Deterministic DeFi vault risk scores with inspectable evidence trails, delivered as an API. Same facts in, same score out — every response can be re-derived from its snapshot. Gate listings, power deposit flows, and give your compliance team a methodology they can actually defend.
GET /v1/scores/morpho-steakhouse-usdc ─── 200 OK ──────────────────────────{ "vault": "morpho-steakhouse-usdc", "score": 1.8, // 1.0 lowest risk — 5.0 "band": "low", "factors": { "technical": "low", "financial": "low", "operational": "low", "systemic": "moderate" … }, "kill_switches_tripped": 0, "evidence": "/v1/facts/morpho-steakhouse-usdc", "engine_version": "2.4.1", "facts_snapshot": "0x3f9a…c21e" // reproducible }
Everything is computed by the same deterministic core: fixed, versioned rules over verified facts. Same facts in, same answer out — every response carries the engine version and a facts snapshot so any score can be reproduced exactly.
Eight weighted factors — technical, financial, operational, systemic, complexity, frontend, legal, chain — plus 14 kill switches that floor the score when tripped.
deterministicThe verified facts under every score: audits, oracle setup, governance, incident history. Human-reviewed before promotion. Missing facts say “pending” — never fabricated.
auditableBalance and policy parameters in, scored portfolio out — risk ceilings, concentration caps, rebalance bands. Planned for institutional partners; not available yet.
future featurePost-Stream-Finance, integrators don't buy grades — they buy accountability. ClariQ scores are pure functions over inspectable facts, not model output. How the facts get found and verified is its own discipline — the facts pipeline.
Scores are pure functions of verified facts: fixed rules, versioned weights, no model output in the number. Re-run any score from its facts snapshot and get the same answer.
An unverified fact renders as “pending,” and the score says so. A risk engine that guesses is a liability.
Fixed factor framework, versioned rules, changelog on every revision. Written for the compliance officer who has to sign off.
Protocols cannot pay for their own scores. Our customers are the businesses that rely on the scores — never their subjects.
Every vault is scored across the same eight dimensions. Weights are fixed and versioned; a single tripped kill switch — an unverifiable contract, an anonymous team holding upgrade keys — floors the score regardless of how well everything else reads.
Audit coverage, contract age, upgradeability, oracle design
Liquidity depth, collateral quality, yield source integrity
Team, governance process, incident response record
Counterparty webs, rehypothecation, contagion paths
Strategy legibility, dependency depth, moving parts
Deposit-path integrity, DNS posture, signing surface
Entity structure, jurisdiction, sanctions exposure
Sequencer trust, bridge dependencies, data availability
Coverage is a machine problem. ClariQ uses AI where it compounds — reading, monitoring, and drafting evidence across 15 chains, continuously — and keeps it out of the one place it doesn't belong: the number.
AI systems read audit reports, governance forums, docs, and on-chain changes around the clock, and draft candidate facts — each with its source attached. It's how a small team maintains 94 protocol fact sets across 15 chains without thinning the evidence.
No candidate fact can touch a score until an analyst has confirmed it against its source. That check is true-or-false — is the timelock really 48 hours? — never a judgment call on risk. Promoted facts keep their provenance: the source field on every entry in /v1/facts is that review, made inspectable.
Verified facts go into fixed, versioned rules. Same facts in, same score out; the engine version and facts snapshot hash on every response mean you can re-run the math, not take our word for it.
Every major embedded-yield launch shipped with a named third-party risk layer. The top curators serve the top exchanges — ClariQ serves everyone else.
Adding an Earn tab? Gate deposits on live risk scores and show users the evidence — without hiring a DeFi risk desk.
Stablecoin yield with concentration caps, risk ceilings, and a methodology document your regulator can read.
Responses are machine-readable and reproducible by design — an agent can act on a score and cite exactly why.
The ClariQ Terminal is not a separate product — it is a live demonstration of the same risk engine partners license. Open vault scores and the verified evidence behind them across 15 chains, then embed that engine via API. Free during beta as an evaluation surface.
Pricing is an assessment of customer needs and integration requirements — not a one-size public rate card. Public vault ratings will always have a free tier.
Start the conversationWe're onboarding a small number of integration partners while the API is in beta — early partners get direct input on the roadmap and NDA access to the full methodology.
Book an integration call