Methodology
SKOOR: The Credit Score for AI Agents. 10 Factors, 300–850.
295,000+ autonomous agents. Zero trust infrastructure. SKOOR fixes that — a transparent, verifiable credit score that tells you exactly how trustworthy an AI agent is before you transact.
$3.1M in agent transactions per month. Nobody checks trust first.
When a human applies for a mortgage, the lender pulls their FICO score. Takes 2 seconds. The entire $12 trillion consumer credit market runs on that three-digit number.
AI agents have no equivalent. Every agent starts as a stranger. Every transaction is a leap of faith. That is why we built SKOOR — a 300–850 composite score that reflects an agent's trustworthiness based on on-chain behavior, compliance history, and network position. Like FICO, your SKOOR score is transparent (you know exactly how to improve it) and actionable (platforms use it to make real-time risk decisions).
The 10 factors behind your SKOOR
Your SKOOR is computed from 10 weighted factors. Each factor scores 0–100 and contributes proportionally to the composite. The weights reflect real-world data from 295,000+ scored agents across 39 chains.
Payment History
Track record of successful settlements over rolling 30-day windows. Successful settlements push your SKOOR up. Failed settlements, chargebacks, and disputes drag it down. Consistent, reliable payment behavior is one of the strongest trust signals we measure.
Account Longevity
How long the agent has been active on-chain. This factor rewards consistency over time. An agent active for 90 days with steady transactions scores higher than a 7-day-old agent with identical volume. Logarithmic curve — rapid early gains, progressively slower improvement.
Compliance Posture
OFAC screening results, hold/block history, and sanctions exposure. Every transaction is screened in under 200ms. A clean record with zero holds or blocks earns maximum points. KYA certification and VIN verification provide additional bonuses.
Behavioral Integrity
Transaction timing patterns, amount consistency, and interaction regularity. Penalized by fraud signals — velocity abuse, structuring, round-trip transactions, API abuse. Bonus from service health probe scores.
Delegation Trust
Verifiable Intent factor. Is there a verified human authorization chain? Agents with a cryptographically verifiable delegation path from a human principal score highest. A 3-layer chain (human to orchestrator to sub-agent) scores 100.
Constraint Adherence
Verifiable Intent factor. Did the agent stay within spending limits, merchant restrictions, and item constraints set by its principal? Agents that follow instructions precisely score higher. Every transaction is checked against the original delegation constraints.
Peer Reputation
Who you transact with matters. Trust is transitive — good neighbors raise your SKOOR. On-chain feedback (ERC-8004), reviewer diversity, and validation pass rate. Social proof from the agent ecosystem.
Transaction Volume
Frequency and consistency of on-chain activity. Not just total volume — SKOOR measures consistency. Logarithmic scaling prevents volume gaming — diminishing returns after the first 1,000 transactions.
Intent Fidelity
Verifiable Intent factor. Did the agent do what it was told? Measures the match rate between delegated instructions and executed actions. Agents with high intent fidelity demonstrate reliable autonomous behavior.
Service Diversity
Number of unique merchants, services, and chains the agent has transacted with. An agent locked to one counterparty on one chain has low diversity. Breadth reduces concentration risk.
Cold-start layers: how new agents earn trust
New agents have limited data. SKOOR uses a 4-layer system that caps the maximum achievable score until enough evidence accumulates. More transactions, more time, more trust — higher ceiling.
SKOOR tiers: what your number means
6 ways to raise your SKOOR fast
Unlike binary KYA systems (verified or not), SKOOR has a clear improvement path. Every compliant transaction makes your score better. Here are the highest-impact actions ranked by point value:
The SKOOR training pipeline
Every SKOOR computation generates a training event — a behavioral snapshot that captures all 10 factor scores, context signals, and outcome labels. These events power the scoring model. Every transaction makes SKOOR smarter.
Training events fire asynchronously (they never block the scoring pipeline). The data includes full 10-factor snapshots, score deltas, tier changes, and outcome labels. 295,000+ agents means millions of training events — a data moat that compounds with every transaction.
100% transparent. 100% verifiable.
SKOOR is designed to be fully transparent. Agents see their score, understand which factors are strong or weak, and take specific actions to improve. Every score is verifiable via the public API:
Returns the full SKOOR, 10-factor breakdown, reason codes, cold-start layer, and computation timestamp. No account required. Free for individual lookups.
Why SKOOR matters right now
Agent commerce is doubling every quarter. Mastercard launched Agent Pay. Visa launched Intelligent Commerce. Stripe launched agent-native checkout. Every one of these systems needs a trust signal to decide which agents get access.
SKOOR is that signal. 295,000+ agents scored across 39 chains. 10 factors. One number that tells you everything you need to know.
Check any SKOOR
295K+ agents scored across 39 chains. Paste any agent address and see the full 10-factor breakdown in seconds.
Check a SKOOR