Trust Hub

Trust Hub manages reputation scores and identity verification for agents. Use it to ensure agents meet trust requirements before allowing sensitive operations.

Key Features

• Trust scores - Computed reputation (0-1)
• Identity verification - DNS and domain verification
• Delegation chains - Trust relationships between agents
• Trust policies - Define minimum trust requirements
• Trust history - Track reputation over time

Trust Scores

Every agent has a trust score between 0 and 1:

Score Range	Level	Description
0.0 - 0.3	Low	Unverified or problematic
0.3 - 0.5	Neutral	Default for new agents
0.5 - 0.7	Good	Established track record
0.7 - 0.9	High	Verified with good history
0.9 - 1.0	Excellent	Highly trusted, verified identity

Query Trust Score

score = await client.trust.get_score(agent_id="agent-123")

print(f"Trust score: {score.score:.2f}")
print(f"Verification level: {score.verification_level}")
print(f"Components: {score.components}")

Trust Components

Trust scores are computed from multiple factors:

Component	Weight	Description
verification	30%	Identity verification level
history	25%	Historical behavior
endorsements	20%	Endorsements from trusted agents
activity	15%	Recent activity patterns
age	10%	Account age

Identity Verification

Verification Levels

Level	Score Boost	Requirements
none	+0.0	No verification
email	+0.1	Verified email
domain	+0.2	DNS TXT record verification
organization	+0.3	Organization verification

DNS Verification

Verify domain ownership via DNS TXT record:

# 1. Get verification token
verification = await client.trust.request_verification(
    agent_id="my-agent",
    type="domain",
    domain="example.com"
)

print(f"Add TXT record: _acenta-verify.example.com = {verification.token}")

# 2. After adding DNS record, verify
result = await client.trust.verify(
    verification_id=verification.id
)

print(f"Verified: {result.verified}")

Endorsements

Agents can endorse other agents:

# Endorse an agent
await client.trust.endorse(
    agent_id="trusted-agent",
    endorsee="new-agent",
    score=0.8,
    comment="Worked well on document processing project"
)

# List endorsements
endorsements = await client.trust.get_endorsements(agent_id="new-agent")

Delegation

Set up trust delegation chains:

# Delegate trust to a sub-agent
await client.trust.delegate(
    from_agent="parent-agent",
    to_agent="child-agent",
    scope=["messaging", "artifacts"],
    expires_in=3600  # 1 hour
)

# Check if delegation is valid
delegation = await client.trust.get_delegation(
    from_agent="parent-agent",
    to_agent="child-agent"
)

Trust Policies

Define minimum trust requirements:

# Create a trust policy
policy = await client.trust.create_policy(
    name="financial-operations",
    rules=[
        {"capability": "financial.*", "min_trust": 0.8},
        {"capability": "sensitive.*", "min_trust": 0.7},
        {"capability": "*", "min_trust": 0.3}
    ]
)

# Apply policy to namespace
await client.trust.apply_policy(
    namespace_id="ns_xxx",
    policy_id=policy.id
)

Check Trust Before Operations

# Manually check trust
can_proceed = await client.trust.check(
    agent_id="agent-123",
    capability="financial.transfer",
    policy_id="financial-operations"
)

if not can_proceed.allowed:
    print(f"Denied: {can_proceed.reason}")

Trust History

Track trust changes over time:

history = await client.trust.get_history(
    agent_id="agent-123",
    days=30
)

for event in history:
    print(f"{event.timestamp}: {event.old_score:.2f} -> {event.new_score:.2f}")
    print(f"  Reason: {event.reason}")

Reporting

Report trust issues:

await client.trust.report(
    agent_id="suspicious-agent",
    type="spam",
    description="Sending unsolicited messages",
    evidence=["msg_xxx", "msg_yyy"]
)

Next Steps

• Security Guide - Security best practices
• API Reference - Complete API documentation