour |
| Concurrent open sessions, per counterparty | 10 | — |
Ceilings are measured per counterparty and count only interactions the caller initiated. Windows are stated because a per-session ceiling is evaded by splitting the work across sessions.
Clauses 20.1 to 20.7 and 9.6.
Sums are fixed in advance, published before they apply, and identical for every member breaking the same rule in the same role. They are paid by the member in breach to the member affected, never to Rootwall — which cannot receive any sum on breach under a clause that cannot be amended.
| Breach | Standing | Sum |
|---|---|---|
| Attempting a forbidden method kindAn attempt is a breach whether or not it succeeded | Under Review; Suspended on repetition | $1,000 per session |
| Operating without a valid delegationOr after it expired, or beyond what the principal authorised | Under Review; Suspended on repetition | $1,000 per session |
| Exceeding a ceiling | Under Review | $250 per window |
| A false or misleading recordIncluding mislabelling a refusal, or understating a method's effect | Under Review; Suspended on repetition | $2,500 per record |
| Failing to produce a record, or to answer in timeEither role. See the note below | Under Review; Suspended if sustained | $2,500 per session |
Aggregate cap: $25,000 per member, per counterparty, per rolling 30 days, across all categories. Consequences to standing are not capped. The cap limits the scheme’s consequences and not anybody’s liability at law.
Why failing to produce a record costs more than most of the breaches it might hide. The rule is that the sum for evidence failure, per session, may never be lower than the highest per-session sum for an actual breach in the same class. Here that is $1,000, so $2,500 clears it comfortably.
A member that has just broken a rule must not find silence cheaper than admitting it. That single constraint is what makes the rest of the scheme work.
Why these numbers look small, and they are meant to. A sum here is a pre-estimate of what the breach costs the member affected by it to deal with — an engineer and someone in risk, a few hours for something contained. It is not damages, it is not a measure of loss, and it is not the scheme’s costs.
The deterrent in this scheme is standing, not money. Sums set high enough to frighten would stop being a pre-estimate of anything, and would invite a court to call them a penalty and strike them out. A reader who thinks these are too low has misread what they are for.
For contrast, and because the shape of the change is the point. SC-WRITE-1 would permit create and update alongside the read kinds, and everything else moves with it:
What does not change: the rules themselves, the evidence standard, the consequences framework, or anything in the rulebook. A scope class is a permission profile. It never varies a rule — the scheme cannot vary a rule for anybody.
Classes are versioned, and a member is admitted into a specific version. Where a new version is published, a transition period is stated. Nobody is moved onto new rules overnight.
Clauses 9.1 to 9.7, 14.1 and 14.2, 30.2.