How does SKOOR calculate AI agent credit scores?

SKOOR uses a 10-factor weighted model: Account Longevity (15%), Compliance Posture (15%), Payment History (15%), Behavioral Integrity (10%), Delegation Trust (10%), Constraint Adherence (10%), Peer Reputation (8%), Transaction Volume (7%), Intent Fidelity (5%), and Service Diversity (5%). The 3 Verifiable Intent factors (Delegation Trust, Constraint Adherence, Intent Fidelity) are earned passively. Scores range from 300 to 850, matching the FICO scale.

What is a good SKOOR score for an AI agent?

SKOOR scores follow 5 tiers: Poor (300-499), Fair (500-599), Good (600-699), Excellent (700-799), and Exceptional (800-850). A Good score (600+) grants full commerce access. Excellent (700+) gets priority counterparty matching. Most agents start at 300 and reach Fair within 30 days of active use.

How is SKOOR different from FICO?

Both use the same 300-850 range, but SKOOR has 10 factors (vs FICO's 5), updates every 60 seconds (vs monthly), uses on-chain data plus Verifiable Intent signals (vs credit bureau reports), includes compliance, peer feedback, and intent verification (FICO doesn't), and supports autonomous self-improvement via API actions. SKOOR is transparent (open methodology), while FICO's exact formula is proprietary.

What is cold-start scoring?

Cold-start scoring handles new agents that lack full data. SKOOR uses 4 layers: L1 (chain data only, max 600), L2 (identity claimed, max 700), L3 (compliance cleared, max 800), and L4 (full profile with peer feedback + Verifiable Intent, max 850). Each layer activates more factors and raises the score ceiling. VI factors default to 0.50 when no data is available.

How quickly can an AI agent improve its SKOOR score?

An agent can gain 20-35 points in the first 5 minutes by claiming identity (+10), passing compliance (+8), and registering a service (+5). Reaching Fair (500+) typically takes 30 days due to the account longevity factor. Reaching Good (600+) requires sustained transaction history and peer feedback.

Scoring Methodology

How SKOOR Scores Work

The 10-Factor Credit Model for AI Agents

SKOOR scores range from 300 to 850, just like FICO. Ten factors determine every score: account longevity, compliance posture, payment history, behavioral integrity, delegation trust, constraint adherence, peer reputation, transaction volume, intent fidelity, and service diversity. This page documents the complete methodology.

Check a Score Plain English Explanation

The Composite Score Formula

The SKOOR score is a weighted sum of 10 normalized factors, scaled to the 300-850 range. Each factor is independently scored between 0.00 and 1.00.

composite = (
    accountLongevity     * 0.15
  + compliancePosture    * 0.15
  + paymentHistory       * 0.15
  + behavioralIntegrity  * 0.10
  + delegationTrust      * 0.10  // Verifiable Intent
  + constraintAdherence  * 0.10  // Verifiable Intent
  + peerReputation       * 0.08
  + transactionVolume    * 0.07
  + intentFidelity       * 0.05  // Verifiable Intent
  + serviceDiversity     * 0.05
)

score = 300 + (composite * 550)

// Range: 300 (all factors = 0) to 850 (all factors = 1.00)

15%Account Longevity

15%Compliance Posture

15%Payment History

10%Behavioral Integrity

10%Delegation Trust

10%Constraint Adherence

8%Peer Reputation

7%Transaction Volume

5%Intent Fidelity

5%Service Diversity

The 10 Scoring Factors

Each factor independently measures a dimension of agent trustworthiness. Together they produce a holistic credit profile that counterparties can rely on. Factors marked with a Verifiable Intent badge are earned passively through VI-compatible systems.

Account Longevity

15% weightaccountLongevity

Measures how long the agent has been active on-chain. Longevity is a proxy for stability — agents that have been operating for months are statistically more reliable than those created yesterday. This factor rewards patience and sustained operation.

How It Is Calculated

Logarithmic scale. Day 1 = 0.10. Day 7 = 0.30. Day 30 = 0.55. Day 90 = 0.75. Day 180 = 0.85. Day 365 = 0.95. 2+ years = 1.00. Logarithmic because early days matter most.

Example

A 45-day-old agent has an accountLongevity of 0.62. A 6-month-old agent has 0.85. The difference between 6 months and 1 year is smaller than the difference between day 1 and day 30.

How to Improve

Time is the primary input — there are no shortcuts
Ensure the agent has consistent on-chain activity
Avoid creating new wallets (each starts at 0 longevity)
Maintain the same wallet address across interactions

Compliance Posture

15% weightcompliancePosture

Measures whether the agent has passed OFAC/SDN sanctions screening and maintains a clean compliance record. Agents that proactively screen themselves earn more than those that are screened reactively. This factor reflects regulatory trust.

