Sky Protocol Capital & Risk Management

Introduction

Sky Protocol operates the third-largest stablecoin system by circulating supply and the largest yield-generating stablecoin in existence. As of February 2026, the protocol carries approximately $9.93 billion in combined USDS and DAI supply — $5.99 billion USDS (60.3%) and $3.94 billion DAI (39.7%) — with December 2025 annualized gross revenue of $435 million and annualized protocol profits of $168 million ^[1][58]. The protocol's trajectory is steep: the Sky Frontier Foundation projects 2026 gross revenue of $611.5 million and a USDS supply target of $20.6 billion — nearly double the 2025 year-end figure ^[2].

The protocol's significance to credit markets crystallized in August 2025, when S&P Global assigned Sky Protocol a B- issuer credit rating — the first credit rating ever awarded to a decentralized finance protocol ^[3]. The rating highlighted three structural concerns: a risk-adjusted capital ratio of only 0.4% as of July 27, 2025; high depositor concentration among a small number of large holders; and governance centralization risk arising from the ability of a small set of actors to unilaterally influence protocol parameters ^[4]. A follow-up analysis by The Defiant noted that S&P saw no quick structural fix for the capitalization weakness, distinguishing between the protocol's revenue-generation capacity and its ability to absorb sudden, large losses ^[5].

This article covers Sky Protocol's capital structure, reserve buffer mechanics, revenue generation and financial performance, risk management framework, asset composition, governance architecture, and operational dependencies.

Several data points frame the protocol's market position. USDS is the third-largest stablecoin by supply and the largest yield-generating stablecoin, with the Sky Savings Rate (SSR) set at approximately 4.5% APY as of February 2026 ^[6]. The protocol is governed by the Sky Atlas, a comprehensive constitutional document that specifies parameters, risk controls, and governance procedures at granular levels of detail ^[7]. Capital is deployed through a network of autonomous Synomic Agents — Primes — of which the currently active Genesis Agents are Spark, Grove, Obex, Skybase, and Core Council EA1 ^[8].

The protocol's eight-year operational history — originating as MakerDAO in 2017 and rebranding to Sky Ecosystem in 2024 — provides a track record that is uncommon in decentralized finance. It has operated continuously through the March 2020 crash (ETH dropped more than 50% in 24 hours), the 2022 crypto winter, and the Terra/Luna collapse without a USDS haircut ^[1]. Black Thursday in March 2020 exposed a critical undercapitalization — a $500,000 surplus buffer against $140 million in DAI supply (0.35%) failed to absorb liquidation shortfalls, resulting in $4 million in bad debt and an emergency auction of 20,600 MKR tokens at approximately $275 per token ^[3]. That episode, now reflected in the S&P analysis, directly informs the protocol's current capital architecture and the 0.4% ratio critique.

The 2026 growth projections carry implications for every layer of the capital structure. A $20.6 billion USDS supply would require proportionally larger liquidity reserves, risk capital buffers, and counterparty diversification. At the projected gross revenue of $611.5 million, profits are forecast at $157.8 million — a 198% increase over actual FY2025 calendar-year profit (which was depressed by the Q1 2025 loss quarter), though below the annualized H2 2025 run rate of $168 million in absolute terms, implying higher expenses from the growth investment: up to 10 new Sky Agents are planned, each requiring Genesis Capital allocation and onboarding ^[2]. The analysis that follows examines whether the current capital architecture is sized appropriately for this growth, and how the protocol's risk frameworks would perform under stress.

Risk Governance Process

Sky Protocol's risk management parameters — capital requirements, rate limits, liquidity thresholds, and asset class constraints — are established and modified through a structured governance process. Understanding this process is essential context for the parameter-specific sections that follow.

Parameter Setting and Modification

Risk parameters are codified in the Sky Atlas, the protocol's constitutional governance document, and in the Laniakea specification for forward-looking framework parameters. Changes to Atlas-defined parameters follow one of three pathways ^[7]:

• Executive Votes: The standard governance pathway for material parameter changes. A governance spell encodes the proposed change, is submitted to the governance contract, and must accumulate more SKY voting weight than the current "hat" (the previously approved spell) before it can be executed after the 24-hour GSM Pause Delay.
• BEAM Authority: For time-sensitive operational parameters (Base Rate, SSR, rate limits), whitelisted operators can execute changes within pre-authorized bounds without the GSM Pause Delay. SP-BEAM governs rate changes; BEAMTimeLock governs rate limit modifications with a 14-day timelock on increases ^[29].
• Standby Spells and MOM Contracts: Pre-approved emergency actions (oracle freeze, debt ceiling zero, PSM halt, market freeze) that can be executed immediately by the Core Council or authorized multisigs without an Executive Vote ^[7].

Who Reviews and Recommends

The Core Council Risk Advisor (currently BA Labs) is the primary entity responsible for risk parameter analysis and recommendation. BA Labs assesses capital requirements, tests interim deployments, monitors encumbrance ratios, and proposes parameter adjustments to governance ^[7]. Scope Facilitators in the Stability scope (currently Steakhouse Financial) provide independent financial analysis and settlement monitoring.

Under the Laniakea target model, the 24-member Core Council (operating under a 16/24 supermajority threshold) will assume collective responsibility for risk governance, with the Risk Advisor role formalized as a standing advisory function rather than a discretionary authority ^[8].

Review Frequency

As of February 2026, risk parameters are reviewed on the following cadences:

• Monthly Settlement Cycle: Capital adequacy, encumbrance ratios, Genesis Capital allocations, and TMF waterfall distributions are assessed at each Monthly Settlement. The transition to Daily Settlement (Laniakea Phase 3) will increase the frequency of capital and risk monitoring to daily ^[39].
• Ad-hoc: Market stress events, material asset composition changes, or breaches of risk thresholds trigger immediate review. Rate changes via SP-BEAM can be executed within hours.
• Bi-weekly Executive Votes: Parameter changes approved through the standard governance pathway are bundled into bi-weekly Executive Vote cycles.

No formal periodic review committee (analogous to a bank's Asset-Liability Committee) has been established in Atlas or Laniakea documentation as of February 2026. The Laniakea roadmap envisions this function being progressively automated through the Sentinel Network (Phases 9–10), with AI-assisted monitoring and formula-driven parameter adjustment replacing manual committee-based review.

Capital Structure and Reserve Buffer

Capital adequacy was identified as the central weakness in the August 2025 B- rating. The protocol's capital structure spans multiple layers — from the narrow Surplus Buffer that absorbs first-loss, to Genesis Capital deployed across the Agent network, to the ultimate backstop of SKY token inflation and USDS haircut. Understanding each layer, its size, and the conditions under which it activates is essential for credit analysis.

Surplus Buffer

The Surplus Buffer is defined as the difference between Sky's total assets and total liabilities. Revenue flowing into the protocol increases the buffer; expenses — including Genesis Capital transfers to Agents — decrease it ^[7]. In practical terms, the Surplus Buffer is the first-loss capital that absorbs bad debt from liquidation shortfalls, smart contract exploits, or RWA defaults before any USDS holder is affected.

S&P's July 27, 2025 analysis placed the risk-adjusted capital ratio at 0.4% — meaning the protocol held roughly $0.40 of loss-absorbing capital for every $100 of USDS in circulation at that date ^[3]. For context, Basel III requires a minimum capital ratio of 10.5% (Pillar 1 plus Capital Conservation Buffer), with UK bank averages around 15%. The protocol's ratio is constrained by a deliberate design choice: rather than maintaining large capital reserves — which in traditional finance would typically be invested in risk-free assets such as T-bills — Sky routes most surplus revenue into SKY buybacks and staking rewards via the Treasury Management Function (TMF) waterfall, with capital retention set at a minimum floor rather than a regulatory target ^[8]. The deployment policy for the Aggregate Backstop Capital (i.e., whether the $50–60 million current buffer and the $309 million long-run target should be held exclusively in risk-free instruments) has not yet been specified in Atlas or Laniakea documentation and remains an open question for the protocol's Asset-Liability Committee equivalent.

Aggregate Capital Architecture

Sky's capital structure distinguishes between two aggregate measures that serve different analytical purposes ^[8]:

The Aggregate Capital Buffer is the broadest measure of capital within the system, comprising: the Surplus Buffer; the Core Council Buffer; the Aligned Delegate (AD) Buffer; and Prime Agent SubProxy capital. This figure represents total resources that could theoretically be deployed to absorb losses, though not all components are equally liquid or accessible.

The Aggregate Backstop Capital is the credit-relevant solvency metric. Following a reclassification implemented in late January 2026, the formula is:

Aggregate Backstop Capital = Total Genesis Capital − Allocated Genesis Capital

This reclassification is analytically significant. The Core Council Buffer was previously included in the Aggregate Backstop Capital calculation; its reclassification to operational capital simultaneously reduced the headline safety margin and changed expense recognition timing. Under the prior treatment, Surplus Buffer transfers to the Core Council Buffer appeared as internal capital movements rather than expenses; under the current accounting treatment, those transfers are recognized as expenses at the point of transfer from the Surplus Buffer ^[9]. A double settlement (November and December 2025 processed together in late January 2026) and the one-time accounting reclassification occurred in the same window, creating a one-off expense spike in Q4 2025/Q1 2026 data ^[9].

As of Q1 2026, the Laniakea framework projects total Allocated Genesis Capital of approximately $120 million, with Aggregate Backstop Capital in the $50–60 million range ^[8]. A target of $125 million Aggregate Backstop Capital has been established for the Genesis Phase, to be achieved through a 25% minimum retention of net revenue at each Monthly Settlement ^[8].

The post-Genesis long-run target is more ambitious: 1.5% of total USDS supply. At the 2026 supply target of $20.6 billion, that implies a long-run backstop target of approximately $309 million ^[10].

Genesis Capital Allocations by Agent

Genesis Capital is the mechanism by which Sky Protocol bootstraps its Agent network. Importantly, Genesis Capital allocation is not an expense in the standard sense — it is a capital deployment that remains within the system as backstop capital, available to be reclaimed in an extreme insolvency scenario ^[8]. The following table summarizes confirmed Genesis Capital commitments as of February 2026:

Nine Genesis Agents are planned in total: four Genesis Stars (Spark, Grove, plus two unannounced Q1 2026), one Genesis Institutional Prime (Obex), and three Genesis Guardian Agents each receiving $25 million ^[8]. Of the three Guardian Agent allocations, $20 million per agent is ring-fenced as a Core Council and GovOps Support Buffer — capital that must be spent entirely in support of general Sky Ecosystem development before Guardians may use it for token buybacks or staking rewards ^[8].

Insolvency Defense Hierarchy

Sky Protocol's insolvency framework operates as a four-level sequential defense. Each level is activated only if the preceding level is exhausted ^[8]. The levels differ in their degree of automation:

• Defense Level 1 — Aggregate Backstop Capital: The first-loss buffer absorbs losses from collateral shortfalls, bad debt, or protocol failures. As noted above, this is the $50–60 million pool as of Q1 2026, with a $125 million target. Trigger mechanism: Automatic. Bad debt from liquidation shortfalls is recognized on-chain at the point of auction settlement — losses are absorbed by the Surplus Buffer without requiring a governance vote or manual intervention.

• Defense Level 2 — SKY Token Inflation: If Aggregate Backstop Capital turns negative following one or more Agent liquidations, the protocol inflates the SKY token supply to raise additional capital through what would function as an emergency debt auction. This mechanism has precedent: in March 2020 (the Black Thursday event), 20,600 MKR tokens were auctioned at approximately $275 per token, raising approximately $5.5 million — with Paradigm Capital acquiring approximately 68% of the issuance ^[3]. Trigger mechanism: Requires governance action. The debt auction (flop) mechanism is initiated by a governance spell when the system's internal accounting recognizes that total debt exceeds total surplus. In 2020, this required an Executive Vote to authorize the MKR auction parameters and kick off the auction. Future implementations under Laniakea may automate detection, but the auction initiation currently requires human-directed governance execution.

• Defense Level 3 — Genesis Capital Backstop Mechanism: In the extreme scenario where the SKY token price collapses to zero and cannot generate further funding, the protocol enacts the Genesis Capital Backstop Mechanism, reclaiming capital from Genesis Agents through a pro-rata haircut. Agents whose capital is reclaimed do not simply absorb the loss: their token holders — who have effectively covered the loss after SKY dilution — receive an airdrop of the newly issued SKY supply ^[8]. Trigger mechanism: Requires governance action. The Core Council would need to execute a spell reclaiming Genesis Capital from Agents.

• Defense Level 4 — USDS Haircut (Final Resort): If the Genesis Capital Backstop Mechanism is insufficient to recapitalize the system, the remaining loss is socialized among USDS holders through a proportional haircut. In this outcome, USDS holders receive an airdrop of the new SKY token supply, enabling them to configure the system to utilize future returns to recover the original 1:1 peg ^[8]. Trigger mechanism: Requires governance action. This is functionally equivalent to an Emergency Shutdown scenario with a modified recovery path.

This four-level view is a simplified Genesis Capital perspective. The Laniakea whitepaper's detailed seven-step loss absorption waterfall maps Level 1 to steps 1–4 (First-Loss Capital, Junior Risk Capital, Agent Token Inflation, and SRC Pool), Level 2 to step 5, Level 3 to step 6, and Level 4 to step 7 ^[8].

