Opencontroller by Lyzr
Opencontroller blog
Answers to the questions enterprises ask about governing AI agents
Agent governance basics
All in this topic- Guide
What is an AI agent control plane?
An AI agent control plane is the system of record and enforcement layer for every agent an enterprise runs: inventory, identity, policy, budgets and evidence.
- Guide
What is Opencontroller? The open control plane for enterprise AI agents
Opencontroller by Lyzr finds every AI agent in your enterprise, gives each one an owner and identity, and controls what it can call, spend and touch.
What is the difference between AI agent observability and AI agent governance?
Observability shows what an AI agent did; governance decides what it may do and acts on it. A side-by-side comparison of questions, data, owners and actions.
How can a small security team govern AI agents?
Five AI agent controls ranked by risk reduction for the effort, a weekly 60-minute review agenda, and what a two-person security team should automate first.
How do shared service centres govern AI agents that serve many business units?
How a shared service centre splits AI agent ownership with business units: who owns what, per-unit policies, chargeback hooks and a five-step intake flow.
How do you govern AI agents embedded in your own mobile and desktop apps?
A checklist for AI agents shipped inside your own mobile and desktop apps: no keys in the client, gateway proxying, user identity, telemetry and kill paths.
How do you govern AI agents built at an internal hackathon?
A sandbox tier, synthetic data, expiring credentials and one promotion gate: a checklist for governing agents built at internal hackathons.
How do approval templates speed up AI agent registration?
Risk-tier approval templates pre-fill controls, evidence and approvers for AI agent registration. Field table plus example low, medium and high templates.
How should you govern AI agents used in the month-end financial close?
Controls for AI agents in the financial close: map reconciliations, accruals and journal entries to SOX ITGC evidence, segregation of duties and thresholds.
Discovery and shadow agents
All in this topic- Guide
How to find every AI agent in your enterprise (including shadow agents)
A step-by-step method to inventory AI agents across AWS, Azure, Google Cloud, SaaS platforms and developer devices, and turn shadow agents into owned ones.
What does AI agent posture scanning actually inspect?
AI agent posture scanning reads configurations, prompts, tool definitions and permissions to flag risky setups. What it inspects, how secrets should be handled.
Agent identity and access
All in this topic- Guide
How to give AI agents an identity (and let them act on behalf of users safely)
Why shared API keys fail for AI agents, and how to give each agent its own identity with scoped, delegated on-behalf-of access using OAuth token exchange.
How do you stop users gaining a service account's access through an AI agent?
On-behalf-of token exchange, permission intersection and authorisation at the tool stop users borrowing an agent service account's access. Decision table inside.
How do you grant MCP server access by identity provider group?
Map IdP groups to policies, policies to per-user credentials, and credentials to MCP servers. Why shared application keys break group-based access.
Runtime control
All in this topic- Guide
How to stop a rogue AI agent: kill switches, quarantine and runtime controls
What a rogue AI agent is, how to detect behavioural deviation early, and the five layers of control that let you block, quarantine or kill an agent in seconds.
What controls do AI agents in procurement need?
PO thresholds, no vendor bank-detail changes, three-way match and separation of duties: a control checklist for procurement AI agents.
Tools and MCP
All in this topic- Guide
How to govern MCP servers in the enterprise: a practical guide
How to approve, register and control Model Context Protocol (MCP) servers so each AI agent can only call the tools it needs, with audit evidence for every call.
What should you do with MCP servers that have no owner?
An ownership-claim workflow for discovered MCP servers: keep them in a shadow state, set a claim deadline, block on clear criteria, and record evidence.
Should MCP access be authorised per server or per tool?
Server-level MCP access decides who reaches a server; tool-level authorisation decides which actions they can take. What each stops, and interim controls.
Gateways and cost
All in this topic- Guide
LLM gateway vs agent gateway vs AI control plane: what's the difference?
An LLM gateway governs model calls, an agent gateway governs agent and tool calls, and a control plane governs the whole agent lifecycle. Here's when you need each.
What is an agent gateway?
An agent gateway mediates agent-to-tool (MCP), agent-to-agent (A2A) and inbound agent calls. What it enforces and how it differs from an LLM gateway.
How do you autoscale an LLM gateway on Kubernetes?
Scale an LLM gateway on Kubernetes with the Horizontal Pod Autoscaler: which metric to use, how to handle streaming connections, and how to size replicas.
How do you attribute LLM usage when many users share one application key?
Three ways to see who is behind traffic on a shared LLM key: per-user virtual keys, end-user identifier fields, and signed token pass-through.
How much does storing AI agent traces cost?
A formula and worked example for AI agent trace volume, why payloads dominate size, and how sampling and tiering cut cost without losing audit evidence.
Evidence and frameworks
All in this topic- Guide
How do you map AI agent controls to compliance frameworks?
One AI agent control set mapped to OWASP agentic, MITRE ATLAS, NIST AI RMF, ISO/IEC 42001, the EU AI Act, DORA and SOC 2, with each framework's legal status.