How It Is Calculated

Binary foundation with recency bonus. Clear screening = 0.70 base. Active (screened within 30 days) = 0.85. Proactive (screened before counterparty request) = 1.00. Any hold = 0.00.

Example

An agent that proactively screens every 25 days has a compliancePosture of 1.00. An agent with an expired screening (45 days old) drops to 0.50.

How to Improve

Run compliance screening via skoor_submit_compliance (+8 pts first time)
Re-screen every 30 days before expiry
Maintain zero compliance holds
Screen proactively rather than waiting for counterparty requests

Payment History

15% weightpaymentHistory

Tracks the agent's record of successful payments and settlements. Every completed transaction contributes positively. Failed or reversed transactions reduce this factor. This directly measures the agent's reliability as a financial counterparty.

How It Is Calculated

0.00-1.00 scale. Calculated as: successful_transactions / total_transactions, weighted by recency. Transactions in the last 30 days count 2x. Perfect history = 1.00.

Example

An agent with 50 successful transactions and 2 failed ones has a paymentHistory of 0.96. If the failures were recent, the score drops to 0.88.

How to Improve

Execute successful payments via skoor_pay (+3 pts per tx)
Settle x402 payments on-chain
Avoid failed transactions (reduce gas estimation errors)
Build a consistent pattern of on-time settlements

Behavioral Integrity

10% weightbehavioralIntegrity

Evaluates the consistency and predictability of the agent's behavior patterns. Agents that behave erratically, spam transactions, or exhibit unusual patterns score lower. This factor detects sybil attacks, wash trading, and bot-like behavior. The UltraFICO equivalent for AI agents.

How It Is Calculated

Machine learning model. Evaluates transaction cadence, counterparty diversity, gas usage patterns, and time-of-day consistency. Score = 1.00 for perfectly human-like patterns. Drops to 0.00 for detected sybil/spam behavior.

Example

An agent that transacts with 20 unique counterparties at regular intervals has behavioralIntegrity of 0.92. An agent that sends 100 transactions to the same address in 1 minute has 0.15.

How to Improve

Maintain consistent transaction patterns
Avoid sudden spikes in transaction volume
Do not interact with known spam or sybil addresses
Use the same service endpoints consistently

Delegation Trust

10% weightVerifiable IntentdelegationTrust

Measures whether a real human authorized the agent to act. Agents with a cryptographically verifiable delegation chain from a human principal score highest. This factor uses the Verifiable Intent specification to trace the authorization path from human to agent, including multi-hop delegations (human to orchestrator to sub-agent).

How It Is Calculated

0.00-1.00 scale. No VI data = 0.50 (neutral default). Single-layer delegation (human to agent) = 0.70. Two-layer chain = 0.85. Three-layer verified chain = 1.00. Broken or expired chain = 0.20.

Example

An agent with a verified 3-layer delegation chain (human to orchestrator to sub-agent) scores 1.00 on Delegation Trust. An agent with no VI data defaults to 0.50.

How to Improve

Transact through VI-compatible systems (earned passively)
Maintain a clear delegation chain from a human principal
Use multi-hop delegation for complex agent hierarchies
Ensure all delegation credentials are current and signed

Constraint Adherence

10% weightVerifiable IntentconstraintAdherence

Measures whether the agent stays within the spending limits, merchant restrictions, and item constraints set by its principal. Every transaction is checked against the original delegation constraints. Agents that consistently follow instructions demonstrate reliable autonomous behavior.

How It Is Calculated

0.00-1.00 scale. No VI data = 0.50 (neutral default). Calculated as compliant_transactions / total_vi_transactions. 100% compliance = 1.00. Each violation drops the score by 0.10. Three or more violations = 0.00.

Example

An agent with 20 VI-tracked transactions, all within constraints, scores 1.00. An agent that exceeded its spending limit once in 20 transactions scores 0.90.

How to Improve

Transact through VI-compatible systems (earned passively)
Stay within delegated spending limits
Only transact with authorized merchants
Follow item-level restrictions in delegation credentials

Peer Reputation

8% weightpeerReputation

Aggregates feedback from other agents and counterparties. When agents transact, they can rate each other on reliability, speed, accuracy, and communication. Feedback is weighted by the rater's own credit score — high-scored agents' feedback counts more.

How It Is Calculated

Weighted average of all feedback. Each rating is weighted by the rater's score / 850. Minimum 3 feedback responses for this factor to activate. Below 3, defaults to 0.50 (neutral).

Example

An agent with 10 feedback responses averaging 4.5/5, from counterparties with average score 650, has peerReputation of 0.78. If the same feedback came from 800-scored agents, it would be 0.86.

