KYC stays with issuers
KYC is performed only by council-approved issuers; validators never see documents. Agent minting adds an issuer-run AML provenance check — sanctions and mixer screening — before certifying funded tiers.
A sovereign Layer-1 for privacy-preserving, KYC-backed identity.
Approved issuers vouch for holders; holders keep their private keys and prove claims with zero-knowledge presentations. No raw KYC, PII or presentation logs ever touch the chain. The same rail extends to AI agents: an agent can prove it is backed by a real, liable, KYC-verified human — without revealing who.
Whitepaper v1.2 (frozen 2026-07-13) consolidates the earlier drafts into one document and integrates the July 2026 decision rounds into the body: the sovereign-L1 decision and its rationale, the token-launch compliance framework, the canonical five-flow economics, and the adopted design evolutions. v1.2 adds the x402 agent-payments ecosystem to the competitive register. It is a concept and market paper — non-normative; the specifications remain the design source of truth and are available on request.
01
Abstract
MintID is a sovereign, permissionless proof-of-stake blockchain for privacy-preserving, KYC-backed identity credentials. It lets approved KYC organisations issue credentials to users while users retain their private keys and control when to prove a claim.
The chain is public trust infrastructure: accepted issuer and verifier records, issuer-status commitments, revocation commitments, staking, incentives, governance actions and immutable audit commitments. It must never become a database of identity documents or a log of identity checks.
A second layer applies the same guarantees to AI agents. A human verifies once; the privacy-preserving accountability of that verification — and, optionally, pre-funded crypto escrow that backs it financially — flows into every agent they deploy.
Chain and token decisions. MintID launches as its own sovereign Layer-1 — the privacy guarantee is chain-wide, not contract-wide, and status freshness cannot ride a foreign chain’s fee market. Testnet tokens are worthless by construction; the genesis allocation follows a published transparency policy, with on-chain vesting, labelled addresses and an independent economic audit before genesis.
Adopted evolutions (July 2026). Five-axis provable scope, attested key-custody class, unlinkable operator context, and a push channel that makes revocation typically sub-second on top of the consensus-enforced 45-second guarantee — plus zero-cost verifier packages, MCP-native verification, and a wallet-local identity statement.
02
The problem
Making an identity credential accountable normally means a public database of identity documents or a log of every check — which destroys privacy and creates honeypots. For AI agents it is worse: today’s stack bolts agents onto non-human-identity plumbing that fails on four fronts at once — accountability, blast radius, sybil resistance and privacy.
Six recurring pains
- Impersonation and confused-deputy attacks.
- Anonymous clients — NHI sprawl with no paper trail.
- Cross-organisational delegation that silently breaks.
- Consent fatigue.
- Revocation that is effectively unsolved.
- Accountability vs. privacy treated as a binary choice.
03
The approach
Invert KYC, fund the backing, and gate disclosure by consent. Only compact cryptographic commitments live on-chain; all KYC, PII and presentations stay off-chain; holders prove predicates in zero knowledge.
What an agent proves
“I am operated by a KYC-verified, liable human, at assurance ≥ G, within scope S, with ≥ F of recourse behind me” — in zero knowledge, without revealing the human and without sibling agents being linkable.
A relying party can settle most disputes without ever learning who the principal is, and obtain identifying detail only with the principal’s explicit, priced consent, or under due process. The posture is identification, not surveillance.
Money-before-identity cascade
- L0 — escrow pays out with no identity revealed.
- L1 — a minimal fraud signal, not PII.
- L2 — full KYC, only under due process.
04
How it works
A sovereign Cosmos SDK application chain (v0.54.2) on CometBFT (v0.39.3) BFT consensus. The language boundary is a security boundary: Go owns consensus-visible state, Rust owns the specialised zero-knowledge cryptography off-chain, and Python owns research, testing and analytics only. The chain never calls an external prover, sidecar or database during block execution.
Consensus — candidate genesis parameters (pending simulation)
active_validator_set_max : 100
target_block_time : 2–3 s
finality SLO (mainnet) : p95 ≤ 5 s
minimum_self_bond : set at genesis
max_effective_power : capped to limit stake concentration
unbonding_period : set at genesis (after economic simulation)Assurance grades
The exact grade stays private. A verifier only learns that the grade meets or exceeds what it requested.
| Grade | Default validity | Meaning |
|---|---|---|
| A1 | 12 months | Basic identity binding and issuer verification. |
| A2 | 12 months | Stronger document and liveness verification. |
| A3 | 6 months | Enhanced due diligence and periodic evidence review. |
| A4 | 3 months | Highest assurance, stricter issuer controls. |
What touches the chain
On-chain
- Accepted issuer and verifier records
- Issuer status-root commitments (heartbeat ≤ 30s)
- Revocation commitments (Merkle / SMT roots)
- Staking, rewards and governance actions
- Evidence commitments — reason code + hash only
Never on-chain
- KYC documents and biometrics
- Names, addresses, dates of birth
- Credential serial numbers or raw payloads
- Presentation logs or per-user issuance records
- Anything that scales with the number of holders
Credential lifecycle
- issue
Issuance
Holder completes KYC with an accepted issuer; the issuer stores encrypted evidence privately and mints a private anonymous credential.