What does PRA SS1/23 mean for AI agents in UK banks?
How the five PRA SS1/23 model risk principles apply to AI agents: inventory, tiering, governance, validation and mitigants, with an evidence map.
What does the FTC Safeguards Rule mean for AI agents handling customer data?
How the GLBA Safeguards Rule elements and the 500-consumer, 30-day FTC notice apply when AI agents touch customer information at non-bank lenders and fintechs.
How do you use the OWASP AI Testing Guide on AI agents?
Which of the 32 OWASP AI Testing Guide tests matter for AI agents, how to run them, and how to link results to OWASP Agentic Top 10 evidence.
Do LLM providers belong in the DORA register of information?
When model APIs and agent platforms count as ICT services in the DORA register, how to record supply chains, and where shadow AI keys break it.
When does an AI agent incident become an SEC disclosure matter?
Form 8-K Item 1.05 applies to material cybersecurity incidents, filed four business days after the materiality call. How agent incidents fit.
What does internal audit look for in AI agent governance?
A checklist of what internal audit tests in an AI agent governance review, by control area, with the test steps and the evidence to have ready.
What does NIST AI 100-2 say about attacks on AI agents?
NIST AI 100-2 E2025 classifies adversarial attacks on AI, adds an AI agent security section and expands indirect prompt injection. How to use it for agent tests.
What does the CISA AI Data Security guidance mean for AI agents?
The May 2025 joint CISA, NSA and FBI AI Data Security information sheet, mapped to AI agent data flows: RAG, memory, tool outputs and logs, with evidence to keep.
Buying and comparison
All in this topic- Guide
How to evaluate an AI agent governance platform: 30 questions for your RFP (2026)
A vendor-neutral checklist of 30 questions to ask when buying an AI agent governance or control plane platform, grouped by discovery, identity, enforcement, evidence and deployment.
What security tests should you run on a self-hosted AI governance platform before go-live?
A go-live security checklist for self-hosted AI governance platforms: auth after pod restarts, unauthenticated route sweeps, cache headers and SSO role mapping.
What disaster recovery should a self-hosted AI governance platform have?
Zone vs region failure, RTO and RPO per component, data-only replication vs warm standby, and what happens to gateway traffic when the platform is down.
How do you verify a self-hosted AI governance platform sends no telemetry out?
A hands-on test plan to prove a self-hosted AI governance deployment sends nothing outside your cloud: deny-by-default egress, flow logs, DNS logs.
Platform Guides
All in this topic- Guide
How do you govern AI agents across every platform they run on?
Native controls per platform, the gaps between them, a seven-point cross-platform baseline and rollout order for Bedrock, Foundry, Vertex, Agentforce and more.
What is Azure AI Content Safety groundedness detection?
How Azure groundedness detection checks answers against sources, its modes and inputs, the April 2027 retirement, and where it fits among guardrails.
How do you govern prompts with Amazon Bedrock Prompt Management?
Pin production to Bedrock prompt version ARNs, split IAM between authors and approvers, and tie each version to a ticket and regression run.
Which Kubernetes Pod Security Standards should AI agent workloads use?
Run AI agent pods under the restricted Pod Security Standard. What it requires, a namespace label example, and a rollout plan that avoids outages.
How do you govern HubSpot's remote MCP server?
HubSpot's remote MCP server uses OAuth 2.1 with PKCE and inherits user permissions. Risks, scopes and an 8-point policy for AI client access.
What admin controls exist for Claude in Chrome?
Claude in Chrome admin controls on Team and Enterprise: org toggle, site allowlist and blocklist, role permissions, and a safe site-list baseline.
How do you govern Claude Code subagents?
Claude Code subagents carry their own tools, models, hooks and MCP servers. How managed settings, permission rules and reviews keep them in policy.
How do you govern event-triggered autonomous agents in Copilot Studio?
Copilot Studio event triggers run on the maker's credentials. How to control them with data policies, environment settings and a launch checklist.
How do you control who can use Snowflake Cortex AI functions?
CORTEX_USER and USE AI FUNCTIONS are granted to PUBLIC by default. How to revoke them, grant narrower roles, and control models with RBAC.
How do you govern Snowflake Cortex Code?
Govern Snowflake Cortex Code with RBAC agent profiles, per-user credit quotas, managed settings, MCP policies and ACCOUNT_USAGE, plus the developer-device gap.
What does Agentforce Command Center show for agent oversight?
Agentforce Command Center metrics, alerts, Data Cloud session tracing and OpenTelemetry export explained, plus the gaps for cross-platform governance.
What does GitHub Agent HQ let enterprise admins govern?
GitHub Agent HQ enterprise controls explained: AI Controls, agent sessions, audit events, custom agents, managed permissions, and the gaps outside GitHub.
How do you govern several Claude organisations at once?
Admin credentials are scoped to one Claude organisation. How to connect each one, merge the inventory, apply one policy set and find the gaps.
Put the framework into practice