Genesis Capital Phase-Out Logic

Genesis Capital is explicitly a temporary mechanism intended to be active only during 2026 and 2027. It applies only to an amount of capital equivalent to the original Genesis Capital provided by Sky to seed each Genesis Agent ^[8].

Phase-out is triggered when two conditions are simultaneously satisfied ^[8]:

1. The Genesis Agent has launched a liquid token with at least $10 million in average daily volume during the preceding month.
2. Aggregate Backstop Capital is at or above $50 million.

When both conditions are met, the Genesis Agent phases out $1 million of Genesis Capital at each Monthly Settlement, plus an additional $1 million for every $10 million of Aggregate Backstop Capital above the $50 million threshold. An illustrative example: if two Genesis Agents have qualifying liquid tokens and Aggregate Backstop Capital stands at $58 million, each phases out $1 million per settlement. If the next month Aggregate Backstop Capital grows to $64 million (above $60 million but below $70 million), each phases out $2 million. Total Genesis Capital can only decrease over time, eventually reaching full phase-out ^[8].

This phase-out design creates a self-limiting backstop: as agents mature and the backstop grows organically through revenue retention, the Genesis Capital mechanism becomes redundant and is wound down.

Emergency Response System and On-Chain Enforcement

Sky Protocol's emergency response architecture operates through several overlapping mechanisms. An emergency situation is defined as any condition requiring immediate intervention to prevent Emergency Shutdown, severe peg divergence, or irreversible harm to the ecosystem ^[7].

The Emergency Response Group comprises: Atlas Axis, Steakhouse Financial, BlockTower Capital, Core Council Risk Advisor (BA Labs), Maker Growth Foundation, Dewiz, and Sidestream ^[7]. Membership is defined by name in the Sky Atlas (Section A.1.9.3) and can only be modified through Executive Votes. Selection criteria are based on operational capacity, domain expertise in protocol risk management, and existing contractual relationships as Ecosystem Actors or Scope Facilitators within the Sky governance structure. As of February 2026, the Emergency Response Group draws from 18 of the 24 active Ecosystem Actors and Scope Facilitators ^[41]. The Core Council holds the authority to force executable spells (governance actions) on Prime agents without requiring a standard token hat vote — a power reserved for emergency use only. Warden sentinels hold halt authority to freeze the Execution Engine. SKY holder full quorum can dismiss all 24 Core Council Guardians ^[7].

Rate limits are the universal on-chain enforcement mechanism for all capital flows. Every capital outflow from the protocol is rate-limited, creating a controlled-velocity constraint on potential losses from compromised parameters or governance attacks ^[8]. The System-wide Outflow Rate Limit (SORL) is set at 25%; the individual rate limit (IRL) is $100,000, with an 18-hour cooldown between increases. A 14-day timelock applies to all new allocation additions; removals are executed immediately ^[7].

The Delegation Intent Policy (DIP) is an on-chain policy defining allowed assets, position limits, velocity limits, and concentration caps for each Prime. The lpla-checker tool cross-checks positions against these limits, calculating the Capital Requirement Ratio (CRR), Total Required Risk Capital (TRRC), and Total Risk Capital (TRC) in real-time ^[7]. On-chain enforcement of risk capital constraints — blocking new deployments when capital ratios are breached — is specified as a future governance proposal rather than a currently active mechanism ^[12]. The encumbrance ratio target is ≤90% (TRRC/TRC); agents exceeding this threshold would face new deployment freezes and mandatory deleveraging timelines ^[12].

Quantitative Loss Budget and Historical Precedent

The Surplus Buffer serves as the protocol's first-loss reserve against four categories of loss: crypto collateral liquidation slippage; RWA haircuts or defaults; smart contract exploits; and oracle failures ^[7].

The Black Thursday precedent establishes the minimum credible stress scenario. On March 12–13, 2020, a 50%+ ETH price drop triggered mass liquidations. The Surplus Buffer held approximately $500,000 against $140 million in DAI supply — a 0.35% ratio — and proved wholly inadequate to absorb the $4 million in bad debt from failed zero-bid auctions ^[3]. The subsequent MKR debt auction at ~$275 per token raised approximately $5.5 million to recapitalize the system ^[3].

The protocol has since invested substantially in buybacks rather than pure capital accumulation. As of December 2025, cumulative SKY buybacks since February 2025 totaled approximately $92.2 million via the Smart Burn Engine ^[1]. As of February 2026, buybacks continue at approximately $300,000 per day — approximately $109.5 million annualized ^[9].

The Smart Burn Engine (SBE) operates under current parameters defined in the Atlas: a temporary fixed-amount configuration of approximately $300,000 per day, with all TMF Steps 4 and 5 currently unified in execution ^[9]. The long-run SBE formula is Burn Rate = (1 − MC/TMC) × 50%, where MC is current market capitalization and TMC is a target market capitalization derived from growth rate and annual profits. This formula is not yet active; activation requires a SBE BEAM governance system that depends on the daily settlement cycle transition ^[9].

Forward-Looking Stress Scenario: 50% ETH Drop in 24 Hours (Q1 2026)

The following scenario applies the Black Thursday precedent to the protocol's current exposure profile, using the best available data as of early 2026 ^[55]:

Current Exposure Baseline

• Total onchain crypto lending exposure: $1.45B as of February 2026, with approximately 70% ETH, 20% wstETH, and the remainder in other collateral ^[58]
• Spark total exposure: $1.54B with $36.94M in risk capital held against $25.79M required (69.83% encumbrance) ^[59]
• System collateralization ratio: 172.57% (Core Vaults)
• wstETH liquidation penalty (chop): 13%
• Oracle delay: 1-hour OSM
• Current Aggregate Backstop Capital: $50–60M (estimated; the $125M target remains in progress)

Scenario Mechanics

A 50% ETH price drop within 24 hours — replicating March 12–13, 2020 — would push all positions below approximately 200% collateralization into liquidation territory. The Black Thursday benchmark produced a 4% bad debt rate on affected supply: $5.67M in bad debt on $140M in DAI supply. Current ETH-correlated exposure across Core Vaults and SparkLend is roughly 10x larger than the 2020 book. A proportional bad debt outcome — $40–60M — would exhaust the current Aggregate Backstop Capital in a single event.

Key Differences from March 2020

Liquidations 2.0 Dutch auctions (replacing the Liq 1.0 English auctions that produced zero-bid outcomes in 2020), substantially deeper DEX liquidity across multiple venues, and multiple competing keeper networks all reduce — but do not eliminate — the risk of cascading liquidation failure. The October 10–11, 2025 market stress event provides a partial validation: the largest DeFi liquidation event to date processed over $20B industry-wide, Aave processed $180M in liquidations with zero bad debt, and Sky/Spark survived with no reported bad debt.

Buffer Adequacy Assessment

The current $50–60M Aggregate Backstop Capital is thin relative to a 2020-severity scenario applied to today's larger book. The $125M target and 25% net revenue retention floor directly address this gap, but the buffer will remain vulnerable to a severe correlated crypto drawdown until the target is reached. The post-Genesis 1.5% of USDS supply target ($150M at current supply; $309M at 2026 target supply) would provide materially more headroom.

Capital Accumulation Timeline

The 25% net revenue retention floor provides a quantifiable path to the $125M Genesis Phase target. Based on the Sky Frontier Foundation's projected $157.8M in 2026 protocol profit ^[2], 25% quarterly retention implies approximately $9.9M per quarter (assuming even quarterly distribution). Starting from the current $50–60M base:

These projections assume no stress events that would deplete the buffer during the accumulation period. A 2020-severity loss event ($40–60M) during the first 12 months would reset the timeline by approximately 4–6 quarters. The 25% floor is a minimum — governance may increase the retention rate at any Monthly Settlement to accelerate buffer building, particularly if the Security and Maintenance budget declines from 21% to the target 10% as projected for late 2026, which would release an additional ~$17M annually for backstop retention ^[8].

Revenue and Financial Performance

Sky Protocol's revenue model is structurally simple: the protocol earns a Base Rate on deployed capital and pays the Sky Savings Rate (SSR) to sUSDS holders, with the spread constituting net revenue. However, the mechanics of how that spread is determined, how income is waterfall-distributed, and how it has evolved across 2025 require precise understanding for a credit analysis.

Revenue Formula and Base Rate Architecture

The core revenue identity, as defined in the Laniakea whitepaper, is ^[8]:

Net Revenue = (Base Rate × Deployed Capital) − (Savings Rate × sUSDS Deposits)

The Base Rate is the key protocol reference rate — the rate charged on all capital deployed by Prime Agents and on collateral in Core Vaults. The relationship between the Base Rate and key derived rates is fixed by Atlas provisions ^[7]:

• SSR = Base Rate − 30 basis points
• Agent Rate (idle capital) = Base Rate − 10 basis points
• Sky Spread = 10 basis points of the 30-basis-point spread between Base Rate and SSR
• Distribution Reward Fee = 20 basis points of the same spread

This architecture means that a 30-basis-point spread between the Base Rate and the SSR is structurally guaranteed regardless of the absolute rate level, providing a minimum revenue floor per unit of supply. At $20 billion USDS supply and 30 basis points of structural spread, the implied minimum contribution from this spread mechanism alone is approximately $60 million annually — before any asset-specific income above the Base Rate.

Income Sources

The full taxonomy of Sky Protocol income sources ^[7]:

1. Stability Fees from the Base Rate on all deployed capital
2. Internal Senior Risk Capital Income (OSRC payments from Prime Agents posting risk capital)
3. External Senior Risk Capital Fees (5% of net interest earned by External SRC providers)
4. Agent Upkeep Fees (paid by Primes for operational support)
5. Agent Creation Fees (one-time fees on new Agent onboarding)
6. Sky Core Vault Income (stability fees from legacy ETH/WSTETH vaults, plus liquidation penalties)
7. LitePSM Income (Coinbase Custody yield on USDC held in the PSM)

The income mix has shifted significantly through 2025 as legacy Core Vaults declined in relative importance and Prime-based income (Spark, Grove, Obex allocations) grew.

Revenue Waterfall

The revenue waterfall governs how gross income is allocated before reaching the TMF Surplus Buffer ^[7]. At a high level, the waterfall operates as follows:

• Step 1: 10% to Security/Stability (subdivided: 5% Core Executor, 3% Reward, 1% Aligned Delegates, 1% Accessibility)
• Step 2: 5% to High Activity Staking Rewards
• Step 3: Variable percentage to Stability Capital Buffer retention
• Step 4: 80% to Smart Burn Buffer + 20% to Standard Activity Staking Rewards

This Atlas waterfall sits above the TMF waterfall, which then governs how the net revenue entering the Surplus Buffer is subsequently allocated ^[10].

Under the TMF (as specified in Laniakea Appendix C), net revenue flows sequentially through five steps ^[10]:

The Net Revenue Ratio is a piecewise hyperbolic function: NRR = Current Yearly Net Revenue / (Net Revenue + 3B USDS Target Constant). At $168 million net revenue (2025 annualized), NRR ≈ 0.053. At the projected $157.8 million 2026 profit, NRR would be similar — meaning the Fortification and Burn steps each receive only approximately 1% of the Step 2 result, while staking rewards receive the bulk ^[10].

FY2025 Financial Performance

FY2025 marked the protocol's first full year under the USDS/SKY branding and demonstrated the revenue power of the model as well as its sensitivity to the savings rate level ^[13][14].

The Q1 2025 loss resulted directly from the elevated SSR level: interest payments to sUSDS holders consumed essentially all gross revenue during a period when the savings rate was at its highest ^[15]. The Q2 2025 recovery — with SSR expenses falling 51% quarter-over-quarter — reflects SSR reductions implemented by governance in response to the Q1 loss. By year-end, the operational expense trajectory was dramatically improved: spell execution costs dropped from approximately $16 million per quarter in Q1 2025 to approximately $0.3 million per quarter in Q4 2025 as legacy overhead wound down ^[9].

FY2025 expense composition: savings costs (DSR + SSR + Integration Boost) constituted approximately $194 million — roughly 68% of total expenses of approximately $285 million. Operational costs were approximately $67 million. This cost-of-capital vs. operational-expense distinction is material for analysts: the savings rate is a competitive market cost that scales with USDS supply and rate levels, while operational expenses represent controllable overhead ^[9].

SKY buybacks accelerated through the year: from approximately $367,000 in early 2025 to $92.2 million cumulative by December 2025, representing over 6.3% of the 23.46 billion total SKY supply repurchased ^[1].

2026 Projections

The Sky Frontier Foundation published 2026 projections in February 2026 ^[2]:

The apparent profit decline from annualized 2025 ($168M) to projected 2026 ($157.8M) despite significantly higher gross revenue reflects the costs of the growth initiative: higher SSR expenses as USDS supply nearly doubles, Genesis Capital allocations to new Agents, and the Security & Maintenance budget under the 21% Genesis Phase rate. Up to 10 new Sky Agents are planned for 2026 launch ^[2].

The revenue composition expected for 2026 includes several materially new income streams: Spark Prime and Spark Institutional Lending launched on February 11, 2026, adding OTC crypto lending revenue (see Asset Composition section below) ^[16]; and additional PSM deployments on new chains. Subsidized borrowing provisions — Spark and Grove are each entitled to borrow 1 billion USDS at a subsidized rate for two years starting January 2026, using the formula t-bill_rate + ((base_rate − t-bill_rate) × T/24) where T is elapsed months — create a ramp structure where the cost to the protocol increases linearly over 24 months ^[7].

