Summary
Pharos Watch monitors crvUSD in production today: Safety, Peg, and Liquidity scores, real-time depeg alerting, mint/burn flow, and dependency tracking are live at pharos.watch/stablecoin/crvusd-curve/. On crvUSD, continuity is not a promise the DAO has to trust: the monitoring already runs in production, in public.
We are bidding Scope 1 only - crvUSD / Mint Markets Risk - as a full-service mandate with a tooling-and-handoff core. If awarded, we will stand up a dedicated Curve-specific instance at curve.pharos.watch: publicly accessible, MIT-licensed, and built so the DAO keeps working infrastructure whether or not we remain the provider. We are not bidding Scope 2 and would welcome a co-provider on it.
Budget: $350,000 per year, split $100,000 in stablecoins + $250,000 in CRV, with 100% of received CRV permanently locked as veCRV if the DAO honors that split, and a 3-month paid trial with an explicit go/no-go checkpoint against the KPIs in Section 3.
1. Scope covered
Committed scope: crvUSD / Mint Markets Risk, full service. This covers pre-vote assessment of new mint markets and PegKeeper pools; continuous monitoring of active markets, PegKeepers, and DAO credit lines where relevant; risk-driven parameter recommendations; critical-stress detection and escalation to the DAO and the eDAO (Curve’s emergency DAO); risk-monitoring support for the Llamalend v1 deprecation and v2 rollout; support for governance posts and executable payloads; and semiannual public reporting on crvUSD market health.
Our read of the immediate risk picture: crvUSD’s peg defense is asymmetric - PegKeepers absorb upside up to their caps, but can only defend downside by withdrawing crvUSD they previously deposited - and YieldBasis (YB) growth now dominates how that inventory builds and drains. The October 10, 2025 crash showed how correlated collateral moves feed through soft liquidation into mint-market bad debt. Day-one priorities: PegKeeper inventory and capacity under YB growth, correlated-collateral stress on mint markets, borrow-rate behavior under peg stress, and continuity on the open items of the crvUSD/YB workstream, alongside v1 deprecation support.
Not bidding: Scope 2 (Llamalend / Isolated Markets Risk). Full Scope 2 coverage requires a scenario stress-testing engine for isolated-market solvency across Llamalend v2’s arbitrary market pairings and new collateral classes (LP tokens, Pendle PTs). That is a different discipline from peg-stability and risk monitoring, currently our core expertise.
We recommend awarding Scope 2 to a team whose core product is lending-market simulation, and we commit to sharing crvUSD collateral, liquidity, and dependency data with that team wherever the scopes intersect: crvUSD used as Llamalend collateral, shared oracle dependencies, and PegKeeper-adjacent liquidity.
Live today vs. formalized under the mandate
Live in production against crvUSD now:
- Safety Score: composite risk rating (A–F / 0–100) synthesizing peg, liquidity, collateral, and dependency risk.
- Peg Score: real-time peg-deviation tracking and peg-health classification.
- Liquidity Score: pool-level depth, concentration, withdrawal risk, and one-sided-liquidity detection.
- DEWS (Depeg Early-Warning System): four-tier real-time alerting with Telegram delivery (Calm, Watch, Alert, Warning).
- Reserve Sync: live crvUSD collateral composition, read on-chain each cycle from every LLAMMA market’s band balances (controllers discovered via the ControllerFactory), with the soft-liquidated crvUSD stock retained in snapshots.
- DDR (Depeg Duration Resolver): active-depeg outcome classification (RECOVERY LIKELY / AT RISK / RECOVERY UNLIKELY).
- Depeg Tracker: historical and live depeg events with duration, severity, and root-cause analysis.
- Mint/Burn Flow: supply-side stress signals from minting and burning activity.
- Compliance Tracker: regulatory and MiCA status per issuer.
- Dependency Map: cross-protocol dependency-risk mapping.
- A smart alert-dispatch infrastructure with PharosWatchBot, which can be repurposed and adapted to the Curve DAO’s needs.
Mint-market scope is Ethereum mainnet: the LLAMMA controllers (WBTC, cbBTC, wstETH, sfrxETH, weETH, tBTC, and ETH collateral) and the PegKeeper pools paired with USDC, USDT, PYUSD, frxUSD, and GHO. Around that core we track crvUSD as a token across 11 deployments on Ethereum plus 10 further chains with context for bridged liquidity and dependency risk, with freeze risk flagged through upstream custody (BitGo/Coinbase for WBTC/cbBTC).
Formalized under the mandate (the data inputs largely exist; the packaged deliverable does not yet):
- Pre-vote risk reports: a structured, repeatable assessment template for new mint markets and PegKeeper pools, with explicit parameter recommendations and a turnaround SLA aligned to governance timelines.