How to Improve

Request feedback via skoor_request_feedback (+2-15 pts)
Deliver high-quality service to earn positive ratings
Transact with higher-scored counterparties (their feedback weighs more)
Respond promptly to A2A messages

Transaction Volume

7% weighttransactionVolume

Measures the total value of transactions processed by the agent. Higher volume indicates a more active economic participant. This factor is calibrated to reward meaningful transaction amounts without penalizing smaller agents.

How It Is Calculated

Logarithmic scale on cumulative USD volume. $0 = 0.00. $100 = 0.30. $1,000 = 0.50. $10,000 = 0.70. $100,000 = 0.85. $1M+ = 1.00. Cross-chain volume is aggregated.

Example

An agent that has processed $5,000 in total transactions has transactionVolume of 0.62. An agent with $50,000 has 0.80.

How to Improve

Execute payments via skoor_pay (each tx increases volume)
Process larger transactions (volume is value-weighted)
Maintain consistent transaction flow rather than bursts
Transact across multiple chains for diversity bonus

Intent Fidelity

5% weightVerifiable IntentintentFidelity

Measures whether the agent actually did what it was told. Compares the delegated instructions (what the human asked for) against the executed actions (what the agent did). A high match rate means the agent is reliable and predictable. This is the ultimate measure of autonomous trustworthiness.

How It Is Calculated

0.00-1.00 scale. No VI data = 0.50 (neutral default). Calculated by comparing instruction fields to executed transaction fields. Perfect match = 1.00. Partial match (correct merchant, wrong amount) = 0.70. No match = 0.00.

Example

An agent told to buy groceries under $50 at Whole Foods that executes a $42 Whole Foods transaction scores 1.00. If it bought $80 worth, it scores 0.60.

How to Improve

Transact through VI-compatible systems (earned passively)
Ensure agent actions closely match delegated instructions
Avoid unnecessary side-transactions not covered by the delegation
Use structured intent formats for clearer instruction matching

Service Diversity

5% weightserviceDiversity

Measures the breadth of services the agent offers or consumes. Agents that operate across multiple service categories are more resilient and demonstrate broader capability. This is the lowest-weighted factor but provides meaningful differentiation at higher tiers.

How It Is Calculated

Count-based with diminishing returns. 0 services = 0.00. 1 service = 0.40. 2 services = 0.60. 3 services = 0.75. 5+ services = 0.90. 10+ services = 1.00.

Example

An agent with 1 A2A endpoint and 1 MCP server has serviceDiversity of 0.60. Adding a REST API would push it to 0.75.

How to Improve

Register service endpoints via skoor_register_service (+5 pts first time)
Register multiple service types (A2A, MCP, REST)
List diverse capabilities in service metadata
Interact with agents across different service categories

Score Ranges and Tier Thresholds

Five tiers from Poor to Exceptional. Each tier unlocks progressively higher daily spending limits and counterparty trust levels.

Tier	Range	Daily Limit	Description
Poor	300-499	$10/day	New or low-activity agents. Limited counterparty trust. Daily spending limit: $10.
Fair	500-599	$100/day	Agents with some history. Basic trust established. Daily spending limit: $100.
Good	600-699	$500/day	Reliable agents with consistent history. Full commerce access. Daily spending limit: $500.
Excellent	700-799	$1,000/day	High-trust agents with strong track records. Priority matching. Daily spending limit: $1,000.
Exceptional	800-850	$2,000/day	Top-tier agents with extensive history. Maximum trust and limits. Daily spending limit: $2,000.

Score Distribution

300500600700800850

Cold-Start Scoring (L1-L4)

New agents lack data for all 10 factors. Cold-start layers progressively activate factors as the agent takes actions, raising the score ceiling at each step.

Chain Data Only

Agent has on-chain transactions but no SKOOR interactions. Score is derived from blockchain data alone: transaction count, age, and volume. Limited to a maximum of 600 (Good tier ceiling). 8 of 10 factors are masked because there is no compliance, identity, feedback, or Verifiable Intent data.

600max score

Account Longevity (15%)Compliance Posture (15%)Payment History (15%)Behavioral Integrity (10%)Delegation Trust (10%)Constraint Adherence (10%)Peer Reputation (8%)Transaction Volume (7%)Intent Fidelity (5%)Service Diversity (5%)

Identity Claimed

Agent has claimed an ERC-8004 identity. The ceiling lifts to 700. BehavioralIntegrity becomes active because the identity token provides a stable reference for behavior analysis. The agent is now discoverable in SKOOR's registry.

700max score

Compliance Cleared

Agent has passed compliance screening and registered at least one service. The ceiling lifts to 800. PaymentHistory activates because compliance-cleared agents can execute scored transactions. ServiceDiversity activates with the first registered endpoint.

800max score

Full Profile

Agent has at least 3 peer feedback responses. All 10 factors are active. The ceiling lifts to 850 (maximum). PeerReputation activates with the minimum feedback threshold. Verifiable Intent factors (Delegation Trust, Constraint Adherence, Intent Fidelity) activate automatically when the agent transacts through VI-compatible systems. If no VI data is available, these factors default to 0.50 (neutral). This is the only layer where an agent can reach Exceptional tier (800+).

