Trust Scores Explained

A complete guide to the trust and confidence scores shown on every verified Clawdbase listing.

Every verified listing on Clawdbase shows two numbers: a trust score and a confidence score. They answer different questions - you need both to make a good judgment.

Trust Score

The trust score measures the creator's verified identity and reputation - not the quality of the agent itself.

A high trust score means the creator has a well-established, verifiable presence across platforms. It answers: "Is there a real, accountable person behind this agent, and do they have a track record?"

The score is calculated by inVerus.ai from signals across platforms - GitHub history, LinkedIn professional data, account age, and cross-platform consistency. It is computed entirely from third-party verifiable data; creators cannot set their own score.

Score tiers

High75 – 100

Strong, established creator identity. Safe for production use.

Medium50 – 74

Partial verification. Good for most use cases - worth a review.

Low0 – 49

Limited verifiable information. Proceed with additional scrutiny.

A low trust score doesn't mean the agent is bad - it means the creator has a limited or unverifiable track record. New legitimate creators start low and build up over time.

Confidence Score

The confidence score measures how certain the assessment is.

Two creators might calculate the same trust score, but if one has 200 data points across five verified platforms and the other has 8 from a recently created account, they are not equally trustworthy signals. Confidence captures this difference.

High Confidence (75–100)92

Strong evidence - the trust score is reliable

High
Medium Confidence (50–74)62

Moderate evidence - score is likely accurate

Medium
Low Confidence (0–49)28

Sparse evidence - treat the score as preliminary

Low

Reading them together

The real signal comes from the combination of both scores.

High Confidence
Low Confidence
High Trust
Low Trust
Best Signal

Strong identity, solid evidence. Most reliable combination for production use.

Promising but Unconfirmed

Early signals are positive but the data is sparse. Could be a new legitimate creator.

Significant Caution

We have solid evidence that this creator does not meet the verified identity bar.

?Insufficient Data

Very little verifiable information. Score is preliminary - treat with caution.

In practice: prefer High trust + High confidence for critical production use. High trust + Low confidence may be a new but legitimate creator worth watching. Low trust + High confidence is a clear warning signal - the evidence is solid and unfavorable.

Trust Equity

Trust Equity is an accumulated reputation layer on top of the base trust score. It grows as:

  • Your registered agents attract GitHub stars and forks
  • Your agents are referenced as parents in other creators' lineage
  • Your agents maintain consistent quality over time

Trust Equity flows downstream through lineage. If you build an agent that others build on top of, your equity flows into their trust signal. Being a foundational builder has compounding value in the ecosystem.

Example lineage tree

OpenInterpreter
AGT-a1b2c3d488
open-interpreter-plugin
AGT-e5f6a7b872
My Custom Agent
AGT-c9d0e1f265
88
Root Trust
+7
Equity flow
65
Your Score

What affects your score

The inVerus protocol weights multiple signals when calculating a creator's score. Account age and activity on established platforms carry the most weight.

GitHub account age & activityVery High
LinkedIn professional historyHigh
Verified account countHigh
Cross-platform identity consistencyMedium
MySanctum profile completenessMedium
Registered agent reputationMedium
Additional platform accountsSupplementary

To see exactly which factors are contributing to your own score and what's missing, visit your MySanctum Score Breakdown.

Developer: querying scores via API

You can fetch trust scores programmatically for any verified creator or agent.

// Fetch trust profile for a creator
const res = await fetch('https://clawdbase.ai/api/identity/openai')
const { trustScore, confidence, tier } = await res.json()
// { trustScore: 88, confidence: 91, tier: 'high' }

// Fetch scores from an agent record
const agent = await fetch('https://clawdbase.ai/api/agents/AGT-a1b2c3d4').then(r => r.json())
console.log(agent.creator.trustScore, agent.creator.confidence)
import httpx

profile = httpx.get('https://clawdbase.ai/api/identity/openai').json()
print(profile['trustScore'], profile['tier'])  # 88, 'high'
curl "https://clawdbase.ai/api/identity/openai"
# { "trustScore": 88, "confidence": 91, "tier": "high", ... }

See the full REST API reference for all available endpoints and response shapes.