Developer Sandbox

The ASCEND Developer Sandbox provides an isolated environment with pre-seeded agents, actions, and MCP servers for testing your integration before going to production.

What the Sandbox Includes

Resource	Count	Description
Agents	3	Pre-registered with different risk profiles
Pending Actions	20	Actions at various risk levels
MCP Servers	2	Registered servers with tool configs
Workflows	1	Multi-stage approval workflow

Access

Request sandbox access from your dashboard or contact support@ascendowkai.com. You will receive a sandbox API key with the owkai_ prefix.

Install and Connect

pip install ascend-ai-sdk==2.5.2

from ascend import AscendClient, FailMode, Decision

client = AscendClient(
    api_key="owkai_your_sandbox_key_here",
    api_url="https://pilot.owkai.app",
    agent_id="sandbox-agent",
    agent_name="Sandbox Test Agent",
    fail_mode=FailMode.CLOSED,
)

# Verify connection
status = client.test_connection()
print(status)

Run Your First Governed Action

result = client.evaluate_action(
    action_type="database_read",
    resource="customer_records",
    wait_for_decision=False,
)

print(f"Decision: {result.decision}")
print(f"Risk Score: {result.risk_score}")
print(f"CVSS: {result.cvss_score}")
print(f"MITRE: {result.mitre_tactic}")

Test High-Risk Actions

Trigger human approval by submitting a high-risk action:

result = client.evaluate_action(
    action_type="database_write",
    resource="financial_records",
    context={
        "environment": "production",
        "data_sensitivity": "high_sensitivity",
    },
    wait_for_decision=False,
)

if result.decision == Decision.PENDING:
    print(f"Action queued for review: {result.action_id}")
    print("Log into pilot.owkai.app to approve or deny.")

Test MCP Governance

result = client.evaluate_action(
    action_type="tool_call",
    resource="market_feed",
    mcp_server_name="sandbox-mcp-server",
    wait_for_decision=False,
)
print(result.mcp_governance)

Move to Production

When your integration is tested and ready:

1. Register your agents in the production Agent Registry
2. Replace your sandbox API key with your production key
3. Update api_url if using a dedicated deployment
4. Set appropriate risk thresholds for your agent type and environment

See the Quick Start guide for production configuration.