- Parameter-recommendation layer: recommendations on
A,fee,loan_discount,liquidation_discount, debt ceilings, PegKeeper caps, and monetary policy: the AggMonetaryPolicy rate response is crvUSD’s primary peg-defense lever, and recommending mint-market parameters without modeling it would be incomplete. Mint markets are LLAMMA markets, so this layer is built LLAMMA-aware. First step: evaluate LlamaRisk’s LLAMMA Simulator from the handoff materials for reuse in discount calibration — the trial commits us to a public verdict on that evaluation plus at least one market calibrated end-to-end (Section 10). If the Simulator is not cleanly reusable, we build our own calibration tooling; the budget and timeline assume that risk. - Oracle-dependency catalog per market: feeds in use, update mechanics, manipulation surface, staleness, and failure modes, including LLAMMA’s own EMA-smoothed, AMM-referential internal oracle, whose smoothing and self-reference create a manipulation and lag surface distinct from a plain external feed. Today this is implicit in our peg and price monitoring; it becomes a named, published deliverable.
- Written public reporting: periodic narrative reports on top of the live dashboards.
- Governance-payload support: a designated contributor drafting Curve governance posts and constructing and verifying on-chain parameter-change payloads.
We would rather the DAO see this delta now than discover it in month two.
2. Optional responsibilities
Committed: public dashboards and frontend risk indicators, delivered through curve.pharos.watch (Section 5) — Safety/Peg/Liquidity scores, DEWS state, PegKeeper health, and mint/burn flow, suitable for embedding in the Curve frontend and for user-facing risk disclosures. The incremental cost is low because the underlying surface already runs.
Available on request: gauge-proposal review and identification of stale, inefficient, or low-quality gauges; CRV emissions justification analysis; and BD support with asset pre-screening, a natural fit given we already screen 344 stablecoins across 108 chains. If the DAO wants any of these committed, we will price and staff them explicitly.
3. KPIs and reporting cadence
- Pre-vote coverage: ≥95% of in-scope mint-market and PegKeeper proposals receive a published risk report before the DAO vote.
- Monitoring coverage: 100% of active in-scope markets and PegKeeper pools under continuous automated monitoring.
- Critical response: human escalation to the DAO/eDAO channel within 4 hours of a critical trigger, with a risk read and a recommended action. Triggers are DEWS Warning tier from day one, joined by solvency-side triggers (soft-liquidation losses, health-distribution deterioration, PegKeeper depletion) as the trial instrumentation ships (Sections 5 and 10).
- Escalation quality: ≥90% of escalations include a specific, actionable recommendation tracked to resolution.
- Miss accounting: every material crvUSD stress event (bad debt, depeg, PegKeeper depletion) receives a public post-incident review stating whether our monitoring flagged it in advance; misses are reported as misses.
- Reporting: quarterly progress updates against these KPIs, plus semiannual crvUSD market-health reports.
- Reusable deliverables: ≥6 documented, reusable assets delivered under perpetual license per mandate period (enumerated in Section 7).
We report all KPIs publicly, so the DAO can audit our performance against our own commitments.
4. Team
Myself, TokenBrice creator and operator of Pharos, leading the mandate. Analyzing DeFi protocols publicly since 2018, active since MakerDAO’s early days. Directly relevant roles:
- GHO Liquidity Committee (October 2023 – February 2024): joined at the Aave community’s request to help repeg GHO - peg monitoring, liquidity strategy, and governance escalation for a CDP stablecoin.
- Liquity: Liquidity Engineer for 18+ months - operational experience in collateral-backed stablecoin risk, liquidity mechanics, and lending-protocol operations.
- DeFiScan: strategy and growth for a decentralization-assessment framework applied to major protocols.
- The DeFi Collective: co-founder and board member of the Swiss non-profit (founded October 2023), with treasury-management responsibilities.
I personally owns the quantitative modeling work under this mandate: discount calibration, debt-ceiling sizing, and rate-policy analysis. He designed and built Pharos’s scoring and classification engines end-to-end, and the trial (Section 10) is structured so the DAO judges delivered calibration work rather than a résumé.
Ike: Pharos Champion, and the front line of the product: content, user support and feedback, outreach, and lead exploration, freeing time for myself to focus on building. Currently part-time; moves to full-time on day one of the mandate, with a natural lane in written reporting and DAO-facing communication.
Staffing plan: the mandate starts with two people, Ike now full-time, and recruitment of a third team member begins as soon as the proposal passes. The trial period (Section 10) is sized so the DAO can judge delivery under this staffing before committing to a full term.