Interest Income Composition

As of Q2 2025, interest income broke down into three primary segments ^[13]:

• Crypto Vaults (~28%): Stability fees from ETH-A, ETH-B, ETH-C, WSTETH-A vaults — legacy overcollateralized lending at the Base Rate
• PSM Income (~37%): Yield from USDC held in the LitePSM, primarily Coinbase Custody sweep yield on the PSM USDC balance
• RWA Yield (~34%): T-bill instruments (BUIDL, JTRSY, USTB), CLO tranches (JAAA), and legacy private credit positions

The shift toward Prime-based income (Spark lending, Grove RWA/CLO) is structural and ongoing. Core Vault revenue has been declining as borrowing demand normalizes and certain collateral types (notably WBTC) are actively being offboarded. The PSM income share will compress as Grove assumes operational ownership and the PSM transitions from a direct yield source to an ASC-managed liquidity tool ^[9].

Instrument-Level Income Breakdown

The most granular publicly available revenue decomposition comes from Block Analitica's October 2024 analysis of the Sky ecosystem's financial performance, annualized from that month's snapshot ^[56]:

Additional income streams outside stability fees: liquidation fees of $3.9M annualized and PSM swap fees of $47K (Sky operates 0% PSM swap fees under normal conditions) ^[56].

The Q2 2025 Protocol Economics Report provides a more recent quarterly breakdown ^[57]:

The Spark Tokenization Grand Prix (April 2025) accelerated RWA income diversification: over $2B was deployed across BUIDL ($500M+, BlackRock), USTB ($300M+, Superstate), and JTRSY ($200M+, Centrifuge/Janus Henderson). At approximately 4.24% T-bill yield, these tokenized treasury instruments generate an estimated $85–90M annually — a revenue line that did not exist prior to Q2 2025 ^[56].

Steakhouse Financial maintains the canonical income dashboard on Dune (dune.com/steakhouse/makerdao), which provides the authoritative source for real-time revenue decomposition at the instrument level ^[57].

Risk Management Framework

Sky Protocol's risk management architecture spans two generations: the current Atlas-based regime, which applies rule-based parameters to individual vault types, and the Laniakea framework under development, which introduces a capital formula, duration matching, correlation categories, and a unified risk capital tower across all asset classes. As of February 2026, the Laniakea framework is in early-phase implementation; the Atlas regime remains operative for most positions.

Laniakea Capital Formula

The Laniakea risk framework computes Total Required Risk Capital (TRRC) (capital formula) for any Prime's portfolio as ^[17]:

Total Capital = Σ Position Capital + Category Cap Penalties

For each position, the calculation depends on the asset's pull-to-par characteristics ^[17]:

For assets with a stressed pull-to-par (SPTP) — e.g., CLO AAA tranches

Duration Capacity = Cumulative liability amount in buckets ≥ required bucket
Matched Portion = min(Position Size, Available Duration Capacity)
Unmatched Portion = Position Size − Matched Portion

Matched Capital = Matched Portion × Risk Weight
Unmatched Capital = Unmatched Portion × max(Risk Weight, FRTB Drawdown)
Position Capital = Matched Capital + Unmatched Capital

For liquid assets without pull-to-par (perpetual exposures, e.g., ETH)

Position Capital = Position Size × max(Risk Weight, FRTB Drawdown)

For collateralized lending (gap risk / liquidation shortfall)

Position Capital = Position Size × max(Risk Weight, Gap Risk CRR)

If the risk weight and forced-loss terms use different exposure bases (notional vs. market value vs. exposure at default), both are computed in dollars before applying the max(...) function ^[17].

Capital Requirement Ratios by Asset Class

The Atlas and Laniakea framework specify CRRs by asset class. CRR is defined as Required Risk Capital / Capital Invested, so a 5% CRR on $100 million implies $5 million of required risk capital ^[7]. CRRs are established through a layered governance process: the Core Council Risk Advisor (BA Labs) proposes per-asset CRRs based on the Laniakea capital formula inputs — risk weights, FRTB drawdown estimates, gap risk models, and stressed pull-to-par durations — and these are ratified through Executive Votes. Interim and blanket CRRs (25% near-term, 100% for testing) are transitional governance decisions set by the Core Council pending completion of granular per-asset modeling. Once the Laniakea framework is fully implemented, CRRs will be formula-driven, with governance controlling the input parameters (risk weights, correlation categories, stress scenarios) rather than setting CRR levels directly ^[17]. As of February 2026, the following CRRs apply or are planned:

The near-term blanket 25% CRR is a transitional measure that significantly overstates required capital relative to the final granular model. Once per-asset CRRs are finalized for each deployed strategy, the aggregate capital requirement will decrease — releasing capacity for incremental deployment.

Spark Risk Capital: Actual vs. Required (February 2026)

The Spark Risk Capital & Requirements Monitor provides real-time visibility into Spark's capital adequacy ^[59]. As of February 25, 2026:

The 69.83% encumbrance ratio indicates Spark holds approximately 43% more risk capital than the minimum required — a meaningful buffer above the minimum threshold. The 1.68% effective risk rate (required capital as a percentage of total exposure) reflects the blended impact of per-position CRRs across Spark's diversified lending book, ranging from 0% on cash stablecoins to 25% on interim deployments. Individual position CRRs are visible at the per-asset level on the dashboard, with the largest capital charges on crypto lending positions carrying gap risk.

Grove Risk Capital: Actual vs. Required (February 2026)

The Grove Risk Capital & Requirements Monitor shows a similar structure for the institutional credit Star ^[60]:

Grove's lower effective risk rate (0.99% vs. Spark's 1.68%) reflects its portfolio composition: T-bill positions (JTRSY at $561M, BUIDL at $380M) carry 0% CRR, while AAA CLO positions (JAAA at $387M combined across two tranches) carry 1.6–2.1% CRR. The highest per-position CRR in Grove's book is 9.9% on the ACROX (Centrifuge) position at $51M. The 75.43% encumbrance ratio indicates Grove operates with less capital headroom than Spark — approximately 33% above the minimum required, compared to Spark's 43%.

Combined Risk Capital Position

The two largest Primes together present a consolidated risk capital picture as of February 2026:

The combined $63.25M in held risk capital across Spark and Grove represents a meaningful portion of the system's capital base, with a blended encumbrance ratio of approximately 72% — indicating the two largest Primes collectively hold approximately 39% more risk capital than their minimum requirements.

Risk Capital Ingression System

Capital adequacy under Laniakea is not simply a binary "enough / not enough" assessment. The ingression system defines how external capital counts toward a Prime's effective capital base. All ingression uses a flat-plus-quarter-circle curve: marginal ingression rates are 100% up to an anchor point, then decline along a quarter-circle to zero at a maximum point ^[18].

Three tiers of risk capital exist with distinct ingression parameters ^[18]:

The market-capitalization-based total cap means that a Prime with a $100 million effective market cap can ingress up to $500 million in total effective risk capital at the flat rate, with a theoretical maximum of approximately $1.285 billion (5× + 10× × π/4) ^[18].

Duration Model — Calibrated to Bank Run Data

The duration matching system addresses the core asset-liability mismatch risk: the danger that Sky deploys capital into long-duration assets while USDS holders — analogous to demand depositors — can redeem instantly ^[19].

The framework uses a Lindy Duration Model for the liability side: each lot of USDS is assigned an expected remaining holding time equal to its current age multiplied by a Lindy factor (conservative 0.5× to 0.7× rather than 1.0×). The protocol then applies a 101-bucket system at 15-day intervals to represent the duration profile of the liability base ^[19].

The structural maximum caps per bucket are the credit-critical parameters, derived from a double exponential decay model calibrated to empirical bank run data:

Individual Cap(t) = A × e^(−λ₁ × t) + B × e^(−λ₂ × t)

Calibration parameters ^[19]:

The calibration targets were fitted to the aggressive end of observed bank run data ^[19]:

Additional calibration inputs include MMF crisis data: September 2008 (26% outflow in 2 weeks) and March 2020 (30% in 2 weeks) ^[19].

Key checkpoints for analyst use (from the 101-bucket table) ^[19]:

A key limitation of the Lindy Duration Model as applied to Sky Protocol merits acknowledgment. The model is calibrated to bank run data from institutions — SVB, First Republic, Credit Suisse — that offered a range of deposit and funding products (term deposits, CDs, bonds, wholesale facilities) to manage asset-liability mismatch. Sky Protocol has no equivalent product diversity: all USDS liabilities are effectively demand deposits with instant redeemability and no maturity ladder. The behavioral stickiness assumptions embedded in the Lindy factor (0.5×–0.7×) may therefore overstate the expected remaining holding time for USDS relative to a traditional bank's deposit base. The conservative calibration parameters (using the aggressive end of observed bank run data) partially offset this concern, and the ASC framework (see Peg Stability section) provides the operational liquidity backstop that banks achieve through funding product diversification. Nonetheless, analysts should note that the duration model represents the protocol's proposed analytical framework for a genuinely novel liability structure, and its parameters may require recalibration as empirical USDS redemption data accumulates.

The implication for the CLO strategy is direct: JAAA CLO tranches with a stressed pull-to-par of approximately 3.5 years (1,260 days) can only be duration-matched against the cumulative 12.57% of USDS liabilities with expected remaining duration of 1,260 days or more. Any CLO exposure beyond this bucket cap — i.e., above approximately 12.57% of the USDS portfolio at the relevant bucket — is treated as unmatched and must be capitalized at max(Risk Weight, FRTB Drawdown) rather than just the base risk weight ^[19].

Correlation Framework and Category Caps

Correlation risk — the danger that the portfolio concentrates into assets that fall together in a stress scenario — is managed through governance-defined Correlation Categories and hard cap concentrations ^[20].

The enforcement mechanism is simple: any exposure above a category cap is subject to a 100% CRR on the excess, creating a hard capital penalty for over-concentration ^[20]. Category examples include "CLOs," "US-based assets," "40–50 month duration instruments," and "real estate." Each category cap is expressed as a percentage of total USDS supply. For example, a "CLOs ≤ 10% of USDS supply" cap would limit total CLO exposure to approximately $986 million at current supply — and any exposure above this would require dollar-for-dollar capital backing ^[20].

The cap calibration methodology converts correlation risk from an abstract covariance matrix into a concrete loss budget constraint ^[20]:

For each scenario s and category c:
cap_percent[c] = min_s ( B(s) / L(c, s) )

Where B(s) is the governance-defined maximum loss fraction tolerable under scenario s, and L(c, s) is the stressed loss per dollar of category c exposure under that scenario. Governance controls include the scenario set, loss budgets, cap update cadence, smoothing rate limits (e.g., ±X% per update), and an emergency freeze capability to halt cap updates while maintaining enforcement ^[20].

A "capacity rights" system prevents new entrants from instantly displacing incumbents from within-cap allocations. Each Prime holds a non-transferable in-cap allocation that adjusts gradually over time based on penalty payment history, with normalization periods of max(asset SPTP, 3 months) ^[20].

Gap Risk and Collateralized Lending

For overcollateralized lending positions — the protocol's largest asset category — the relevant risk measure is gap risk: bad debt arising when collateral prices crash faster than liquidations can execute ^[21].

The Laniakea approach models gap risk empirically ^[21]:

1. Historical crash data analysis (flash crashes, black swan events) to characterize worst-case price gaps
2. Health factor distribution modeling under stress — not assuming all positions simultaneously reach HF = 1.0
3. Expected bad debt calculation for instantaneous price gaps of X%, derived from historical worst cases
4. Capital sizing to absorb the worst-case scenario at the chosen confidence level

Liquidation loss given default (LGD) is amplified by four factors specific to on-chain markets: auction slippage due to limited AMM depth; oracle latency or stale feeds; keeper congestion during peak market stress; and collateral correlation when all liquidators are selling the same assets simultaneously ^[21].

Sparklend positions and other overcollateralized crypto lending positions have no pull-to-par (they are perpetual, with no maturity), meaning they cannot be duration-matched and must be capitalized on a max(Risk Weight, Gap Risk CRR) basis at all times ^[22].

Rules-Based vs. Case-by-Case Governance

The Laniakea framework is explicitly rules-based at the position and category level. CRRs, FRTB drawdowns, gap risk models, and correlation caps are all formula-driven; governance controls the parameters of these formulas rather than adjudicating individual positions ^[17][20]. Cases falling outside defined asset structures escalate to governance for bespoke treatment — typically receiving the blanket 25% near-term CRR or 100% interim CRR during the assessment period ^[7].

Transition status for the Laniakea risk models as of February 2026: implemented for Lending Markets (Spark, Aave, Morpho, Fluid, Maple); pending implementation for Perpetual Positions, Direct Exposures, Bond-Like Instruments, Cash Stablecoins, and RWA ^[7].

