openai-agents-trust 0.1.0

Trust & governance layer for OpenAI Agents SDK — policy enforcement, trust-gated handoffs, and Merkle audit trails

openai-agents-trust

Trust & governance layer for the OpenAI Agents SDK. Adds policy enforcement, trust-gated handoffs, and tamper-evident audit trails using native SDK guardrails and hooks.

Built by AgentMesh — the open-source trust layer for multi-agent systems. Similar integrations merged into Dify (65K ⭐), LlamaIndex (47K ⭐), and Microsoft Agent-Lightning (15K ⭐).

Install

pip install openai-agents-trust

Quick Start

from agents import Agent, Runner
from openai_agents_trust import (
    trust_input_guardrail,
    policy_input_guardrail,
    GovernanceHooks,
    TrustGuardrailConfig,
    PolicyGuardrailConfig,
    TrustScorer,
    GovernancePolicy,
    AuditLog,
)

# Shared governance state
scorer = TrustScorer()
audit = AuditLog()
policy = GovernancePolicy(
    name="production",
    max_tool_calls=20,
    blocked_patterns=[r"DROP TABLE", r"rm -rf", r"eval\("],
    min_trust_score=0.7,
)

# Create guardrails
trust_config = TrustGuardrailConfig(scorer=scorer, min_score=0.7, audit_log=audit)
policy_config = PolicyGuardrailConfig(policy=policy, audit_log=audit)

# Attach to agents
agent = Agent(
    name="researcher",
    instructions="You are a research assistant.",
    input_guardrails=[
        trust_input_guardrail(trust_config),
        policy_input_guardrail(policy_config),
    ],
)

# Run with governance hooks
result = await Runner.run(
    agent,
    input="Analyze this data",
    run_config=RunConfig(hooks=GovernanceHooks(policy=policy, scorer=scorer, audit_log=audit)),
)

# Verify audit integrity
print(f"Audit entries: {len(audit)}")
print(f"Chain valid: {audit.verify_chain()}")

Trust-Gated Handoffs

from openai_agents_trust import trust_gated_handoff

billing_agent = Agent(name="billing", instructions="Handle billing.")
support_agent = Agent(name="support", instructions="Handle support.")

triage = Agent(
    name="triage",
    handoffs=[
        trust_gated_handoff(billing_agent, scorer=scorer, min_score=0.8),
        trust_gated_handoff(support_agent, scorer=scorer, min_score=0.6),
    ],
)

Features

Feature	SDK Hook	Description
Trust Guardrail	InputGuardrail	Blocks agents below trust threshold
Policy Guardrail	InputGuardrail	Enforces blocked patterns, tool limits
Content Guardrail	OutputGuardrail	Validates output against policies
Governance Hooks	RunHooksBase	Tracks tools, audits handoffs, scores trust
Trust-Gated Handoff	is_enabled	Disables handoffs to untrusted agents
Merkle Audit	—	Tamper-evident chain of all decisions

License

MIT