What ATCE is
The Adaptive Trust, Compliance and Execution Engine is how JIL puts trust ahead of value movement. It rules on each regulated act inside a sovereign domain before the act exists. A chartered entity may add local bylaws. The verdict on an act that stays inside the domain is a domestic ruling. The verdict on a hop that leaves the domain is a corridor ruling. The acts include: issuance, mint and burn, treasury transfers, cross-border settlement, redemptions, emergency controls. It is not a sidecar dashboard. It is how the chain decides whether an act may exist.
Status, 2026-11-12. The validator image was created 2026-08-21 and can deny a crossing. The rule that no transfer moves outside an ATCE ruling, including a release with no ATCE context, was committed on 2026-11-01 and is not in that image. The signing set on jil-mainnet-2 is eight.
Fail-closed
If identity is below the required level, a sanctions match fires, a limit is exceeded, or an attestation is missing, the transfer is not settled. The refusal is reason-coded and hash-committed. A regulator can open the reason. A bypassed off-chain gatekeeper cannot rewrite that history.
Before, not after
Public chains settle first and hope compliance catches up. ATCE inverts that. Proof is produced before value moves. That ordering is the difference between an audit trail and a control.
What ATCE actually evaluates
Two things meet at the moment of movement. The first is the policy profile, which lives on-chain as part of the cell's state, not in a document held somewhere else. It pins the identity level a counterparty must hold, the jurisdictions the cell will accept value from and send value to, the per-transaction ceiling, the risk score at which the cell stops, the sanctions posture, and the attestation that must be present.
The second is the context that travels with the transaction itself: the counterparty's identity level and its supporting attestation, the origin and destination jurisdiction, the amount, a attested risk score, and whether a sanctions screen fired. Because that context is carried by the transaction and signed, every validator sees the same facts.
ATCE takes those two inputs and returns three things: a verdict, a reason code for that verdict, and a single SHA-256 commitment binding the profile, the context and the outcome together.
Two layers
A sovereign domain is a gated garden. Anything that leaves the domain goes through ATCE and a corridor. ATCE also runs inside the domain, between the root and its hubs and annexes. That is two layers: domain to domain, and inside the domain. In an account-to-account move both accounts are ruled, because one may carry restrictions the other does not.
Before ATCE issues its final ruling it calls an agentic review service that can stop a transaction even when every check is green. It cannot approve one the rules refuse. This was added to improve trust along the corridor.
Both accounts are ruled
A plain transfer between two accounts is ruled on both of them, because account A may carry restrictions that account B does not. The ruling reads each account's own identity level, jurisdiction, risk, sanctions flag, expiry, freeze and zone, and its hash binds both accounts' record versions.
A restricted sender paying an unrestricted recipient is held. An unrestricted sender paying a restricted recipient is held. A transfer passes only when both accounts clear and sit in the same zone; an unzoned account counts as no zone. This is the rule the validators enforce on every transfer.
The record: a 10 year commitment
JIL commits to keep the transaction record for 10 years from creation, then releases it. Ten years is the maximum. It holds the ATCE rules that applied, the agentic approval, the buyer and seller details including the KYB and KYC checks, and the ruling. It is write once, read many (WORM) and immutable.
JIL holds the WORM record locally, not on the chain, and each ATCE transaction record has a unique hash. The Layer-1 references the transaction by that hash. The chain carries no names, documents or KYC files; those stay in the encrypted WORM record. Anyone given the record can recompute the hash and compare it with the one the Layer-1 references. The retention of 10 years from creation is a commitment JIL makes with each sovereign domain, by allocating petabyte-level storage capacity for it when its license is signed.
KYC and KYB are a condition of settlement
This is the part institutions usually ask about twice, because on other chains it is not true. Identity is not checked at an application's front door and then trusted for the rest of the journey. The required identity level and its attestation are evaluated inside consensus, on the transfer itself. A counterparty that has not satisfied KYC, or a business that has not satisfied KYB, does not generate a failed compliance report. It generates a transaction that never settles.
The same evaluation runs on the receiving side. When value crosses from one cell to another, the sending cell's policy is enforced on departure and the receiving cell's on arrival. Neither jurisdiction has to trust the other's onboarding, because each one re-runs its own rulebook against facts that traveled with the value.
One engine, two borders
ATCE is a single primitive with two callers, which is why it is a keystone rather than a feature. The federation arrival gate calls it when value crosses between cells. The Digital Currency Platform calls the identical evaluation on every transfer of a Currency Object, which is what that currency's Financial DNA means in practice.
One engine means one rulebook, one verdict format and one commitment scheme. A supervisor who has learned how a cross-border release is judged has also learned how a domestic transfer of the national currency is judged.