Asset Composition and Collateral Strategy

Sky Protocol's asset composition spans crypto-native collateral in Core Vaults, real-world assets deployed through the Prime Agent network, OTC crypto lending products, and peg stability reserves. The portfolio has evolved substantially since the Endgame initiative, shifting from a primarily ETH-backed vault model toward a diversified multi-asset structure managed by specialized Stars. As of February 2026, the USDS backing portfolio totals approximately $9.89 billion across six asset categories — stablecoins ($5.72B), onchain crypto lending ($1.45B), short-duration treasury bills ($941M), AAA corporate debt ($812M), OTC crypto lending ($800M), and other positions ($162M) ^[58]. This section details the current composition, examines the new OTC crypto lending products, analyzes the AAA CLO strategy, and quantifies operational dependencies on third-party service providers. The asset mix's yield diversification and liquidity profile is relevant to evaluating the projected $20.6 billion USDS supply target and risk capital coverage under the CRR framework.

Balance Sheet Overview

As of February 25, 2026, the Sky Risk & Analytics Dashboard reports the following USDS backing composition across the entire protocol ^[58]:

The dominance of stablecoins (57.85%) reflects the PSM USDC reserves, multi-chain PSM3 deployments, and stablecoin allocations across Spark, Grove, and Obex. The combined non-stablecoin risk asset backing — crypto lending ($1.45B), T-bills ($941M), CLOs ($812M), OTC lending ($800M), and other ($162M) — totals approximately $4.17 billion, representing the yield-generating asset base that supports protocol revenue and the Sky Savings Rate.

The Core Vault collateral base within the crypto lending segment is concentrated in Ethereum ecosystem assets: approximately 70% ETH (across ETH-A, ETH-B, ETH-C vault types), approximately 20% liquid staked ETH (wstETH-A), with the remainder in positions being actively wound down, including WBTC ^[7]. The overall system collateralization ratio on Core Vaults was 172.57% as of January 2026.

This vault-level collateralization ratio significantly overstates system-wide resilience when viewed in isolation. The relevant credit metrics are the risk-adjusted capital ratios described in the Capital Structure section, not the gross collateralization ratio on the crypto vault book. A 172% collateralization ratio on volatile crypto assets provides a meaningful buffer against gradual price declines; it provides materially less protection against sudden correlated drawdowns, oracle failures, or liquidation cascades — scenarios directly analogous to the Black Thursday event.

Sky Direct Exposures

Beyond the Core Vault book, Sky Protocol has deployed substantial capital through its Prime Agent network into a diversified set of instruments. The following table summarizes key direct exposures as of February 2026 ^[7][23][24]:

Cash stablecoins — USDC, USDT, pyUSD — are classified as eligible reserve assets ^[7].

OTC Crypto Lending (Spark Prime and Spark Institutional)

Spark's OTC crypto lending products launched on February 11, 2026, introducing a new revenue category ^[16].

Spark holds exclusive rights to OTC crypto lending on behalf of Sky Protocol. Two products launched simultaneously ^[16]:

• Spark Prime is margin lending through the Arkis prime broker engine. Borrowers post collateral across centralized exchanges, DeFi venues, and custodians simultaneously, enabling multi-venue collateral management. The Arkis system monitors positions continuously and initiates automatic position unwinding if risk thresholds are breached. The primary target client segment is crypto hedge funds seeking leverage without liquidating existing positions across fragmented venues.

• Spark Institutional Lending is fully custodial lending through Anchorage Digital, a regulated institutional custodian holding a national trust charter (OCC). BTC is the collateral asset. Initial LTV is 80% (implying 125% collateralization); margin call triggers at 85% LTV (approximately 117.6% collateralization). Maximum authorized exposure is $500 million. The Atlas-specified CRR for this product is 3%, implying $15 million in required risk capital at maximum deployment ^[7].

Prior to the February 2026 formal product launch, Spark had demonstrated capacity for institutional-scale deployment: approximately $500 million in USDC was supplied to Coinbase's BTC-backed loan market over a three-month period, and approximately $1 billion into the PayPal PYUSD market on Morpho ^[16]. These deployments provide a performance track record to supplement the formal product documentation.

AAA CLO Strategy (Grove)

Grove launched on June 24, 2025 with an initial $1 billion allocation to the Janus Henderson Anemoy AAA CLO Strategy (JAAA) — the largest single RWA deployment in Sky Protocol history at that date ^[23]. The structure employs Centrifuge infrastructure for on-chain tokenization; Anemoy is a BVI Segregated Portfolio Company licensed by the BVI Financial Services Commission; and Janus Henderson serves as portfolio manager, with the same team managing Janus Henderson's $21 billion AAA CLO ETF.

The Avalanche expansion — targeting $250 million in JAAA and JTRSY tokens — was announced in July 2025, leveraging Centrifuge's cross-chain capabilities ^[24]. The higher CRR for Avalanche-held CLO positions (2.1% vs. 1.6% on Ethereum) reflects the additional bridge and cross-chain protocol risk that cannot be captured in the underlying CLO risk weight alone ^[7].

Additional CLO developments include: Galaxy CLO 2025-1 (Cedar Grove Ltd) authorized by DAO Resolution on December 11, 2025; and BNY/Securitize STAC with approximately $100 million planned ^[7].

The stressed pull-to-par treatment for CLO AAA tranches is notable. Normal pull-to-par is approximately 2.5 years; the 1.4× stress modifier applied in the Laniakea framework (calibrated to 2008–2009 prepayment slowdowns, where CLO prepayment rates dropped from 28% to 9–15%) extends the stressed pull-to-par to approximately 3.5 years ^[22]. This stressed duration determines which liability bucket the CLO exposure must be matched against, directly limiting the maximum CLO concentration at any given USDS supply level.

Legacy RWA Wind-Down

The governance policy for legacy RWA positions established under MakerDAO governance before the Endgame initiative specifies that these positions must remain "for as long as necessary, optimized for yield if possible," and should be wound down or offboarded when feasible ^[7].

Notable legacy positions in wind-down or transition:

• Andromeda (RWA015): Short-term treasury position; designated safe; being transitioned to Grove operational management ^[7]
• Legacy HVB and other private credit: Bespoke lending arrangements predating the Halo framework; being wound down on counterparty-specific timelines; no standardized LCTS interface or Halo Unit token
• WBTC Vaults: Debt ceiling set to zero; offboarding in progress ^[7]

The collateral offboarding threshold under Atlas governance is set such that any vault with below $20 million in total debt triggers mandatory offboarding consideration ^[7]. Direct Executive Votes can execute offboarding without a prior Governance Poll — a streamlined process that reduces execution lag when winding down unwanted risk positions ^[7].

Specific legacy position status as of February 2026:

The Laniakea implementation roadmap designates Phase 1.3 — "Legacy Cleanup & Core Halos" — as the formal process for standardizing retained legacy assets as Core Halo entries and winding down the remainder. The general governance policy permits direct Executive Votes to execute offboarding without requiring a prior Governance Poll, streamlining the process. No firm completion date has been published for full legacy RWA elimination; the Atlas states these positions must remain "for as long as necessary, optimized for yield if possible" — wind-down is opportunistic rather than scheduled, contingent on counterparty loan maturity and real estate market conditions.

Actively Stabilizing Collateral and Liquidity Constraints

The Actively Stabilizing Collateral (ASC) framework governs the protocol's peg defense capability. The ASC requirement imposes a minimum liquidity buffer that limits how much of the USDS liability base can be supported by illiquid assets ^[25].

Key ASC parameters ^[25]:

• Prime Agents must hold at least 5% of their Collateral Portfolio in Actively Stabilizing Collateral
• Resting ASC must provide buy support at a minimum price of 0.999 USD per USDS (10 basis points downside spread), through directly and atomically executable bids or verifiable market-making arrangements
• Latent ASC — cash stablecoins convertible to resting ASC within 15 minutes — may not exceed 25% of total ASC
• Demand Absorption Buffer (DAB) must equal 25% of required ASC, consisting of USDS available for sale at ≤1.001 USD per USDS

During a Peg Defense Event (triggered when average USDS price on LayerZero-connected DEXes falls below 0.999 USD), each Prime must buy USDS at a rate of at least 6.25% of its ASC requirement every 6 hours ^[25].

The duration-bucket framework imposes additional structural liquidity constraints by capping allocations to long-duration assets ^[19]:

Under the structural cap system, approximately 76% of the portfolio could theoretically be held in assets with a duration above 720 days (24 months), but the cumulative caps significantly constrain this in practice: the capacity at bucket 48 (720 days) is approximately 23.5% cumulative, declining to 12.6% at the JAAA duration bucket ^[19].

The ASC parameters and liquidity constraints listed above — the 5% minimum, 0.999 price floor, 25% latent cap, DAB requirements, and duration bucket structural caps — are defined in the Sky Atlas and the Laniakea specification documents. Modifications to these parameters require Executive Votes (or, for certain rate-related parameters, BEAM authority within pre-authorized bounds). The Core Council Risk Advisor (BA Labs) is responsible for monitoring compliance and recommending parameter adjustments. Review cadence is governance-driven: parameters are assessed as part of the Monthly Settlement Cycle (transitioning to Daily Settlement under Phase 3), with ad-hoc review triggered by market stress events or material changes to the asset composition. No fixed periodic review schedule (e.g., quarterly ALCO) has been formalized in Atlas or Laniakea documentation as of February 2026.

Operational Dependencies and Third-Party Concentration

Sky Protocol's investment operations involve material dependencies on third-party service providers:

Chronicle Protocol's oracle network uses Scribe aggregated signatures from 25 validators. The oracle system is a systemic dependency: a coordinated manipulation or extended outage would impair liquidation mechanics across the entire crypto vault book ^[7].

No active Arrangers are listed in Atlas as of February 2026 ^[7].

Peg Stability and Liquidity Management

Sky Protocol's peg stability architecture combines a legacy Peg Stability Module (PSM) infrastructure with the emerging Actively Stabilizing Collateral (ASC) framework introduced under Laniakea. Together, these mechanisms constitute the protocol's first line of defense against USDS depeg events. Analysts must understand both layers: the PSM provides immediate, deterministic arbitrage capacity backed by USDC reserves, while ASC formalizes per-Prime liquidity obligations across the entire agent ecosystem. The transition from PSM-centric stability to distributed ASC obligations is actively underway as of February 2026, with Grove assuming operational ownership of the LitePSM as the first step. This section quantifies the current parameters, describes peg defense trigger mechanics, and assesses rate management tools that allow monetary policy to adapt to liquidity conditions without requiring full governance cycles. The general process through which these parameters are set and reviewed is described in the Risk Governance Process section above.

PSM Architecture

The LitePSM (Lite Peg Stability Module) functions as a 1:1 USDC-to-USDS conversion backstop deployed on Ethereum mainnet, enabling risk-free arbitrage at any USDS price deviation from $1.00 ^[26]. As of February 2026, the LitePSM operates with tin (fee-in) and tout (fee-out) both set at 0%, meaning swaps incur no protocol fee and arbitrage is profitable at any nonzero price differential ^[26]. Key DC-IAM (Debt Ceiling Instant Access Module) parameters:

The 400M USDS buffer is maintained by keeper networks via fill(), trim(), and chug() operations, shifting the minting cost from individual users to protocol infrastructure ^[26].

PSM3 contracts extend the arbitrage mechanism to Layer 2 networks: Base, Arbitrum, Optimism, and Unichain are all supported as of February 2026 ^[26]. Rate limits on L2 PSM3 contracts prevent rapid drawdowns that could strain cross-chain bridge capacity.

Grove is in the process of assuming operational ownership of the LitePSM from Sky Core, after which Grove manages the LitePSM as an ASC asset under ALM rules ^[25]. Coinbase Custody currently holds the USDC deposited in the PSM, earning yield on that balance for Grove's benefit. The transition preserves Sky Core's active oversight until full ASC implementation is operational.

An emergency halt mechanism — the LITE_PSM_MOM (MOM contract) — can pause all PSM swaps without requiring the standard 24-hour GSM Pause Delay ^[27]. This capability proved essential as a design safeguard after the March 2023 USDC depeg event demonstrated that PSMs can transmit rather than contain external stablecoin shocks.

USDT Integration

As of February 2026, USDT is integrated across the protocol through five simultaneous channels: SparkLend USDT markets, Aave Core USDT allocations, Curve sUSDS/USDT and USDC/USDT pools, Maple USDT, and Uniswap v4 USDT/USDS pools. The Atlas classifies USDT as a Cash Stablecoin alongside USDC and pyUSD, assigning it 0% Capital Requirement Ratio — identical treatment to USDC — and imposing no exposure limit ^[7].

USDT TVL on Spark grew from $25 million in late July 2025 to over $900 million by October 2025 — a 3,500% increase in under four months — validating organic demand before Q1 2026 infrastructure expansions ^[46]. The Spark Savings USDT vault carries a supply cap of 2 billion USDT, signaling governance's view of the addressable scale.

On February 23, 2026, Sky launched dedicated USDC and USDT Risk Capital Vaults on Morpho. The USDT vault deploys deposits exclusively into the stUSDS/USDT market at 86% LLTV, offering yields above 8% annually ^[40]. The vault design creates a structural USDS demand flywheel: borrowers must acquire USDS and deposit it as stUSDS collateral to borrow USDT, generating recursive USDS demand from every dollar of USDT deposited. This represents the "substantial strategic increase in USDT liquidity allocation" referenced in the Sky Frontier Foundation's 2026 outlook ^[2].