Two properties of Pharos bear on this bid. It is MIT open-source (github.com/TokenBrice/pharos-watch) and transparently funded (pharos.watch/funding/): the DAO can audit how a Safety Score is computed or a DEWS threshold is set. And it is independent: our published crvUSD classification (CeFi-Dependent / Crypto-Collateralized, with freeze risk flagged via BitGo/Coinbase custody on WBTC/cbBTC) is the kind of unflattering but honest call that independence requires; and it stays on the dashboard while we submit this bid.
Disclosure: I am also one of the three founders of Polaris, a decentralized-stablecoin project that is pre-launch and in development, with no live token or markets. Pharos’s open-source methodology means any rating it publishes can be audited against the same rules applied to every other asset. Should Polaris ever intersect this mandate’s scope - as proposed collateral, a PegKeeper pairing, or any assessment target - the conflict will be handled in full transparency: we will disclose it in the relevant report, recuse from both the pre-vote recommendation and the ongoing monitoring judgments it touches, and defer to an independent reviewer designated by the DAO.
5. Tooling, models, monitoring approach, and dependencies
Continuous automated monitoring is the base layer: every active crvUSD market and PegKeeper pool scored across peg, liquidity, collateral, and dependency dimensions, with state changes surfaced through DEWS’s four alert tiers in real time. Analyst judgment sits on top: escalations, pre-vote reports, parameter recommendations, and written reports are authored by the team, informed by the automated layer.
curve.pharos.watch. Under the mandate we fork our production stack into a dedicated Curve instance: public, MIT-licensed, focused on crvUSD mint markets, PegKeepers, and the mandate deliverables (oracle catalog, pre-vote reports, parameter dashboards). Because it is open source, it remains useful to Curve contributors, a successor provider, or a Scope 2 co-provider after our mandate ends. Reusable, transparent risk infrastructure is mandate objective 5; this is our concrete implementation of it.
Escalation mechanics. DEWS alerting is automated and runs 24/7 — critical triggers page the team the moment they fire and are simultaneously delivered to the DAO/eDAO escalation channel. The 4-hour KPI covers the human layer: an on-call rotation across the team (two people at mandate start, three after the hire) delivers an initial risk read and recommended action within the SLA.
To be built or formalized: the LLAMMA-aware parameter-recommendation layer (including the LlamaRisk Simulator reuse evaluation), the controller-level solvency instrumentation below, the oracle-dependency catalog, the pre-vote report pipeline, and the written-report workflow — all additive to the live data layer. We will start by reviewing LlamaRisk’s handoff materials — reports, simulations, models, dashboards, and repositories — and classifying what is reusable, what should be deprecated, and what should be rebuilt, rather than rebuilding by default.
Dependencies: on-chain data via RPC/indexing for Curve pools, crvUSD mint markets, and PegKeepers; the oracle feeds referenced by each market; and market and liquidity data for collateral and PegKeeper-paired assets. Where Swiss Stake or the backend team operate authoritative data infrastructure, we prefer consuming it to duplicating it.
Limitations: Today we read every LLAMMA market’s band balances on-chain each cycle: live collateral composition per market, plus the soft-liquidated crvUSD stock retained in snapshots. We do not yet surface per-band collateral distributions (read, but aggregated), per-controller health distributions, live debt and collateralization ratios, on-chain PegKeeper state, or borrow-rate observation. Those are committed trial deliverables (Section 10), and the solvency escalation triggers in KPI 3 activate as they land. Full isolated-market scenario simulation for Llamalend is the Scope 2 boundary we do not bid. Consistent with the call for proposals (CFP), we own the quality of our analysis, models, alerts, and recommendations; the DAO and eDAO retain final decisions and execution.
6. Operated services during the mandate
- Continuous automated monitoring of 100% of active crvUSD markets and PegKeeper pools.
- DEWS real-time alerting, with a DAO/eDAO escalation channel and Telegram delivery for Curve contributors.
- Pre-vote risk assessment for in-scope proposals within the review window (Section 9).
- Parameter-recommendation service for active markets and PegKeepers.
- Incident response: threshold-driven escalation per runbooks, with concrete recommendations under critical stress.
- Public reporting: quarterly KPI updates and semiannual market-health reports.
- curve.pharos.watch: operated, maintained, and extended throughout the mandate.
7. DAO-owned or DAO-accessible deliverables after the mandate
- curve.pharos.watch: the full Curve-specific monitoring instance, MIT-licensed, forkable and operable without us (a Pharos API key with sufficient req/min limits and no-expiry will be provided to the maintainer).
- Documented crvUSD monitoring models and scoring methodology (Safety/Peg/Liquidity, DDR, mint/burn flow as applied to crvUSD).
- DEWS alert-threshold configuration and incident-response runbook, operable by a Curve contributor.
- Pre-vote risk-report template, plus all reports produced during the mandate.