- status
Status
Each active issuer publishes a signed status root every 30 seconds or faster; an emergency root can be published immediately.
- present
Presentation
The holder proves a predicate in zero knowledge. The proof is valid for 10 seconds and is never broadcast, indexed or committed to the chain.
- renew
Renewal
Renewal creates a new credential with a new holder secret that does not publicly link to any prior credential — even from the same issuer.
- revoke
Self-revocation
A holder can revoke without issuer cooperation, revealing neither their identity nor a normal credential identifier.
05
Network roles
Every role has a narrow, enforced boundary. No single role can both see identity and decide consensus.
| Role | May | Must not |
|---|---|---|
| Holder | Control credential secrets, prove predicates, self-revoke. | Delegate signing or publish credential identifiers. |
| Accepted issuer | Perform KYC, issue credentials, publish status roots. | Put KYC data on-chain or inspect normal presentations. |
| Approved verifier | Request minimal claims, validate chain state and proof. | Request unapproved claims, replay a proof, or log it on-chain. |
| Validator ("miner") | Run consensus, validate deterministic state transitions. | Access KYC documents or judge real-world identity. |
| Compliance council | Approve / suspend issuers and verifiers, adjudicate bonds. | Move funds alone or approve a person’s identity. |
06
Tokenomics
Mechanism, not numbers
The token model is deliberately parametric. The architecture — hard cap, declining emission, fee-and-burn, issuer bonds and a separate reserve — is fixed. The maximum supply is already fixed: 93,924,792 tokens. Every other concrete value (allocation, vesting, rates) is set before genesis by independent economic simulation. No figures are invented here.
Hard-capped supply
Maximum supply is an immutable hard cap. The reward pool is minted only through a published, declining emission curve with a long horizon.
Monetary policy
MAX_SUPPLY = 93,924,792 tokens — immutable hard cap (ratified 2026-07)
GENESIS_MINTED = initial permitted distribution
REWARD_POOL = MAX_SUPPLY - GENESIS_MINTED
E(epoch) = min(remaining_reward_pool,
scheduled_declining_epoch_emission)
net_supply_change(epoch) = E(epoch) - burns(epoch)Fees and burns can make supply net-deflationary, but nothing promises perpetual deflation.
Sources of burn
- Issuer admission, renewal and active-registry lease fees
- A defined portion of status-publication fees
- A defined portion of issuer-bond and validator slashes
- Agent mint and recurring lease fees (protocol cut)
- Optional, governance-configured protocol-state rent
A recurring tax is never imposed on validator stake merely to force deflation — that would undermine security.
Validator & issuer economics
Consensus reward — per validator i, per epoch
R_i = E(epoch) × (P_i × S_i) / ΣW
P_i = capped voting power
S_i = signed-vote performance (0..1)Validators and their delegators earn a performance-weighted share of scheduled emission. There is no reward for subjective KYC work. Every active issuer locks a slashable native-token bond; off-chain services (indexing, archived proofs, verifier APIs) are paid direct fees that never influence consensus rewards.
BTC / XMR remediation reserve
A separate treasury reserve — not token backing, not a bridge, not escrow collateral (the escrow collateral rails are a separate, registry-governed mechanism). It is used only for proven issuer-fraud remediation and critical security incidents, behind a council threshold plus an independent custodian signature, an on-chain proposal, a reason code and a timelock. Routine operations, validator rewards, grants and price support are prohibited.
Agent-layer flows
| Flow | Direction | Token effect |
|---|---|---|
| Agent mint fee | Principal → issuer (+ protocol cut) | Protocol cut partially burned. |
| Agent lease (recurring) | Principal → issuer (+ protocol cut) | Recurring burn scaling with agent population. |
| Verification / mirror fees | Relying party → operators | Direct fees — never influence consensus rewards. |
| Issuer bond | Issuer locks native token | Slashable by council for proven fraud. |
| Disclosure bid | Relying party → principal | Payment for consent — never slashed; protocol cut partially burned. |
| Dispute bond | Relying party → arbitration | Returned if upheld; slashed if frivolous. |
The distinction is deliberate: a bond is slashable, adjudicated stake on the dispute rail; a bid is a non-slashable payment for consent on the commercial rail. No payment is ever the lawful basis for disclosure.
The canonical public form of these economics is the five-flow structure — agent mint and lease, issuer economics, disclosure brokerage, the dispute and arbiter market, and premium verification services — presented on the economics page.
Financial-assurance escrow
Distinct from the native token, principals self-custody USDC or EURC in their own wallet, with a smart-contract lock from tier F2 up. USDC and EURC are the launch collateral; the ratified target set adds BTC as the first post-launch volatile rail and, subject to legal review, XMR — each through the protocol’s collateral registry with per-asset over-collateralization. Assurance is published as bands (F0–F4), never as amounts; the band values are illustrative until pilot-calibrated. There is no fiat custodian and no chargebacks.