The strategic rationale spans four dimensions relevant to credit analysis. First, addressable market expansion: USDT's approximately $142 billion market capitalization is roughly 5–6 times larger than USDC's, representing a massive untapped inflow pool. Second, counterparty diversification: the legacy PSM architecture was heavily USDC-concentrated, creating single-issuer dependency on Circle. USDT in ASC-qualifying Curve pools provides peg-defense liquidity denominated in a different issuer's liability, meaning a Circle-specific failure cannot simultaneously impair all peg stabilization capacity. Third, the Morpho vault mechanism generates USDS supply growth through leveraged stUSDS positions — a supply-growth flywheel rather than simple yield generation. Fourth, Spark retains the full profit and loss on USDT investments held in SparkLend, creating direct financial incentive for aggressive USDT deployment without requiring governance coordination ^[7].

ASC Framework

The Actively Stabilizing Collateral (ASC) framework formalizes per-Prime minimum liquidity obligations ^[28]. Key structural parameters:

• Minimum ASC: Each Prime must hold at least 5% of its Collateral Portfolio in ASC at all times ^[28].
• Resting ASC: Must provide buy support at or above 0.999 USD per USDS (10 basis points below peg). Qualifying forms include USDC in LitePSM and PSM3, cash stablecoins in Curve or Uniswap pools paired with USDS, and USDC in GUNI pools ^[25].
• Latent ASC: Cash stablecoins convertible into Resting ASC within 15 minutes via fully automated process. Latent ASC is capped at 25% of total ASC. Qualifying assets include cash stablecoins in SparkLend, Aave, Morpho, and in Prime ALM Proxies ^[25].
• Demand Absorption Buffer (DAB): Each Prime must maintain a DAB equal to 25% of its required ASC, consisting of USDS available for sale at no more than 1.001 USD per USDS. This provides sell-side peg stabilization during periods of USDS excess supply ^[25].

As of February 2026, the stablecoin component of USDS backing — $5.72 billion, or 57.85% of total backing — represents the pool from which ASC obligations are primarily drawn ^[58]. ASC is designed to displace the legacy PSM as the primary liquidity management mechanism over time, distributing peg defense obligations across the full Prime ecosystem rather than concentrating them in a single centralized pool.

The Laniakea target state envisions ASC as a comprehensive ALM layer across all Primes, with the legacy PSM becoming an implementation detail of Grove's liquidity operations rather than a protocol-level mechanism ^[25].

Peg Defense Events

A Peg Defense Event is triggered when the average USDS price on LayerZero-connected DEXes falls below 0.999 USD per USDS ^[28]. Upon trigger:

• Each Prime with ASC obligations must immediately begin purchasing USDS at a rate of at least 6.25% of its ASC requirement every 6 hours ^[25].
• Peg defense may be executed by selling other collateral for USDS, or by using newly generated USDS to borrow other assets (e.g., USDC on Aave) and purchasing USDS on the open market ^[25].

The Laniakea framework notes that near-term penalties for ASC failures and peg defense shortfalls are not yet explicitly defined and are scheduled for future specification. Enforcement currently relies on detection, reporting, and reputational consequences rather than automated on-chain penalties ^[25].

Rate Management and Run Risk

PSM ASC levels serve as a primary monetary policy signal. The Base Rate adjustment table:

High PSM ASC indicates excess reliance on centralized stablecoin reserves and prompts rate decreases to encourage vault borrowing and collateral diversification. Low ASC signals insufficient liquidity buffers and prompts rate increases to generate protocol revenue and rebuild reserves ^[29].

The Sky Savings Rate is bounded via SP-BEAM: maximum 30%, minimum 2%, with adjustments limited to 400 basis points per step ^[29]. SP-BEAM allows authorized operators to modify rates without the 24-hour GSM Pause Delay ^[29]. Dynamic IAM (DC-IAM) auto-adjusts debt ceilings to maintain target available debt levels without governance intervention.

Hot Money and Concentration Risk

The Laniakea duration model employs a double exponential decay function calibrated to empirical bank run data to model USDS liability "hot money" risk ^[19]:

Individual Cap(t) = A × e^(-λ₁ × t) + B × e^(-λ₂ × t)

Parameters:

At a 30-day horizon, the model implies that up to 25% of the USDS supply may redeem during acute stress — calibrated to SVB's 25% single-day outflow and First Republic's 37% two-day outflow in March 2023. At 90 days, the cumulative outflow cap reaches 45%, consistent with First Republic losing 57% of deposits over Q1 2023 and Credit Suisse losing 29% in the same period. Money market fund crisis data — September 2008 (26% in two weeks) and March 2020 (30% in two weeks) — anchors the short-end calibration ^[19].

S&P Global's August 2025 rating report flagged Ethena's USDe as constituting approximately 11% of Sky's total assets at the time of the rating, applying a 1,250% Basel III risk weight to that exposure ^[30]. The Q1 2025 episode — where SSR was raised to 12.5% to attract capital, generating Ethena inflows followed by a $5M protocol loss when rates normalized — exemplifies the hot money dynamic the duration model attempts to quantify ^[15].

Star Architecture and Agent Risk Governance

The Star Prime architecture distributes Sky Protocol's capital deployment across specialized semi-autonomous entities, each operating within governance-defined mandates, exclusivity domains, and risk limits. As of February 2026, four Synomic Agents (Stars) and one Institutional Prime are operational, with two additional Stars in the Genesis Phase pipeline. Each Star operates under a Program Accord specifying capital constraints, borrowing terms, and operational boundaries. Stars are functionally analogous to operating subsidiaries: they hold and deploy protocol capital, generate revenue, and bear first-loss risk on their respective portfolios — but unlike subsidiaries, governance retains override authority through on-chain controls and can recall capital, freeze operations, or revoke authority without requiring legal proceedings. The concentration of credit and operational risk within each Star, and the correlation risk across Stars, constitutes the primary portfolio-level risk vector for Sky Protocol.

Genesis Agent Overview

Spark's subsidized borrowing arrangement: up to 1B USDS for two years from January 2026 at a subsidized rate, with the requirement to maintain at least 1B USDS deployed to remain eligible for full monthly reimbursement ^[31]. Grove operates under the same 1B USDS minimum borrowing threshold for eligibility.

Exclusivity Rules

Exclusivity rules prevent cannibalization across Stars and define each agent's addressable capital:

• Spark: Exclusive rights over Maple Finance, cryptocurrency OTC lending, and Established DeFi Lending categories ^[31].
• Grove: Exclusive rights over Real World Assets, excluding Treasuries held for ASC purposes or stablecoin liquidity management ^[31].

Exclusivity rules function as the primary mechanism partitioning risk domains across the Star ecosystem. Violation of exclusivity by one Prime results in 100% CRR on the non-exclusive portion of the exposure under the Correlation Framework's penalty regime ^[20].

Star Risk Standards and Enforcement

All Primes are subject to CRR-based capital requirements as defined in the Laniakea capital formula. For each position, capital is calculated as:

Position Capital = Matched Portion × Risk Weight
                 + Unmatched Portion × max(Risk Weight, FRTB Drawdown)

Interim deployments — those operating before standardized Laniakea infrastructure is in place — require 100% CRR on all deployed capital during the testing period ^[31]. The Core Council Risk Advisor (BA Labs) must specify testing parameters for each interim deployment before it may proceed. This requirement applies to Phase 0 legacy deployments currently being standardized as Core Halos under Phase 1.

On-chain rate limits constrain all capital flows. The Initial Rate Limit is $100,000 per initialization target, with a Second-Order Rate Limit (SORL) of 25% per 18-hour cooldown window constraining rate increases ^[20]. These parameters were calibrated against Type 1 (direct extraction) and Type 2 (slippage grinding) attack models, with total weighted harm bounded at approximately $2.36M per initialization target at adopted parameters.

Governance Oversight Delegation

The Laniakea target governance model specifies a Core Council of 24 Guardians operating under a 16/24 supermajority (2/3rds threshold) requirement ^[20]. Quarterly rotation schedules replace 6 of 24 seats per quarter, completing a full cohort cycle in four quarters via SKY holder polls.

As of February 2026, Core Council EA1 — the first Core Executor Agent — was approved in December 2025 with an initial target membership of 7 ^[31]. The transition from the current SKY token Executive Vote model to the 24-Guardian Council model is underway under the SpellGuard system.

Key SpellGuard authorization controls:

• SpellCore: Core Council hat requiring 16/24 Guardian approval. Controls Council Beacons, BEAM ownership override, Synome write rights, Prime SpellGuards, and Council composition ^[20].
• Prime SpellGuard: Dual-key authorization requiring both a top-down core spell payload AND the Prime token holder hat. Prime SpellGuards can only trigger Halo spells — they cannot modify Prime-level state directly ^[20].
• SKY Holder Supremacy: SKY holders retain the ability to dismiss the entire Core Council via override, maintaining ultimate sovereignty while delegating day-to-day governance ^[20].

Preventing Correlated Risk Across Stars

The Correlation Framework (Laniakea correlation-framework.md) is the primary mechanism controlling concentration risk across Stars ^[20]. Governance defines arbitrary Correlation Categories (examples: "CLOs," "US-based assets," "real estate," "bridge exposures") and assigns hard caps as a percentage of total USDS supply. All category caps are enforced through the capital formula: exposure above a cap attracts 100% CRR on the excess portion, eliminating any leverage benefit on over-cap positions.

Cap calibration uses scenario stress engines under two methods:

• Method A (Independent Caps): cap_percent[c] = min_s(B(s) / L(c, s)) where B(s) is the maximum portfolio loss fraction tolerated under scenario s, and L(c, s) is the stressed loss per dollar of category c exposure under scenario s. Conservative because it ignores cross-category interactions.
• Method B (Joint Optimization): Caps set jointly across categories to satisfy scenario constraints while maximizing portfolio utilization ^[20].

Bridge risk (protocol risk, liveness risk, censorship risk) is managed via the same Correlation Framework category caps — no separate bridge-specific risk module exists ^[20]. Stars' exclusivity rules naturally partition risk domains, with Spark covering DeFi/crypto lending and Grove covering RWA.

Grove Stack Single Points of Failure

Grove's operational infrastructure involves multiple layers of legal and custodial intermediation, each representing a potential point of failure:

The Atlas acknowledges the substitution risk explicitly: "Future iterations of the Artifact will specify operational processes owned by Grove" ^[31]. No substitution plan for key counterparties — Janus Henderson, Coinbase Custody, or Centrifuge — is specified in current documentation. The Andromeda RWA portfolio is transitioning to Grove, representing an ongoing operational migration with associated execution risk.

Governance and Decentralization

Sky Protocol's governance system is in transition from the legacy SKY token Executive Vote model toward the Laniakea-specified Core Council structure. Understanding both the current state and the target state is essential for analysts assessing governance risk. The current model concentrates substantial discretionary authority in a small number of actors — Facilitators, Aligned Delegates, and the co-founder — while the Laniakea model specifies a 24-member rotating Council with quantified supermajority thresholds and AI-assisted monitoring. The distance between current practice and target state is significant, and the pace of transition is a key rating variable. S&P's August 2025 B- rating explicitly cited governance centralization as a constraint; the protocol's response has been to accelerate the Core Council formalization and implement the SpellGuard dual-key authorization system.

Current Governance Structure

Sky Protocol governance operates through six Scopes — Governance, Support, Stability, Protocol, Accessibility, and Agent — each defining a domain of protocol authority ^[32]. The SKY token became the sole governance token on May 19, 2025, replacing MKR following the 1:24,000 conversion ^[32].

Voting operates on a continuous approval basis with bi-weekly Executive Votes executing approved parameter changes. The "hat" mechanic establishes a victory threshold: any new proposal must accumulate more SKY votes than the current hat before it can be executed. This design prevents minority proposals from executing but also means that high hat levels can block legitimate parameter changes if voter turnout declines.

The GSM Pause Delay of 24 hours provides a 24-hour window between proposal approval and execution, during which Protego contracts can cancel malicious proposals ^[33]. Emergency exceptions (MOM contracts, Standby Spells, BEAMs) bypass this delay for time-critical risk management.

Scope Advisors and Ecosystem Actors

Scope Facilitators are granted broad discretionary authority, including the ability to "supersede existing Atlas provisions" in their Scope domain, subject to community review ^[32]. This discretionary authority is a key governance centralization concern noted by analysts: a small number of Facilitators can modify effective protocol behavior without requiring full Executive Vote cycles, creating a gap between formal governance parameters and operational reality.

As of February 2026, 11 active Aligned Delegates earn compensation ranging from 48,000 to 400,000 USDS per year, with total compensation capped at 1% of Step 1 Capital under the Atlas ^[32]. Delegates include Bonapublica, PBG, BLUE, Cloaky, and others. The Laniakea target model consolidates the current Facilitator, Aligned Delegate, and Guardian roles into a single "Guardian" role operating within the Core Council structure.

As of February 2026, the protocol's operational contributor base comprises 24 active entities across six governance Scopes ^[41]:

The "Scope Advisor" role referenced in earlier Atlas versions was deprecated in February 2026 Atlas edits. The Emergency Response Group — the rapid-response team authorized to intervene during crisis events — draws from 18 of these entities.

Governance Concentration Concerns

Rune Christensen, the protocol co-founder, holds approximately 9% of governance voting tokens — a figure cited directly in S&P's August 2025 rating report as a governance concentration risk ^[30]. Low voter turnout is flagged as a compounding factor: if a large proportion of SKY tokens are not actively voting, the effective concentration of voting power among active participants is materially higher than the nominal 9%.

