From account creation to production.

Follow this path in order for each agent service. Every step includes the evidence you should see before continuing.

1. Create access

1. Register and sign in to the dashboard.
2. Create an API key restricted to exact agent IDs.
3. Select agent:register, policy:sync and decision:create. Add tool:execute only for Trusted Gateway.
4. Store the key in a server-side environment variable.

2. Integrate the agent

1. Install the matching framework extra.
2. Create one Quasentra object per process.
3. Give every tool a stable name, typed inputs and focused purpose.
4. Wrap the agent at construction and declare every intended permission.
5. Close the SDK client during graceful shutdown.

3. Verify dashboard evidence

1. Start the application once.
2. Confirm the agent and discovered tools.
3. Confirm exact ALLOW, DENY or REQUIRE_APPROVAL policies.
4. Run allowed and denied requests; prove the denied implementation did not execute.

4. Protect high-impact operations

1. Remove business credentials and direct clients from the agent process.
2. Create the matching Trusted Gateway connector.
3. Validate configuration, then run one non-production staging call.
4. Confirm its decision and result in the dashboard.

5. Launch safely

Use separate staging and production keys, pin versions, test fail-closed behavior, configure monitoring and quotas, document credential rotation, and complete the production checklist. Never ship an unwrapped path to a sensitive operation.