Set before genesis
- Token name, ticker and denomination
- Genesis distribution
- Emission decay curve and epoch schedule
- Fee split and burn percentages
- Minimum issuer bond and slashing ranges
- Validator cap, minimum self-bond and unbonding period
- Financial-assurance band amounts (F0–F4)
07
Governance
Protocol upgrades and versioned parameters are decided by token-weighted on-chain governance; the compliance council handles the operational gates: recommended as a 5-of-9 multisig, it approves, suspends and revokes issuers and verifiers, and adjudicates bonds. Moving the external reserve additionally requires an independent custodian or auditor signature.
Every governance action carries a proposal id, action type, target, policy version, reason code, an evidence commitment, an effective time and the approving signers with their threshold. Council decisions publish a non-sensitive evidence commitment and reason code — never the sensitive content.
Disputes are handled by a registered, KYC-approved and bonded arbiter market, selected by stake-weighted sortition and slashable — an on-chain registry analogous to the verifier registry.
Council capture is contained by threshold quorum, independently selected members, key rotation, public commitments and custodian co-control.
08
Roadmap
A security-first build sequence. No mainnet before independent application, cryptography, infrastructure and economic audits, testnet attack exercises, a bug bounty and an incident-response plan. Current status: specifications frozen (v1.2); chain foundation underway.
- M0
Protocol charter
Threat model, policy, state schemas and a draft economic model — boundaries approved.
- M1
Deterministic chain
The chain daemon, staking, slashing, rewards and upgrades on a local testnet, with reproducible builds.
- M2
Trust registry
Issuer, council, bond and status modules; status heartbeat and emergency root — no holder data.
- M3
Verifier controls
Approved-verifier and origin registry with exact-origin, request-key and nonce-replay checks.
- M4
Credential core
Anonymous-credential profile and issuance / presentation adapters — A1–A4 and multi-claim proofs in 10 seconds.
- M5
Self-revocation
Zero-knowledge self-revocation with an independent cryptography audit; no correlation or replay break.
- M6
Public testnet
External validators, issuers and verifiers; chaos tests, SLOs met and governance drills.
- M7
Mainnet genesis
Audited release with governance and reserve controls active; all launch gates and docs complete.
Agent layer — one launch scope
The committed agent model is a single launch scope, not a phase ladder: issuer-mediated agent credentials, self-custodied per-agent escrow and a consented commercial disclosure rail. One piece of new cryptography — the proof-of-harm circuit for the dispute rail — is a launch-blocking item: it ships only after independent audit, never silently. A shared liability pool and an insurance dispute backstop are explicitly deferred and unscheduled.
09
Risk & compliance
Lawful basis, cleanly split
No PII on-chain. Commercial disclosure rests on the data subject’s in-the-moment consent; the dispute rail on legal obligation or a court order. No payment, bid or bond, is ever the lawful basis — and a non-disclosing path to the transaction always exists.
Unmasking is last resort
Full-KYC unmasking is arbiter-decided, issuer-released and council-supervised — triggered only when both parties reject the arbiter’s settlement and the arbiter makes a neutral finding. Never on payment.
Deliberately out of scope
No identity proofs or per-user records on-chain, no validator access to KYC, no proof-of-work, no cross-chain bridges, no wrapped collateral, no stablecoin backing, no lending, no token-price support and no slashing for subjective claims.
10
Glossary
- Holder
- The user who controls credential secrets and proves predicates.
- Accepted issuer
- A council-approved KYC organisation that issues credentials and publishes status roots.
- Verifier / relying party
- An on-chain-registered party that requests minimal claims with exact-origin binding.
- Validator
- Runs consensus and validates deterministic state; never inspects KYC. Retained commercially as "miner".
- Compliance council
- The operational compliance body — a threshold-vote council (recommended 5-of-9) for issuer/verifier gates, bonds and treasury co-signing; software and parameters are governed token-weighted.
- Assurance grade (A1–A4)
- A private grade, proven by threshold as grade ≥ requested.
- Status root / heartbeat
- A compact signed issuer commitment published every 30 seconds or faster.
- Presentation
- A 10-second, nonce- and origin-bound zero-knowledge proof. Never committed to the chain.
- Principal
- The KYC-verified human or organisation deploying agents; bears accountability.
- Agent credential
- A fresh per-agent credential, unlinkable to the principal and to sibling agents.
- Accountability escrow
- The issuer’s private off-chain mapping from agent to human, revealed only under due process.
- Disclosure bid vs. dispute bond
- A non-slashable payment for consent versus slashable, adjudicated stake.
- Money-before-identity cascade
- L0 escrow payout (no identity) → L1 minimal fraud signal → L2 full KYC under due process.
Read the specifications
This page summarises the normative specifications SPEC-1, SPEC-2 and SPEC-3. Get in touch for the full documents.
Specifications frozen (v1.2); chain foundation underway. Every value shown as “TBD” is finalised before genesis through economic simulation and audit. Nothing here is an offer of securities or financial advice.