The depth of concentration was demonstrated in the November 2024 governance vote on whether to rebrand the protocol back to "Maker." Four whale wallets collectively cast 62,452 MKR — accounting for 98% of all votes in favor of retaining the Sky brand — while only approximately 20 addresses participated in total ^[47][48]. Each of the four whales held roughly 20% of the actual voting weight. A single opposing whale cast nearly all of the 14,864 MKR supporting reversion to Maker. The vote outcome — 79% for Sky, 18.5% for Maker — was determined entirely by five addresses.

The broader token distribution is highly concentrated. Third-party analytics indicate the top three SKY addresses collectively hold approximately 76% of total supply ^[44]. The original venture capital investors — a16z (which purchased 6% of MKR for $15 million in 2018) and Paradigm (which acquired 5.5% for $27.5 million) — both liquidated their governance token positions in July 2023, preceding the Endgame transition ^[42][43]. The VC exits were preceded by public opposition to the Endgame proposal, with a16z voting against the plan despite support from over 80% of the community. The absence of institutional anchor investors with long-term governance alignment is a structural concern for analysts evaluating governance stability.

The token migration from MKR to SKY provides additional context. At launch, only 26% of MKR had migrated to SKY; by September 2025, approximately 61% had converted, with a 1% quarterly penalty on the conversion rate activated on September 18, 2025 to incentivize stragglers. As of February 2026, approximately 81% of MKR has been converted. Coinbase completed its mandatory exchange-held migration in January 2026. The 60.3% SKY staking rate as of Q4 2025 reflects staking-for-yield participation, which is distinct from governance voting — staking for rewards does not require active participation in Executive Votes.

The October 2020 flash loan incident remains the canonical governance attack case study: B.Protocol borrowed 13,000 MKR tokens from dYdX, used them to pass an unauthorized governance vote, and exited before detection ^[33]. While subsequent ESM threshold increases (from 150,000 MKR to 300,000 MKR equivalent) have raised the cost of such attacks, the incident demonstrates that governance security requires both token concentration limits and operational detection speed ^[33].

BA Labs holds the Core Council Risk Advisor role while simultaneously operating under a conflict-of-interest waiver for its work with Ethena — a situation that creates at minimum a perception risk that risk advisory functions may be influenced by commercial relationships with counterparties whose assets are being assessed ^[30].

Governance concentration remains a structural feature that the Laniakea roadmap addresses progressively. The Core Council expansion from 7 initial members to the target 24-member rotating body is the primary mitigation mechanism: a fully staffed Council operating under 16/24 supermajority requirements would structurally prevent any single actor — including the co-founder — from unilateral influence. The SpellGuard dual-key authorization system reduces single-actor risk by requiring both Core Council approval and token holder confirmation for executive actions. The ongoing MKR-to-SKY migration (81% complete) and staking incentive programs are broadening the active token holder base, though meaningful dilution of co-founder voting influence relative to active turnout has not yet occurred. Sky Protocol acknowledges this concern and has committed to the Core Council expansion timeline as the concrete structural response ^[8][20].

Decentralization Progress Since Mid-2025

SKY became the sole governance token on May 19, 2025. Core Council EA1 was formally approved in December 2025 — the first step toward the 24-Guardian target ^[32]. The SpellGuard dual-key system, which requires both top-down Core Council authorization and bottom-up token holder approval for executive actions, represents a structural improvement in the governance attack surface relative to the prior single-key Executive Vote model.

DeFiScan, an independent protocol decentralization assessment tool, rates Sky Protocol at "Stage 0" — the lowest tier — citing insufficient exit windows for upgrades (18-hour GSM delay versus the 7-day minimum required for Stage 1) and high centralization dependency on USDC/Circle ^[44]. The assessment notes that aligned delegates hold a majority of voting power, amplifying the effective concentration beyond nominal token distribution.

The Laniakea whitepaper targets a "99% automated / 1% human judgment" governance model as the steady-state architecture, with human governance reserved for edge cases, emergencies, and parameter updates that exceed the authority of automated systems ^[20].

Structural Changes Since the S&P Rating (August 2025 — February 2026)

The following timeline documents concrete governance and protocol changes executed through Executive Votes since the August 8, 2025 S&P rating, drawn from the on-chain executive spell history ^[53]:

Ranked Delegate participation remained consistent throughout: six delegates — AegisD, BLUE, Bonapublica, Cloaky, Tango, and Sky Staking — maintained greater than 75% participation across all Executive Votes. Monthly compensation for the active delegate cohort totaled approximately $20,000–$24,000 ^[53].

DeFiScan's independent protocol decentralization assessment remains at Stage 0 (unchanged since mid-2025). Upgradeability, Autonomy, and Exit Window are all rated "High risk." The exit window has been reduced from 30 hours to 18 hours (the current GSM Pause Delay), but reaching Stage 1 requires either a 7-day exit window or the implementation of a qualifying security council ^[54].

AI Tools in Governance

The Sentinel Network (Laniakea) introduces three distinct AI-capable formations for protocol governance and operations ^[34]:

• Baseline Sentinel (stl-base): Holds exclusive access to the Execution Engine (pBEAMs). Runs the public Base Strategy. Falls back to Base Strategy if Stream disconnects. Cannot be operated by the same party as the Warden. Available from Phase 9.
• Stream Sentinel (stl-stream): Operated by Ecosystem Actors (DevCos, Trading Firms). Sends trading intent to Baseline for validation — does not hold execution keys. Earns carry only when outperforming the counterfactual Base Strategy: Carry = (Actual PnL - Simulated Baseline PnL) × Performance Fee Ratio ^[34].
• Warden Sentinel (stl-warden): Independent monitoring with halt authority. Can freeze the Execution Engine. Must be operated by independent parties to prevent correlated failure with Baseline. Multiple Wardens per formation reduce Time to Shutdown (TTS) and therefore reduce Operational Risk Capital requirements ^[34].

Skychain is a proposed Sky-native EVM blockchain designed for AI agents — optimized for throughput, with state rent replacing permanent storage, and no human UI trade-offs ^[35]. As of February 2026, Skychain is in exploratory research status; no deployment decisions have been made ^[35].

USDS Freeze Function

The Sky Atlas states: "USDS Stablecoin must be a permissionless and useful currency available to anyone" ^[33]. No token-level blacklist or individual-address freeze mechanism exists in the USDS contract. Protocol-level emergency controls — oracle freeze (OSM_MOM), SparkLend market freeze (FREEZER_MOM), LitePSM halt (SingleLitePsmHaltFactory), SBE breaker (SPLITTER_MOM) — affect protocol operations but do not freeze individual USDS balances or prevent peer-to-peer transfers ^[33].

The Solana LayerZero Freezer Multisig operates on a 2/4 signing threshold and can halt the Solana bridge but does not affect mainnet USDS ^[33]. Adding a token-level freeze function would require a full Executive Vote and would represent a fundamental architectural change to USDS's permissionless design — a change that Laniakea documentation specifically positions as contrary to USDS's identity as a decentralized yield-generating stablecoin ^[20].

Cross-Chain Expansion and Bridge Risk

Sky Protocol's multi-chain expansion significantly increases the complexity of risk management. Each additional chain introduces bridge risk (smart contract vulnerability, validator compromise, liveness failure), liquidity fragmentation risk, and operational risk from maintaining consistent governance signals across multiple execution environments. The current SkyLink architecture handles bridging for USDS, sUSDS, and SKY tokens, while the Spark Liquidity Layer manages cross-chain capital allocation across six chains. Rate limits are the primary on-chain tool constraining cross-chain damage in the event of a bridge exploit or governance failure.

Multi-Chain Deployment

Spark Liquidity Layer covers six chains as of February 2026 ^[36].

SkyLink Bridge Architecture

SkyLink is the protocol's native bridging infrastructure for USDS, sUSDS, and SKY tokens across EVM-compatible chains ^[36]. Base deployment uses the native Optimism Stack bridge. Solana deployment initially used Wormhole NTT and has been approved to migrate to LayerZero OFT as of November 2025, with $824M in cumulative Solana bridge volume and $111.4M USDS outstanding on Solana as of December 2025 ^[36].

Solana-specific rate limits: 10M USDS per day, calculated on a net basis (withdrawals offset deposits within the accounting window) ^[36]. The Token Bridge quorum is 4/7 validators ^[36]. Rate limit decreases on bridge flows are always executable instantly; rate limit increases require a 14-day timelock via BEAMTimeLock ^[20].

Chain Halt Policies

No named "chain halt policy" exists in Atlas or Laniakea documentation. The functional equivalent is achieved through a combination of:

1. Instant rate limit decreases to zero, effectively halting new bridge flows.
2. Solana LayerZero Freezer Multisig (2/4 signing) capable of halting the Solana bridge ^[33].
3. Warden Sentinel halt authority (Phase 9+), which can freeze the Execution Engine controlling bridge flows ^[34].
4. SORL (Second-Order Rate Limit) constraining rate limit increases to 25% per 18-hour window, preventing rapid re-opening after a halt ^[20].

Multi-Chain Liquidity Cascade Prevention

Bridge risk concentration is managed through the Correlation Framework category caps rather than through dedicated bridge risk modules ^[20]. Each chain's exposure is bounded by its applicable correlation category cap. Foreign Primes — which receive bridged assets on non-mainnet chains — do not maintain separate vaults between Prime and Foreign Prime: capital flows directly through ALM Proxy transfers, meaning cross-chain exposure is captured within the originating Prime's collateral portfolio and subject to its CRR requirements.

Per-Prime ASC requirements apply on each chain where a Prime operates, creating per-chain liquidity buffers that must be maintained independently. This means a liquidity stress event on Base cannot drain Ethereum mainnet ASC beyond the cross-chain bridge rate limits.

Catastrophic Failure Safety Architecture

The SparkLend Security Access Multisig operates on a 3/5 signing threshold (VoteWizard, Prism, Jem, Derek, LucasManuel), providing rapid emergency response for SparkLend-specific incidents without requiring a full governance cycle ^[33].

The "instant decrease, constrained increase" invariant is the fundamental asymmetry in Sky Protocol's rate limit and debt ceiling governance: risk exposures can be reduced immediately in response to adverse conditions, but increases must wait through timelocks. Losses are reflected in settlement immediately; gains accrue gradually ^[20].

Spark Liquidity Layer Risk Limits

Spark Liquidity Layer allocations are governed on a per-instance basis through on-chain RateLimits contracts:

• maxAmount: Hard cap on instantaneous allocation to each instance.
• slope: Rate at which allocation can increase over time after a withdrawal.
• SORL: 25% per 18-hour window constrains rate limit increases (Second-Order Rate Limit) ^[20].
• IRL: $100,000 initial rate limit per initialization target ^[20].
• Cooldown: 18-hour cooldown between rate limit increase cycles.

No single aggregate limit across all Liquidity Layer instances is specified in Atlas; governance configures each instance independently.

Smart Contract, Oracle, and Upgrade Risk

Smart contract and oracle risk management involves layered safety mechanisms operating across different time horizons: the 24-hour GSM Pause Delay provides advance notice of parameter changes, MOM contracts provide sub-hour emergency response, and the Emergency Shutdown Module provides a last-resort protocol freeze. Each layer has been activated during historical stress events, providing operational validation of the designs. Chronicle Protocol's oracle infrastructure, securing approximately $12.6B in total value across DeFi as of Q1 2025, represents a systemic single-provider dependency that the OSM's 1-hour delay partially mitigates. The modular Diamond PAU architecture (EIP-2535) allows new capabilities to be added as facets without full contract redeployment — but individual facets still require GSM Pause Delay before activation.

Upgradeable Modules and Authority

The Diamond PAU architecture deploys modular action facets that can be added to Prime and Halo operational surfaces without redeploying the entire contract system ^[20]. All parameter changes — including new facet additions — are subject to the 24-hour GSM Pause Delay unless executed via a designated exception pathway ^[33].

Exceptions to the GSM Pause Delay:

Contract Pausing Matrix

The CLIP_BREAKER_FAB's permissionless activation tier is particularly notable: if a collateral asset's price decreases by more than the preset Breaker Price Tolerance between consecutive oracle updates, any address can trigger the liquidations circuit breaker without requiring governance action ^[33]. This removes the requirement for timely Core Council response in fast-moving market events.

Oracle Architecture

Chronicle Protocol provides oracle infrastructure for Sky Protocol, using Scribe aggregated signatures across 25 validators to achieve approximately 60% gas reduction compared to individual validator submissions ^[37]. As of Q1 2025, Chronicle secures approximately $12.6B in total value across DeFi ^[37].

The Oracle Security Module (OSM) introduces a mandatory 1-hour delay between price submission and vault-visible price. During this window, an oracle manipulation attempt can be detected and the oracle frozen via OSM_MOM before the corrupted price affects vault liquidations. This delay is the primary technical mitigation for oracle manipulation risk.

The Atlas mandates the Chronicle v3 oracle for ETH, STETH, and WBTC "until at least January 1st 2026" ^[33]. This single-provider mandate creates oracle concentration risk: a Chronicle Protocol compromise would affect all major collateral asset pricing simultaneously. The OSM's 1-hour detection window is the primary mitigation.