- Oracle-dependency catalog for crvUSD markets.
- Parameter-recommendation model documentation, including assumptions and limitations.
- API access for integrators.
We believe handoff survivability to be a first-class deliverable because the DAO has just lived through a provider departure.
8. Ownership, access terms, documentation, and handoff plan
The DAO receives a perpetual, non-exclusive license to all models and code delivered under the mandate, plus operational access, as the CFP specifies; Pharos retains ownership and the right to commercialize the tooling to other DAOs and TradFi/DeFi actors. In practice, MIT licensing exceeds those terms: the Curve instance is forkable by anyone, including a successor provider, without our cooperation.
During the mandate, designated Curve contributors receive operational access to the monitoring stack, the escalation channel, dashboards, and the API. Models, thresholds, assumptions, limitations, and runbooks are documented for reuse by Curve contributors, and deliverables are handed over progressively during the mandate rather than at the end. The handoff covers operations, not just code: RPC and data-feed configuration, alert-threshold values, and deployment runbooks transfer alongside the repositories, so “operable without us” means exactly that. If the mandate ends or is not renewed, we provide a defined transition window with runbook walkthroughs for whoever the DAO designates.
9. Expected coordination needs with Curve contributors
- Swiss Stake / backend team: agree data interfaces early (RPC/indexing, oracle feed references, PegKeeper and mint-market state). We consume authoritative infrastructure where it exists and are open to the hybrid model where monitoring is shared with existing contributors.
- Governance timeline: the 95% pre-vote KPI needs pipeline visibility. We ask for a minimum 5-business-day review window between a proposal reaching in-scope status and its vote, and early notice of new mint-market and PegKeeper candidates.
- eDAO: an agreed escalation channel and contact protocol for critical triggers, so the 4-hour SLA is actionable on both sides.
- Scope 2 co-provider: data sharing and coordination on shared oracle and collateral risk, per Section 1.
- LlamaRisk handoff: access to the transition materials referenced in Section 5.
10. Budget request, payment structure, and proposed trial
Budget: $350,000 per year, split $100,000 in stablecoins + $250,000 in CRV.
The stablecoin component covers the mandate’s fixed operating costs - the two salaried team members (Ike full-time plus the third hire) and infrastructure. The CRV component carries everything else, and it exists to make an unusual commitment possible:
If the DAO honors this split, 100% of the CRV we receive is permanently maintained as veCRV. The stable component exists precisely so that no CRV ever needs to be sold. Under this structure, the mandate produces zero CRV sell pressure - the CRV compensation is entirely permanently-locked veCRV in the pharos-watch.eth wallet - and Pharos accumulates a growing, permanent governance stake aligned with the DAO.
For context: LlamaRisk’s most recent renewal ask (March 2026) was $360,000 per year in crvUSD plus a 2,500,000 CRV vest, for a mandate substantially broader than both risk scopes combined. Our cash ask is under a third of that, for one scope, on infrastructure that already runs.
Alternative, at the DAO’s preference: a pure CRV linear vest matching prior precedent, same total, with the permanent veCRV commitment at 50% - fixed salary costs would then have to come from the liquid half.
Receiving address: the Pharos multisig (0x5d698362EDb8AEa1C2b2483096BDeE3265D860DB / pharos-watch.eth).
Trial: 3 months, paid pro-rata ($25,000 in stablecoins + $62,500 in CRV), with an explicit go/no-go checkpoint against the Section 3 KPIs. The trial’s flagship deliverables are the two capabilities this proposal asks to be trusted on:
- A public verdict on the LLAMMA Simulator handoff - reusable or rebuilt - with at least one mint market’s parameters calibrated end-to-end and published.
- Controller-level solvency instrumentation live on curve.pharos.watch: per-band collateral distributions, health distributions, live debt and collateralization ratios, PegKeeper state, and borrow-rate observation, wired into the KPI-3 escalation triggers.
The trial lets the DAO evaluate delivered coverage: pre-vote reports, escalation quality, response times, before committing to a full term. If we do not hit the KPIs in the trial, the DAO should not renew, and we would not expect it to.
Author’s note. Many of you around here already know me. I’ve been involved with Curve one way or another since close to inception, and helped hundreds, if not thousands, understand its core mechanisms, especially in the French-speaking community. I’ve been critical of Curve’s shortcomings when warranted, but I have always admired that Curve and veCRV defined what tokenomics should be, to me: fully onchain, immutable, and granting governance participants tangible power over revenue streams. If this proposal passes, the Pharos team and I will be honored to play a more active part in Curve’s next chapter.
— TokenBrice, on behalf of the small but eager to grow Pharos Watch team
pharos.watch · crvUSD coverage · github.com/TokenBrice/pharos-watch · funding transparency