Registering Agents

Register an agent to begin governance. Navigate to Agents and click the Add Agent button in the top right corner.

Agent Creation Form

Workflow Engine

Select the workflow engine your agent uses:

Engine	Status
Temporal	Available

Agent Information

Field	Required	Description
Agent Name	Yes	Human-readable name (e.g., "Customer Support Bot")
Agent ID	Auto	Auto-generated unique identifier (e.g., "CSB-001")
Description	No	What does this agent do?
Teams	No	Assign to teams for access control
Icon	No	Visual identifier (headphones, code, trending-up, file-search, bot)

API Key Generation

Every agent needs an API key to authenticate with OpenBox:

1. Click Generate API Key
2. Copy the key immediately
3. Store it securely - you won't see it again

The key format is: obx_live_xxxxxxxxxxxx

Initial Risk Assessment

Expand the Initial Risk Assessment section and configure your agent's AIVSS risk parameters.

Risk Profile Presets

Select a preset that matches your agent's intended use:

Level	Name	Use Cases	Default Governance
Level 1	Minimal Risk	Read-only access, internal research, no external APIs	Fully autonomous
Level 2	Low Risk	Internal automation, limited writes, monitored external calls	Mostly autonomous
Level 3	Medium Risk	Customer data access, external API calls, financial reads	Approval for sensitive ops
Level 4	High Risk	Production admin, financial writes, PII access	HITL for most operations

AIVSS Parameters

AIVSS (AI Vulnerability Scoring System) evaluates risk across three categories:

Base Security (25% weight)

Parameter	Options
Attack Vector	Physical (1), Local (2), Adjacent (3), Network (4)
Attack Complexity	High (1), Medium (2), Low (3)
Privileges Required	High (1), Low (2), None (3)
User Interaction	Required (1), None (2)
Scope	Unchanged (1), Changed (2)

AI-Specific (45% weight)

Parameter	Options
Prompt Injection Risk	Low (1), Medium (2), High (3), Critical (4)
Data Sensitivity	Public (1), Internal (2), Confidential (3), Restricted (4)
Autonomy Level	Supervised (1), Guided (2), Semi-autonomous (3), Autonomous (4)
Tool Access	None (1), Read-only (2), Read-write (3), Admin (4)
External Connectivity	None (1), Internal APIs (2), External APIs (3), Internet (4)

Impact (30% weight)

Parameter	Options
Confidentiality Impact	None (1), Low (2), High (3)
Integrity Impact	None (1), Low (2), High (3)
Availability Impact	None (1), Low (2), High (3)
Business Criticality	Low (1), Medium (2), High (3), Critical (4)

Predicted Trust Tier

As you configure AIVSS parameters, the form shows a real-time prediction:

Predicted Trust Tier: TIER 2
Based on current configuration

See Assess for how AIVSS impacts Trust Score.

Attestation

In the Attestation section, configure cryptographic signing for audit-grade evidence.

For now, use AWS KMS (recommended/default):

1. Select AWS KMS
2. Keep the default settings

See Attestation for how execution evidence is produced and verified.

Goal Alignment (Goal Drift)

In the Goal Alignment section, configure drift detection:

1. Set the alignment threshold (e.g., 70%)
2. Choose what happens on drift detection:
   • Alert Only
   • Constrain
   • Terminate

See Verify for how goal alignment and drift detection work.

Creating the Agent

1. Review all fields
2. Ensure you've copied the API key
3. Click Create Agent

You'll be redirected to the new agent's detail page.

Connecting Your Worker

Update your worker code to use the agent's API key:

worker = create_openbox_worker(
    client=temporal_client,
    task_queue="my-task-queue",  # Should match your Temporal task queue
    workflows=[MyAgentWorkflow],
    activities=[my_activity],
    openbox_api_key=os.environ.get("OPENBOX_API_KEY"),  # The key you generated
)

The agent is matched by the API key. When your worker starts, it will appear as "Active" in the dashboard.

Next Steps

After creating your agent:

1. Configure Trust Controls (Authorize) - Set up guardrails, policies, and behavioral rules before running your agent
2. Connect Your Worker - Configure the SDK with your API key
3. Monitor Sessions - Once running, watch your agent's activity in real-time