Delayed Liquidation Loss

The March 2020 Black Thursday event provides the primary historical reference for delayed liquidation loss ^[3]:

Zero-bid auctions resulted from Ethereum network congestion preventing keepers from submitting bids before auctions ended. Liquidations 2.0 (April 2021) replaced English auctions with Dutch auctions (DOG/CLIP mechanism), designed to maintain keeper profitability even under network stress. Current keeper compensation structure: 100 DAI flat tip plus approximately 2–3% chip (percentage of collateral) per liquidation ^[3].

For less-tested collateral types (e.g., staked ETH), the gap risk framework applies: governance models instantaneous price gaps and health factor distributions under stress scenarios, rather than relying solely on the historical auction mechanics that failed on Black Thursday. ETH is classified as having "infinite" pull-to-par (no maturity), meaning it cannot be duration-matched and must be held as unmatched exposure subject to FRTB drawdown capital ^[20].

WSTETH carries a 150–175% liquidation ratio and 13% liquidation penalty (chop) under current parameters, reflecting the additional slippage risk associated with the staked ETH redemption queue during market stress ^[3].

Oracle Concentration

The mandatory single-oracle-provider requirement through January 1, 2026 created a concentrated dependency that has since expired, but governance has not publicly announced a multi-provider transition as of February 2026. Chronicle Protocol's aggregated signature design (25 validators, Scribe mechanism) provides internal redundancy within the oracle network, but represents a single aggregation layer whose contract-level compromise would affect all Chronicle-sourced prices simultaneously.

Regulatory Environment

Regulatory uncertainty is explicitly identified as a rating constraint in S&P's August 2025 assessment. Sky Protocol's position in the regulatory landscape is genuinely novel: USDS lacks a centralized issuer subject to existing stablecoin regulatory frameworks, yet functions economically as a stablecoin at multi-billion-dollar scale. The Laniakea whitepaper explicitly positions USDS's decentralized nature as a differentiator from payment stablecoins — a strategic framing that simultaneously argues for regulatory exemption and presents reputational and enforcement risk if regulators disagree. As of February 2026, no major jurisdiction has issued definitive guidance specifically addressing decentralized yield-bearing stablecoins of the Sky Protocol type.

GENIUS Act (United States)

The GENIUS Act (Guiding and Establishing National Innovation for US Stablecoins) represents the primary US stablecoin regulatory framework under consideration as of February 2026. The Act targets "payment stablecoins" — instruments marketed for payment use and redeemable for US dollars.

The Laniakea whitepaper explicitly positions USDS as "a decentralized yield-generating stablecoin focused on capital formation — distinct from payment stablecoins regulated under frameworks like the GENIUS Act" ^[6]. This framing attempts to place USDS outside the GENIUS Act's definitional scope by emphasizing yield-generation and decentralization rather than payment utility. Whether US regulators will accept this framing is uncertain.

The absence of a centralized issuer is the most significant structural argument for USDS being outside GENIUS Act scope: the Act's reserve requirements, governance disclosures, and redemption obligations assume an identifiable issuer entity. Sky Protocol's distributed governance model has no such entity. However, the practical risk is that regulators may look through legal form to economic function, treating USDS as a payment stablecoin on the basis of its dollar peg and liquidity mechanics regardless of its governance structure ^[30].

MiCA Framework (European Union)

The EU's Markets in Crypto-Assets Regulation (MiCA) establishes stablecoin issuer requirements for e-money tokens and asset-referenced tokens. MiCA requires compliant issuers to maintain 100% liquid reserves, provide redemption at par, meet governance disclosure standards, and obtain regulatory authorization.

DAI was delisted from major EU crypto exchanges including Binance, Kraken, and Coinbase by March 31, 2025 under MiCA's stablecoin provisions. USDS faces the same delisting risk if its regulatory status is not resolved before the July 1, 2026 deadline when all transitional periods expire ^[5]. Sky Protocol's claim that it is "a decentralized, community-driven system" with "no single entity" in control (Laniakea whitepaper) provides the legal theory for MiCA exemption, but regulators retain discretion to apply the framework based on economic substance ^[6].

Circle's USDC — which forms the primary ASC base for Sky Protocol — is a MiCA-compliant e-money token. This creates a paradox: Sky Protocol's reserve infrastructure is held in MiCA-compliant assets, yet its primary stablecoin products may themselves be non-compliant. A regulatory action against USDS in EU markets would reduce the addressable depositor base without necessarily affecting the asset side of the protocol's balance sheet.

Freeze Function Considerations

Multiple regulatory frameworks (GENIUS Act, MiCA, OFAC compliance regimes) may require stablecoin issuers to maintain the capability to freeze specific addresses for sanctions compliance or law enforcement purposes.

USDS does not have a token-level freeze or blacklist capability, consistent with the Sky Atlas requirement that "USDS Stablecoin must be a permissionless and useful currency available to anyone" ^[33]. Protocol-level emergency controls exist but do not affect individual USDS token transfers or balances. Adding a freeze function would require an Executive Vote and would represent a fundamental change to USDS's permissionless design — a change that the Laniakea whitepaper explicitly argues would undermine USDS's value proposition relative to regulated payment stablecoins ^[6].

This creates a structural regulatory tension: frameworks requiring freeze capability are incompatible with USDS's current permissionless architecture unless the protocol introduces a centralized issuer function. Sky Protocol's regulatory strategy appears to rest on jurisdictional differentiation — maintaining permissionless USDS while relying on regulated partners (Coinbase Custody, Centrifuge, Janus Henderson) for the asset side of the balance sheet.

Criticism and Risk Assessment

The S&P B- rating issued on August 8, 2025 — the first credit rating ever assigned to a DeFi protocol — synthesizes the key vulnerabilities identified in the preceding sections into a formal credit opinion. The rating carries a Stable Outlook but explicitly states that an upgrade is unlikely within 12 months. The primary constraints — capital inadequacy, depositor concentration, governance centralization, and regulatory uncertainty — map directly to structural features of Sky Protocol's architecture. This section details each constraint and the protocol's structural responses.

S&P B- Rating Analysis

S&P Global's August 8, 2025 rating assigned USDS/DAI a B- Stablecoin Stability Assessment ^[30]. Key findings:

S&P applied Basel III-equivalent risk weights to assets, treating USDe — a delta-neutral synthetic dollar backed by perpetual futures funding — at the maximum 1,250% risk weight given its novel risk profile and lack of regulatory clarity ^[30]. At this risk weight, Ethena's 11% asset share consumed a disproportionate share of Sky's implied risk-weighted capital.

Capital Inadequacy

Sky Protocol's 0.4% capital ratio — measured as Surplus Buffer (approximately $70M at time of rating) divided by total USDS supply — falls dramatically below traditional bank minimum capital requirements (4–8% for Tier 1 capital under Basel III; 8–10.5% including conservation buffers). S&P acknowledged that Sky was undercapitalized by the protocol's own admission during H1 2025 ^[30].

Several structural features exacerbate the capital shortfall:

1. SBE buybacks: The Smart Burn Engine prioritizes SKY token buybacks over capital retention. Since activation in 2024, the SBE has directed significant protocol income to buybacks rather than Surplus Buffer accumulation. As of December 2025, annualized SKY buybacks totaled approximately $102.2M versus $168M in annualized operational profit — a 61% payout ratio that leaves limited margin for capital building ^[6].
2. Genesis Capital transfers: Transfers of Genesis Capital to bootstrap new Stars (Spark, Grove, Obex) reduce the Surplus Buffer by the transferred amount. Projected total Allocated Genesis Capital of approximately $120M by Q1 2026 exceeds the current Surplus Buffer, creating negative net Aggregate Backstop Capital at current operating levels pending the 25% revenue retention mechanism taking effect ^[8].
3. Maintenance budget: The current Genesis Phase Security and Maintenance Budget consumes 21% of Net Revenue — a figure the Laniakea framework targets to reduce to a permanent 10% cap by end of 2026, which would materially increase capital retention capacity ^[8].

The Laniakea capital formula and Aggregate Backstop Capital target of $125M (with a 25% of net revenue retention floor until the target is met) represent the primary credible path toward capital improvement ^[8]. The Spark Risk Capital Monitor demonstrates partial progress: as of February 2026, Spark holds $36.94M in total risk capital against $25.79M required — a 69.83% encumbrance ratio that provides meaningful headroom at the individual Star level ^[59].

Governance Centralization

The co-founder's 9% token holding, while not a majority, represents the single largest voting bloc in a system where active turnout is low and voting power is fragmented among 11 Aligned Delegates and a broader retail token holder base. In practice, concentrated positions in low-turnout continuous approval voting systems can be decisive for marginal proposals.

Facilitators' statutory discretion to supersede Atlas provisions creates an execution risk that formal governance metrics understate: the parameter-space of protocol behavior is wider than on-chain governance records suggest, because Facilitator actions may not require Executive Vote approval ^[32].

The flash loan precedent (B.Protocol, October 2020) established that economic governance attacks are feasible, and the protocol's mitigation — ESM threshold increase to 300,000 MKR equivalent — addresses the specific borrowing-based attack but not alternative forms of governance manipulation ^[33].

BA Labs's dual role as Core Council Risk Advisor and holder of a conflict-of-interest waiver for Ethena work represents an unresolved structural conflict. The Risk Advisor role involves assessing capital requirements and testing parameters for interim deployments, which directly affects the treatment of Ethena-related exposures ^[30].

Depositor Concentration and Hot Money

The Q1 2025 SSR-driven hot money cycle provides the clearest evidence that depositor concentration risk is not merely theoretical. The protocol raised the SSR to 12.5% to attract capital — primarily via Ethena's sUSDS yield strategy — generating significant inflows. When rates normalized, Ethena redemptions created a net outflow that produced a $5M protocol loss, the first loss quarter in Sky Protocol's history ^[15].

The duration model (see Peg Stability section) attempts to address this risk analytically by calibrating bucket capacity limits to empirical bank run data. However, the model's parameters were estimated ex post using SVB, First Republic, and Credit Suisse data — institutions with diversified depositor bases and longer operating histories than Sky Protocol. Calibration uncertainty for a newer protocol with a narrower depositor composition may mean the model's outputs are systematically optimistic.

No explicit "hot money depositor limits" — e.g., per-counterparty concentration caps on USDS supply — exist in the current Atlas or Laniakea framework. The Correlation Framework's category caps address asset-side concentration; liability-side concentration limits are not yet specified ^[20].

Protocol Responses

The Laniakea risk framework — correlation framework, capital formula, duration model, daily settlement cycle — was published or advanced after the August 2025 rating. The rating therefore reflects the protocol's state before these risk governance frameworks were introduced.

Structural responses underway as of February 2026:

Future Developments

Sky Protocol's forward-looking capital and risk architecture is defined by the Laniakea specification — a multi-phase roadmap that progressively replaces manual governance processes with formula-driven, eventually autonomous, risk management. The transition from the current Genesis Phase to full Laniakea implementation represents the most significant structural evolution since the protocol's founding.

Disclaimer: This section contains forward-looking information based on published protocol roadmaps and governance proposals as of February 2026. Actual implementation timelines, parameter values, and feature availability may differ materially from descriptions herein. The Laniakea specifications are draft documents subject to ongoing revision.

Laniakea Implementation Phases

The Laniakea roadmap comprises eleven phases (0 through 10), progressing from legacy exception handling through full autonomous sentinel-operated governance ^[38]:

Phase 3 (Daily Settlement) is the rate-limiting dependency for Phase 4 and beyond. The LCTS lock/settle behavior is synchronized with the daily settlement window (13:00–16:00 UTC), and srUSDS exchange rate updates occur at the 16:00 Moment of Settlement ^[38].

Daily Settlement Transition

The daily settlement cycle (Phase 3) establishes the canonical timing contract for all subsequent phases ^[39]:

• Active Window: 16:00 UTC to 13:00 UTC (21 hours) — data collection, normal deposits and withdrawals, parameter staging.
• Processing Window (Lock): 13:00 UTC to 16:00 UTC (up to 3 hours) — calculation finalization, prepayments, verification, settlement-critical transactions only.
• Moment of Settlement: 16:00 UTC — new OSRC allocations, duration capacity, srUSDS exchange rate updates take effect; late payment penalties begin accruing.

In Phase 3, GovOps automation performs computations and submits required transactions in the absence of stl-base. All settlement outputs use epoch identifiers and schemas that stl-base will use in Phase 9, ensuring a clean handoff ^[39].

Penalty forgiveness: During the early rollout period, the Core Council retains an explicit forgiveness process for operational failures. Forgiveness must be recorded as a signed governance statement in Synome referencing the original penalty event — preventing silent exception handling while allowing operations to proceed through early implementation difficulties ^[39].

Sentinel Network and AI Governance

Phase 9 deploys the Baseline Sentinel (stl-base) and Warden Sentinels (stl-warden) — the execution and safety planes of sentinel formations ^[34]. stl-base activates sealed-bid auctions for OSRC (Originated Senior Risk Capital) capacity and Duration Bucket capacity, transitioning allocation from Core Council direction to market-based pricing.

