Agent Governance Cerone SDK (Beta)

A lean governance system for AI agents. Cryptographic identity and run-time validation for AI agents in production.

AI agents are being deployed into production systems that handle real decisions, real data, and real consequences, yet most teams have no runtime control over what those agents actually do. Agent Governance fixes that.

Agent Governance enforces cryptographic identity, semantic intent validation, and continuous trust scoring on every agent action; ensuring agents only do what they were authorized to do, and nothing else. Deploying AI agents in production without this layer of protection is an unacceptable operational and security risk. It is essentially a lightweight layer that does not aim to introduce any technical debt.

The SDK is currently in early-access beta for selected teams building production AI systems that require accountability, auditability, and control.

Install

pip install agent-governance-cerone-sdk

Access (Beta)

API access is currently managed through early access, trials and POCs are highly welcomed:

• Request access: https://www.homersemantics.com/ai-agent-governance-and-oauth
• Reach out to Anant at anantdhavale@gmail.com or info@homersemantics.com
• Receive approval and API credentials
• Review terms in TERMS_OF_SERVICE.md

Quick Start

from agent_governance import AgentGovernanceClient
import os

client = AgentGovernanceClient(api_key=os.getenv("AGENT_GOVERNANCE_API_KEY"))

# Create agent
agent = client.create_agent(
    purpose="Email classifier",
    capabilities=["read_email", "move_email", "send_reply"]
)

# Validate action before executing
response = client.validate(
    agent_id=agent.agent_id,
    action="move_email",
    parameters={"email_id": "123", "folder": "spam"}
)

if response.result.value == "approved":
    mailbox.move_email("123", "spam")
else:
    print(f"Action rejected: {response.violations}")

What the SDK Provides

• Authenticated client access to Agent Governance services
• Agent registration and identity-bound operations
• Sync and async validation calls
• Batch validation support
• Governance-oriented telemetry surfaces

Usage Terms

Usage limits and commercial terms are defined in TERMS_OF_SERVICE.md.

Current free trial: up to 25 validations in the first 5 days.

License

Proprietary commercial license. See LICENSE.

Contact

• Reach out to Anant at anantdhavale@gmail.com or info@homersemantics.com
• If you are using Agent Governance, we would highly appreciate any feedback.
• This is a redux of Cerone, with an easy to understand name.