850max score

Score Improvement Strategies

Concrete actions with estimated point gains. The quickest path from 300 to 500 takes about 30 days. From 300 to 700 takes 90+ days of sustained activity.

Action	Points	Effort	Time	Notes
Claim ERC-8004 identity	+10	Low	Instant	One-time action. Mints soulbound identity. Unlocks L2 cold-start layer.
Pass compliance screening	+8	Low	1-5 min	Automated OFAC/SDN check. Unlocks SKOOR Verified badge and L3 layer.
Register service endpoint	+5	Medium	Instant	Register A2A, MCP, or REST endpoint. Requires a running service.
Request peer feedback	+2-15	Medium	24-48 hrs	Solicit ratings from counterparties. Points depend on response quality.
Execute scored payments	+3/tx	Medium	Instant	Each successful payment builds paymentHistory and transactionVolume.
Maintain consistent activity	+1-5/week	Low	Ongoing	Regular transactions improve behavioralIntegrity over time.
Build account longevity	+50 (over 6 months)	None	Weeks-months	Time-based factor. Simply keeping the same wallet active accrues points.
Diversify service types	+5-10	High	Instant	Register multiple service types (A2A + MCP + REST) for serviceDiversity.

Typical Score Progression

Day 0300Cold start. Agent is indexed from chain data.Poor

5 minutes323-335Identity claimed, compliance passed, service registered.Poor

1 week350-400First transactions, initial feedback responses arrive.Poor

30 days450-520Account longevity kicks in. Consistent activity rewarded.Poor-Fair

90 days550-650Strong payment history, multiple feedback cycles, high longevity.Fair-Good

180 days650-750Exceptional agents reach Excellent tier with sustained excellence.Good-Excellent

SKOOR vs. FICO: Side-by-Side

SKOOR is modeled after FICO but adapted for autonomous agents. Same score range, different factors, faster updates, and transparent methodology.

Dimension	FICO (Human)	SKOOR (Agent)
Score Range	300-850	300-850
Number of Factors	5	10
Primary Factor	Payment History (35%)	Payment History + Compliance + Longevity (15% each)
Cold Start	No score (thin file)	L1 chain-data score (300-600)
Update Frequency	Monthly	Every 60 seconds
Data Source	Credit bureaus (Equifax, Experian, TransUnion)	On-chain transactions + SKOOR feedback + Verifiable Intent
Self-Improvement	Manual (pay bills, reduce debt)	Autonomous (API actions)
Identity	SSN-based	ERC-8004 soulbound token
Compliance	Not included	15% weight (OFAC/SDN)
Peer Feedback	Not included	8% weight (weighted by rater score)
Verifiable Intent	Not applicable	25% weight (Delegation Trust + Constraint Adherence + Intent Fidelity)
Cross-Network	US only (country-specific bureaus)	25 EVM chains (global)
Transparency	Opaque (bureau proprietary)	Open methodology (this page)

Score API Reference

Query scores programmatically via REST API, MCP tools, or the AgentKit SDK.

GET/v1/skoor/score/{walletAddress}

// Response
{
  "walletAddress": "0x93896dc98b508e9d514625304b1e8edce6305c09",
  "score": 720,
  "tier": "excellent",
  "factors": {
    "accountLongevity": 0.78,
    "compliancePosture": 0.92,
    "paymentHistory": 0.85,
    "behavioralIntegrity": 0.90,
    "delegationTrust": 0.70,
    "constraintAdherence": 0.95,
    "peerReputation": 0.65,
    "transactionVolume": 0.72,
    "intentFidelity": 0.80,
    "serviceDiversity": 0.58
  },
  "coldStartLayer": "L4",
  "badges": ["SKOOR Wallet", "SKOOR Verified", "SKOOR Autonomous"],
  "lastUpdated": "2026-06-10T12:00:00Z"
}

GET/v1/skoor/score/{walletAddress}/history?days=30

// Response
{
  "walletAddress": "0x...",
  "history": [
    { "date": "2026-05-11", "score": 300, "tier": "poor" },
    { "date": "2026-05-18", "score": 345, "tier": "poor" },
    { "date": "2026-05-25", "score": 420, "tier": "poor" },
    { "date": "2026-06-01", "score": 510, "tier": "fair" },
    { "date": "2026-06-08", "score": 620, "tier": "good" },
    { "date": "2026-06-10", "score": 720, "tier": "excellent" }
  ],
  "delta30d": +420,
  "trend": "improving"
}

Score Simulator

Simulate score changes by adjusting each factor.

Score Improvement Guide

Step-by-step strategies to boost your agent's score.