AG9Architect AI agent scope
Managed agent runtime review
A configuration review of the managed agent runtime you build on: its policy layer, gateway, identity, network isolation and guardrails.
Price on requestAsk about AG9How pricing works
What it is for
Managed agent runtimes give you security controls out of the box, but only if they are switched on and set well. AG9 reviews the settings against your threat model.
What we look at
For the runtime you use, we review policy, gateway, identity, network isolation, guardrails and logging. Examples of what the providers offer today:
- Amazon Bedrock AgentCore: Policy, generally available since 3 March 2026, lets teams write rules in natural language that convert to the Cedar policy language and are enforced at an AgentCore Gateway, which allows or denies each agent-tool request.[1]
- Gemini Enterprise Agent Platform: Model Armor for Agent Gateway, generally available since 24 June 2026, applies content guardrails to prompts and responses passing through the gateway.[2]
- Microsoft Foundry Agent Service: each agent can have its own Microsoft Entra identity, and agents can run inside your Azure virtual network.[3]
The full list of runtimes we work with is on the platforms page. Naming a provider here implies no partnership or endorsement.
What you receive
A configuration review, findings mapped under the TriOngle Method, and a hardened baseline you can apply and keep under change control.
Framework mapping
The frameworks this scope is aligned with, the version we map to, and what we map. Aligned with means we use the framework; it does not mean any body has certified TriOngle.
| Framework | Version | What we map |
|---|---|---|
| Managed agent runtime documentation | As published, checked 4 Nov 2026[1][2][3] | The provider’s own controls and their settings |
| OWASP Top 10 for Agentic Applications | 2026 edition, 9 Dec 2025[4] | ASI02 Tool Misuse and Exploitation; ASI03 Identity and Privilege Abuse |