K
Governance · checked, not asserted

Governance

Governance changes how Skew grows — fees, listings, incentives, treasury. It can never change the math that protects your funds.

math-locked kernel
SIMULATION ONLYLocal simulation workspace.No transaction, wallet signature, program id, or funds movement is implied by this render.
The boundarySolvency is not on the ballot

A governance mistake can misallocate the treasury or mis-tier an asset — it can never create bad debt or unfund a claim. Aave and Maker let governance change the risk math; Skew puts that math beyond the vote.

Can govern
the market · futarchy
  • Future fee policy (new versions)
  • Asset-tier metadata
  • Listing bonds & incentives
  • Treasury spend & grants
  • Integrations & launch sequencing
  • Risk-cap defaults
Can constrain
governance / security council · fast to restrict
  • Halt a market → close-only
  • Disable a reference source (fail-closed)
  • Tighten caps / OI
  • Staged-rollout limits
  • Security reserves
  • Future factory admission
Cannot govern
nobody · immutable in code
  • Settlement & WCL math
  • Clamp-then-snap rounding
  • Receipt finality
  • Backed-profit rule
  • Existing fee-policy interpretation
  • Custody invariants & withdrawal path
How decisions are mademechanism · MetaDAO futarchy
1Growth proposal

Anyone bonds a proposal — fees, listings, incentives, treasury.

Kernel boundary

Touches settlement / WCL / rounding / receipts / custody? Auto-rejected — never reaches a vote.

hard wall
2Pass / fail markets

Conditional markets price each outcome (MetaDAO-style futarchy).

3TWAP finalization

Time-weighted pass-market price decides; resists last-minute manipulation.

4Timelock + exit

Delay + a user-exit window before an Expand action takes effect.

Execute

The program applies it — to growth surfaces only, never the kernel.

Growth proposals only. The kernel boundary is enforced on-chain by the program — a proposal that would touch a math-locked surface cannot be formed, let alone pass.

Separation of Powers

Power to make the protocol safer is fast and broad. Power to expand risk is slow, timelocked, and always preceded by a user-exit window.

RestrictFast - broad
  • Halt a market → close-only
  • Disable a reference source (fail-closed)
  • Tighten caps & open-interest limits
  • Staged-rollout limits
Takes effect immediately.
ExpandSlow - timelocked
  • Enable a new reference source
  • Raise caps & open-interest limits
  • Admit a new permissionless factory
  • Loosen staged-rollout limits
Timelock + user-exit window before activation.
What governance fundsfees → treasury → growth
Fee surfacesillustrative
Execution (maker / taker)
Per fill, by asset tier
0.50 / 3.50 bps
Premium fee
Option writers; capped
≤ 12.5% of premium
Notional (taker)
Forward / RFQ block, by tier
2.5 – 8.0 bps
WCL reservation
Risk-reservation fee
0.25 – 2.0 bps
Duration
Carry, by tier
0.01 – 0.05 bps/day
Settlement
At maturity
1.0 – 2.0 bps
Secondary transfer
Of max(notional, WCL) + settle
1.0 + 0.5 bps
Builder / DAO share
Permissionless verified markets
20% / remainder
Immutable per version

A fee policy is content-addressed (fee_policy_hash). A change creates a new version for future formation only — existing positions, contracts and receipts keep their original hash forever. Governance can approve a new version; it can never rewrite a live one, and fees are segregated from payoff conservation (Σ payoffs = 0, fee pool ≥ 0).

Fees → treasury → growthillustrative
Fees collected
$1.84M
30d, across surfaces
Classified pool
$1.71M
Only after they leave the liability pool
DAO treasury
$12.4M
Governable by futarchy
Growth spend · futarchy-allocated
Audits & security 34%Incentives 26%Listings & integrations 18%Grants 14%Buybacks 8%

Segregated from escrow — fees enter the pool only after they are no longer user liabilities; the treasury can be misallocated by a bad vote, but it can never reach a funded claim.

Constitution - Immutable Parameters

Governance cannot override
  • Settlement mathImmutable

    How a contract pays out at settlement is fixed in code — no vote changes a payoff.

  • WCL mathImmutable

    The worst-case-loss / escrow computation — the solvency inequality — is immutable.