The verdict is part of the block
ATCE is deterministic and runs in consensus. Every validator re-runs the same evaluation over the same inputs and reaches the same answer, so the decision becomes part of the application state hash rather than a log line written afterwards. Three consequences follow, and each one matters to a supervisor.
- It is replayable. Anyone holding the block can re-derive the verdict and get the same result.
- It is attributable. The commitment is anchored to CourtChain and resolves through a public verify link.
- It is not discretionary. No operator can approve a transfer the published policy refuses, because the refusal is reached independently by every node.
The agentic review does not make the ruling discretionary. Its result is signed and recorded with the ruling, and it can only add a hold: it can stop an act the published policy allows, and it can never approve an act the policy refuses.
Fail-closed is a deliberate default rather than a safety net. A missing input, an unset limit or an empty jurisdiction list denies. A misconfigured cell moves nothing; it does not quietly move everything.
Why this is built the way a supervisor would want it built
Most compliance technology is reconciliation. Value moves, and a second system tries to reconstruct afterwards whether it should have. That model produces findings months later, drawn from records the supervised firm controls.
ATCE moves the control to the only point at which refusing is still possible, and produces evidence as a by-product of the decision rather than as a later report. The supervisor is not asked to trust an attestation the operator wrote. They are handed a commitment they can re-compute themselves.
It also leaves sovereignty where it belongs. JIL does not author anyone's rulebook. The cell operator publishes the policy profile, and ATCE guarantees only that the published policy is the policy that actually executed.
Why the consensus layer is called ACL
ACL stands for the Attested Consensus Layer, and the name is a design statement rather than branding. On a conventional chain, consensus answers one question: do the nodes agree on the ordering of these transactions? Identity, policy and jurisdiction are somebody else's problem, handled above the protocol or beside it.
ACL answers a second question in the same pass: is each act attested, carrying an identity, a policy verdict and a jurisdiction that the network itself evaluated? Agreement on ordering and agreement on admissibility are produced together, by validators that are named rather than anonymous.
That is the property regulated finance has always needed and general-purpose chains were never built to provide. It is also why the layer earns its own name: this is not a ledger with a compliance product bolted on, it is a consensus layer whose output is an attested record.
Status. Measured 2026-11-12, the every-transfer rule is not in the validator image created 2026-08-21. That image can deny a crossing. The three-arm Attested Consensus Layer seal is not the process on these hosts. Arming it in another cell is opt-in and requires a coordinated validator upgrade carried out with that cell operator. Yes. ATCE is patent pending. So is ACL, and so are all 177 patent claims. Nothing here is a granted patent.
Buyer, seller, currency, and the corridor journal
On a corridor hop, ATCE addresses every layer. The buyer brings the Tier 1 rules, the hub rules, and the annex rules. The seller brings the same three. When the movement uses a layer 1 or a stablecoin, that asset's rules on the currency site apply as well. The buyer Tier 1 and the seller Tier 1 have each enabled that asset. A transfer can take several hops, and each hop retrieves the rules again. A missing rule or a refusal stops the hop, and the refusal is still written down.
Each rule carries a version. A change appends a new policy number and does not rewrite the number already named. The policy word is that name. The ATCE log stores the number. The policy table stores the version chains, write once, and the corridor journal stores the policy text and the transaction: when it was created, who bought what, who sold, the source of funds or of the commodity, the history of that source, and the buyer and seller ATCE accreditations. The versioned table and the corridor journal are Designed. The corridor journal is the Layer-1 trust record on this host. It is not the payment-integrity product at jilsovereign.com, and a court still decides admission. Read The Corridor Journal.
The every-transfer rule, including refusal of a release that carries no ATCE context, was committed on 2026-11-01. It is not in the validator image created 2026-08-21. That image can deny a crossing.
The seller ATCE elects where a model may run. The buyer list does not elect the model, and an empty seller list allows no model. The agentic review can watch the parties and hold an act; eight yes votes do not clear that hold.
In practice
- KYC and KYB travel with the transfer
- Jurisdiction zones and sanctions are evaluated in consensus
- Per-transaction limits and reserve policy are live state, not PDFs
Questions people actually ask
Does ATCE replace a compliance officer?
No. It enforces the rules the cell operator publishes. Local law and the Statement of Authority still govern.
Can a contract talk around ATCE?
No. Policy is evaluated in consensus. A contract that would violate the rulebook does not settle.
Who writes the policy ATCE enforces?
The cell operator does, under its own law and its Statement of Authority. JIL supplies the engine that makes the published policy binding; it never supplies the rulebook.
What happens when a rule changes?
A change of the live pin is a governed transaction. The design keeps the old version as a policy number in a write-once table, cited by the corridor journal, so a past transfer can be re-read under the rule that applied; that table is Designed.
Keep reading
The white paper and a briefing are the next step if you are evaluating a domain.
