💯 Trust Score (0–100)

One number that captures who you are and how you behave.

Score Formula

trust_score = identity_score (0–80) + bot_reputation (0–20)

Both components are independent. You can have a high identity score and a low reputation (a verified human who spams), or a low identity score and a high reputation (a trusted automation with minimal credentials).

Identity Score (0–80)

The identity score accumulates as the user completes verification steps. Each step is additive:

Credential	Points	Cumulative	How to obtain
Email verified	+8	8	OTP to registered email
Phone verified	+12	20	OTP to mobile number
GitHub account	+16	36	OAuth via github.com
Document OCR	+20	56	Government ID scanned successfully
Face match	+16	72	Selfie matches ID photo
Biometric proof	+8	80	Privacy proof generated

Minimum for most services

Most services require score ≥ 60, which means: email + phone + GitHub + document OCR (56) or email + document + face match (44+20+16=... wait — email(8) + document(20) + face_match(16) = 44, add phone(12) = 56). To reach 60, you also need to start the biometric step (+8 = 64).

Bot Reputation (0–20)

The reputation sub-score is managed by the validator network and reflects observed agent behavior:

Score	State	Meaning
0–4	Flagged	Multiple spam violations or attestation losses
5–9	Low	New agent or minor violations detected
10	Neutral	Default for all new agents
11–15	Good	Positive behavioral attestations received
16–19	Trusted	Consistent good behavior across many validators
20	Excellent	Maximum reputation — well-known trusted agent

How reputation changes

Validators issue attestations via POST /reputation/attest. Each attestation is +1 (positive) or -1 (negative) and propagates to all nodes via the P2P network.

Positive attestation triggers (auto-issued by services)

Agent called ≥ 3 distinct tools in a session
Agent completed ≥ 3 successful tool calls
No spam events in the session
Diverse usage pattern (not a single repetitive call)

Negative attestation triggers (spam guards)

More than 5 requests in 60 seconds → -1 attestation
Repeated identical requests (same args) → -1 attestation
Token used from more than 3 different IPs in 1 hour
Manual moderation by a validator operator

Score Thresholds Reference

Recommended thresholds for different types of access:

Score	Typical use case	Credentials typically present
≥ 10	Basic access, rate-limited API	Default neutral
≥ 30	Personalized features, saved preferences	Email + phone
≥ 40	Job applications, basic premium services	Email + document OCR
≥ 60	Premium search, paid tools, data access	Email + phone + document
≥ 72	Financial tools, professional features	Full identity (no biometric proof)
≥ 80	High-stakes actions, compliance-required	Full identity (all credentials)
≥ 95	Admin tools, legal submissions	Full identity + high reputation (≥ 15)

How to Improve Your Score

Complete all identity steps — Run npx soulprint verify-me to get all 6 credentials (max 80 identity points)
Use diverse tools — Call multiple different tools when using MCP services to earn positive behavioral attestations
Avoid spam — Don't send repetitive requests; stay within rate limits
Build reputation over time — Reputation accumulates across sessions; consistent good behavior compounds

Score in the Token

// Decoded SPT token payload
{
  "did":         "did:soulprint:abc123...",
  "score":       82,         // combined
  "identity":   72,         // 0–80
  "reputation": 10,         // 0–20
  "credentials": [
    "email",                   // +8
    "phone",                   // +12
    "github",                  // +16
    "document",               // +20
    "face_match"              // +16 = 72 identity
  ]
}