  • Clamp-then-snap roundingImmutable

    Reference clamping + lattice rounding is fixed: loss ceil, winner credit floor.

  • Receipt finalityImmutable

    A settlement receipt is final and deterministically replayable; never re-opened.

  • Backed-profit ruleImmutable

    Only realized, fund-backed profit is ever withdrawable.

  • Existing fee-policy interpretationImmutable

    A live position keeps its original fee_policy_hash forever; a fee change is future-only.

  • Custody-accounting invariantsImmutable

    Integer conservation holds every transition: funded claims ≤ funded sources.

  • Immutable withdrawal pathImmutable

    Withdrawals can never be paused, frozen, or vetoed by governance.

Honest scope

What you give up for solvency

Skew is solvent by construction: each side escrows at least its worst-case loss, payoffs sum to zero, and the realized reference is clamped into the declared collar before payoff — so the vault is provably non-negative for any payoff, any asset, any volatility. That guarantee has an honest price, stated plainly:

Z3 / Kani · 1.2M payoff cases
No leverage on the default path1

Every position prepays its full worst-case loss up front (the WCC path). There are no margin calls and no liquidation engine — the cost is capital efficiency, paid honestly instead of risked.

Collared and capped only2

Each instrument declares a finite settlement range. A real move past the collar settles at the collar — disclosed up front, never hidden. Truly unbounded payoffs are excluded by design, not mispriced.

On-chain computability ceiling3

Payoffs too high-dimensional to soundly collateralize are rejected at registration, not silently admitted. If it cannot be proven bounded, it does not list.

Pricing is oracleless and derived entirely on-chain (batch-clearing tick / spot-pool TWAP behind a nine-guard firewall). A stale, bad, or manipulated reference can at most redistribute funds already escrowed between the two counterparties — it can never mint withdrawable profit or reach protocol funds. The reference is not a solvency primitive.

Reference policy & asset tiersgovernable within hard limits

Spot-Source Allow-List

Oracleless markets reference on-chain spot TWAPs from protocol-admitted venues.

  • Orca
    Concentrated-liquidity TWAP
    Enabled
  • Raydium
    AMM + CLOB hybrid TWAP
    Enabled
  • Meteora
    Dynamic-vault TWAP
    Enabled
Enabling is timelocked - added only after a delay and user-exit window.
Disabling is immediate & fail-closed - a suspect source is dropped at once.

Trust Tiers

A market's tier governs cross-margin eligibility and risk params. Promotion never requires a discretionary vote.

TierCross-marginRisk paramsMarkets
T0Provisional
Isolated onlyTightest OI cap - low leverage1
T1Established
Isolated onlyRaised OI cap - moderate leverage6
T2Trusted
Cross-eligible (same tier)High OI cap - high leverage4
T3Blue-chip
Full cross-marginMax OI cap - max leverage3
Objective on-chain merit

Promotion granted automatically when measurable on-chain thresholds (depth, age, uptime) are met. No vote, no discretion.

No vote required
Concordance

Several independent stake-weighted verifiers must independently agree. Disagreement blocks promotion - it cannot force one through.

Stake-weighted quorum

Bounded Residual Risks

Explicitly bounded
  • Genesis trust

    Initial tier assignments expire into objective merit; nothing stays trusted by fiat.

  • Upgrade authority

    Restrict-only fast path; every expansion is timelocked with a user-exit window.

  • External price-source integrity

    Multi-venue TWAP, confidence bands, fail-closed disabling on divergence.

  • Sustained participation

    Lazy keeper-free settlement - solvency does not depend on a liveness committee.

  • Host-chain liveness

    Forced-inclusion orders are censorship-resistant; state resumes on chain recovery.

Audit trailtrust evidence

Collateral flow

WCL/WCC audit trail

Track how worst-case collateral moves from reservation to escrow, release, and settlement.

0 bad debt
Reserved lock1
$154.4K
Maximum loss reserved before execution
Matched / escrowed2
$36.97B
Fully funded claims after match
Released collateral3
$5.6K
Unfilled or excess reserve returned
Settled receipts4
$24.4K
2 receipts replayed
Residual / bad debt5
$0.00
No unpaid settlement residue

Throughput

595K
fills / sec - single market
1.47M
fills / sec - across 42 markets
200K
orders / sec - intent submission