Phase 10 adds Stream Sentinels (stl-stream), completing the three-component formation structure. The compounding loop enabled by Stream formations:

flowchart LR
    A["Public Capital"] --> B["Private Intelligence"]
    B --> C["Better Streams"]
    C --> D["More Carry"]
    D --> E["More Intelligence"]
    E --> A

carries potential for accelerating capital efficiency gains but introduces novel concentration risk if Stream operators (Ecosystem Actors) converge on similar strategies ^[34].

Operational Risk Capital requirements for sentinel-era operations are bounded by:

ORC ≥ Rate Limit × TTS

where TTS (Time to Shutdown) is the worst-case detection-to-halt window for Warden Sentinels. Premium warden coverage (3+ independent operators, automated detection) targets TTS of approximately 1 hour, implying ORC of approximately $4.2M for a $100M/day rate limit — a 95.8% ORC reduction versus single manual warden coverage (24-hour TTS requiring $100M ORC for the same rate limit) ^[34].

Generator System

Phase 6 replaces the current per-Prime MCD ilk architecture with a unified USDS Generator PAU connected to Primes through ERC-4626 vault interfaces ^[38]. The architectural transformation:

flowchart LR
    subgraph BEFORE["Before: per-Prime ilks"]
        P1B["Prime 1"] --> I1["ilk 1"] --> MCD1["MCD"]
        P2B["Prime 2"] --> I2["ilk 2"] --> MCD1
        P3B["Prime 3"] --> I3["ilk 3"] --> MCD1
    end

    MCD1 -..->|transforms to| ERC

    subgraph AFTER["After: Generator PAU"]
        P1A["Prime 1"] --> ERC["ERC-4626 Vault"]
        P2A["Prime 2"] --> ERC
        P3A["Prime 3"] --> ERC
        ERC --> GEN["Generator PAU"]
        GEN --> ILK["ilk"] --> MCD2["MCD"]
        GEN --> SR["srUSDS via LCTS"]
    end

This simplifies governance scope separation (Generator vs Prime vs Halo) and provides the stable interface required for the Phase 8 multi-Generator architecture — which will enable independent governance scopes, separate risk capital token structures, and cross-Generator restrictions for different stablecoin types ^[38].

Scaling to $100 Billion

The Endgame's long-stated aspiration is $100 billion in USDS supply — a figure referenced by the protocol's founder as the ultimate target since the original Endgame proposal. No official balance sheet composition breakdown at this scale has been published, but three structural frameworks embedded in Laniakea constrain the feasible asset mix ^[45]:

First, the duration model's structural caps impose a minimum liquidity floor. At the 30-day horizon (Bucket 2), the cumulative cap is 75.18% — meaning under stress, up to 24.8% of the USDS supply could be demanded within 30 days. At $100 billion, this implies a minimum of approximately $25 billion in assets redeemable within 30 days (T-bills, money market funds, PSM stablecoins, and overnight facilities). The protocol cannot allocate below this floor regardless of yield opportunity cost.

Second, the Correlation Framework's category caps prevent over-concentration into any single asset class. Exposure exceeding a category cap faces 100% CRR — dollar-for-dollar capital backing that eliminates leverage benefit. While specific numeric caps are governance-set and not yet published, the framework's example parameters suggest constraints such as "CLOs ≤ 10% of USDS supply" (approximately $10 billion at $100 billion scale).

Third, the Aggregate Backstop Capital target of 1.5% of USDS supply would require a $1.5 billion solvency buffer at $100 billion — a 12x increase from the current $125 million Genesis Phase target. Building this buffer from the 25% net revenue retention mechanism would require sustained profitability over multiple years.

Based on these constraints and current strategic direction, the envelope of feasible compositions at $100 billion USDS supply is approximately:

The $100 billion target originates from Sky founder Rune Christensen's March 2024 statement that the Endgame initiative aims to "raise DAI supply to $100 billion and beyond" ^[52]. The Laniakea TMF formula models $100B and $1T+ as plausible scaling endpoints. No specific timeline exists, but the Laniakea framework mathematically defines what a $100B balance sheet must look like through its constraint system. The following table derives binding floors and ceilings directly from the duration model, ASC parameters, backstop target, and correlation caps:

Constraint-Derived $100B Composition Envelope

These constraints are mutually reinforcing: the duration model's short-end floors set minimum liquidity thresholds, the ASC framework defines peg defense capacity requirements, and the correlation caps limit asset-class concentration — collectively shaping the feasibility envelope for the $100B asset mix.

The current USDS backing composition as of February 2026 provides a baseline for evaluating the $100B trajectory: stablecoins constitute 57.85% ($5.72B), with yield-generating assets totaling $4.17B across crypto lending, T-bills, CLOs, and OTC lending ^[58]. Reaching $100B would require the yield-generating asset base to grow approximately 24x from current levels — implying substantial deepening of institutional credit markets, cross-chain deployment, and new asset categories.

Citigroup's September 2025 stablecoin market forecast projects $1.9 trillion (base case) to $4.0 trillion (bull case) by 2030 ^[50]. At Sky's current approximately 3.3% stablecoin market share, a $1.9 trillion market implies approximately $63 billion in USDS supply by 2030 — suggesting $100 billion would require meaningful market share gains beyond the current trajectory.

2027 Outlook

No official 2027 projections have been published by Sky Protocol or the Sky Frontier Foundation. The following structural milestones are confirmed for 2027 based on the Laniakea implementation documents:

The Genesis Capital mechanism — the temporary backstop system seeding the Agent network — is explicitly designed to be in place "only during 2026 and 2027," after which it phases out as Agents mature and launch liquid tokens ^[8]. The Security and Maintenance budget is targeted to decline from 21% to a permanent 10% cap before the end of 2026, releasing additional revenue for backstop building and buybacks ^[8].

By 2027, the Laniakea roadmap anticipates Phases 1 through 3 as complete (monthly then daily settlement cycles operational) with Phase 4 (LCTS launch, srUSDS token, first Portfolio Halo) likely in progress. The Sentinel Network deployment (Phases 9–10), which would bring AI-operated execution and monitoring to the protocol, remains in later phases without firm dates.

Extrapolating from the 2026 SFF target of $20.6 billion USDS supply and assuming growth deceleration from 124% (2025→2026) to 40–50% (2026→2027), a mid-case 2027 USDS supply estimate falls in the $29–31 billion range. At current net margins, this would imply gross revenue of approximately $900 million to $1.1 billion and net revenue of $300–400 million. The Aggregate Backstop Capital target of 1.5% would require approximately $435–465 million at this supply level.

In September 2025, Sky pitched Hyperliquid's USDH stablecoin with a $8 billion balance sheet and cited its B- S&P credit rating as a credential — the first instance of the protocol leveraging the rating commercially ^[51]. Sky has since used the B- rating as a commercial credential in institutional business development ^[51].

Related Articles

The following Ori articles provide deeper coverage of specific topics referenced in this analysis:

• USDS — Sky Protocol's decentralized yield-generating stablecoin; peg mechanics, issuance, and supply dynamics.
• Sky Savings Rate — The yield mechanism paid to sUSDS holders; monetary policy transmission.
• Peg Stability Module — LitePSM architecture, buffer mechanics, DC-IAM parameters, and March 2023 crisis history.
• Actively Stabilizing Collateral — ASC/DAB framework, per-Prime obligations, and peg defense event mechanics.
• SkyLink — Cross-chain bridging for USDS, sUSDS, and SKY; bridge architecture and rate limits.
• Spark — Spark Prime's SparkLend, Liquidity Layer, and Savings product architecture.
• Grove — Grove's institutional credit and RWA deployment; CLO structure; custodian stack.
• Obex — Institutional Prime / incubator structure; capitalization and backstop terms.
• Sky Stars — Overview of the full Star Prime ecosystem and Genesis Phase structure.
• Smart Burn Engine — SKY buyback mechanism; SBE breaker; capital retention trade-off.
• Emergency Shutdown — ESM architecture; cage() mechanics; collateral claim process.
• Oracle Architecture — Chronicle Protocol; OSM delay; oracle concentration risk.
• Liquidations — Dutch auction mechanism (Liquidations 2.0); keeper economics; gap risk framework.
• Sky Atlas — The governance document encoding current protocol parameters.
• Laniakea — The forward-looking protocol design authored by Rune Christensen.
• MiCA — EU Markets in Crypto-Assets Regulation and implications for USDS/DAI.
• Sky Governance Voting — SKY token voting mechanics, hat system, and Aligned Delegate structure.
• Genesis Capital — Backstop capital system, insolvency defense waterfall, and phase-out mechanics.
• Aligned Delegates — Role, compensation, and voting authority of ADs in current governance.
• Direct Deposit Modules — D3M architecture; SparkLend integration; D3M breaker.
• Sky Direct Exposures — Current asset allocation detail by counterparty and instrument type.

Sources

1. Sky Ecosystem Whitepaper v2.0, December 2025 | GlobalFinTech Series
2. Sky Frontier Foundation 2026 Outlook | PR Newswire
3. S&P Assigns First-Ever Credit Rating to a DeFi Protocol, Rates Sky at B- | CoinDesk
4. S&P Global Rates Sky Protocol B- | The Block
5. S&P Sees No Quick Fix for Sky Protocol's Weak Capital and Centralization | The Defiant
6. Sky Ecosystem Whitepaper | Laniakea
7. Sky Atlas — Constitutional Governance Document | Sky Atlas
8. Genesis Capital | Laniakea
9. Current Accounting | Laniakea
10. Appendix C: Treasury Management Function | Laniakea
11. MSC #5 Settlement Summary, January 2026 — Spark and Grove | Sky Forum
12. Risk Monitoring | Laniakea
13. Sky Protocol Economics Report Q2 2025 | Sky Forum
14. Sky Protocol Reports $168M Annualized Profit Following 61.5% Reduction in Operating Expenses | StableDash
15. DeFi Savings Protocol Sky Slumps to $5M Loss as USDS Interest Payments Wipe Out Profit | CoinDesk
16. Spark Looks to Build a Safe Bridge Between Onchain Capital and TradFi | CoinDesk
17. Capital Formula | Laniakea
18. Risk Capital Ingression | Laniakea
19. Duration Model | Laniakea
20. Correlation Framework | Laniakea
21. Collateralized Lending Risk | Laniakea
22. Asset Type Treatment | Laniakea
23. Grove Announces Launch with $1 Billion Allocation to Tokenized Janus Henderson Anemoy AAA CLO Strategy | BusinessWire
24. Sky's Grove Expands to Avalanche with $250M RWA Plan | CoinDesk
25. Actively Stabilizing Collateral | Laniakea
26. Lite Peg Stability Module Parameter Values — A.3.3.2.7.1.1.2 | Sky Atlas
27. PSM Breaker Exception — A.1.9.3.2.9 | Sky Atlas
28. Minimum ASC Requirement — A.3.3.2.2 | Sky Atlas
29. SP-BEAM Rate Authority — A.1.9.3.2.10.2 | Sky Atlas
30. S&P Stablecoin Stability Assessment: USDS/DAI | S&P Global
31. Spark / Grove Program Accords — A.2.8.2.2.2.4 | Sky Atlas
32. Six Scopes — A.0.1.1.13 | Sky Atlas
33. GSM Pause Delay — A.1.9.3.1.2 | Sky Atlas
34. Sentinel Network | Laniakea
35. Skychain | Laniakea
36. Solana Bridge Rate Limit — A.1.9.4.1.3.2.2 | Sky Atlas
37. Chronicle Protocol | Chronicle Labs
38. Implementation Roadmap | Laniakea
39. Phase 3: Daily Settlement | Laniakea
40. Announcing sky.money USDC & USDT Risk Capital Vaults | Morpho Forum
41. Sky Fusion — Active Ecosystem Actors and Scope Facilitators
42. Venture Capital Firm a16z Unloads $7M of MKR Tokens | CoinDesk
43. Paradigm Moves $3.5M in MakerDAO's MKR Tokens | CoinDesk
44. Sky Protocol Stage 0 Governance Analysis | DeFiScan
45. MakerDAO's Endgame Plan in Five Steps | DL News
46. USDT TVL on Spark Surpasses $900 Million | BlockchainReporter
47. Sky Whales Kill Plan to Rebrand Protocol Back to Maker | CoinTelegraph
48. Just Four Entities Account for Nearly All Votes to Keep Sky | The Block
49. RWA Vault Debt Ceiling Reductions, 6s Capital Stability Fee Increase | Sky Governance
50. Stablecoin Market Could Reach $4 Trillion by 2030 | CoinDesk / Citi
51. Sky Pitches GENIUS-Compliant USDH With $8B Balance Sheet | CoinDesk
52. MakerDAO Endgame to Launch, Aims for $100 Billion DAI to Rival Tether | CoinTelegraph
53. Sky Ecosystem Executive Votes | GitHub
54. Sky Protocol Stage 0 Analysis | DeFiScan
55. Exploring the Sky Ecosystem: Collateral Portfolio Analysis | Block Analitica
56. Exploring the Sky Ecosystem: Financial Performance Analysis | Block Analitica
57. Protocol Economics Report Q2 2025 | Sky Forum
58. Sky Risk & Analytics Dashboard — Supply Breakdown | Sky — USDS backing composition as of February 25, 2026
59. Spark Risk Capital & Requirements Monitor | Sky — Spark required vs. held risk capital as of February 25, 2026
60. Grove Risk Capital & Requirements Monitor | Sky — Grove required vs. held risk capital as of February 25, 2026