Google Cloud launches IAM Helper to let AI agents handle access management

What IAM Helper is

Google Cloud describes IAM Helper as an agent-native orchestration and interface layer for Google Cloud IAM. It augments the company's core IAM engines, including Policy Intelligence and Privileged Access Manager (PAM), and makes IAM skills callable directly by autonomous code assistants, deployment pipelines and custom agents.

The problem it targets, according to the company, is that developers who rely on agents to provision cloud resources, write Terraform and deploy microservices at machine speed see their CI/CD pipelines blocked even faster as security friction scales. Manual troubleshooting of permission blockers slows engineers and administrators, while shortcuts can lead to broad agent over-permissioning that puts least privilege at risk.

IAM Helper shifts the burden of policy evaluation away from human administrators and LLMs, Google Cloud says. Complex orchestration is offloaded to the Google Cloud backend, which the company says maximizes token efficiency.

Agent Skills and built-in guardrails

The workflows are packaged as IAM Helper Agent Skills using the open Agent Skills standard. They combine end-to-end workflows, helper scripts and guardrails, and can orchestrate multiple underlying functions and APIs. Google Cloud says they work across common agents: Gemini CLI, Gemini Cloud Assist, ADK, Claude Code and Cursor.

Every skill separates read-only discovery from policy changes. Diagnostics and simulations run automatically, while any change to access policies pauses for human-in-the-loop (HITL) confirmation based on customizable risk tiers.

Google Cloud lists four skills:

• iam-helper-for-policy-management generates and validates IAM v1 allow bindings and IAM v2 deny rules, and blocks risky configurations such as public access (allUsers) and broad basic roles (roles/owner, roles/editor) before they are applied.
• iam-helper-for-troubleshooting identifies the exact missing permission or blocking rule behind a 403 PERMISSION_DENIED error. Developers can check whether they can unblock themselves or generate a structured ticket for an administrator (Requester flow). Security administrators can trace inherited allow and v2 deny rules across organization, folder and project hierarchies (Resolver flow).
• iam-helper-for-policy-simulator walks an agent through a plan, simulate, analyze and apply flow, testing proposed changes against 90 days of historical access logs to confirm existing traffic won't be disrupted.
• iam-helper-for-privileged-access-management manages on-demand credentials through PAM, with requester, administrator and approver workflows. Elevated access is audited, time-bound and automatically revoked once the task is done.

Two example scenarios

Google Cloud offers two scenarios. In the first, a developer's coding agent hits a 403 error while deploying a new application. Instead of waiting on a support ticket, the agent uses the troubleshooting skill to identify the missing Pub/Sub permission in seconds, then uses the privileged access skill to request a one-hour JIT grant.

In the second, an infrastructure agent meets a permission block during a routine deployment pipeline. Rather than failing the build, the pipeline generates a revised, least-privilege policy and verifies against 90 days of logs via the policy simulator skill that active services won't break. The change is then routed to an admin for quick approval.

MCP tools underneath

All the skills are based on a common set of foundational MCP Tools. These are atomic, read-only functions exposed via Google-managed MCP servers across core IAM endpoints: Policy Troubleshooter, Policy Analyzer, Policy Simulator and the IAM v1 and v2 APIs.

Google Cloud says teams no longer need to write custom API wrappers. Agents can query the backend and get exact answers without parsing verbose policy documents in the prompt. Hierarchy traversal, group expansion and policy evaluation run on dedicated IAM endpoints, which the company says keeps skills fast and accurate with minimal token usage.

What it means for teams

Teams can also use the MCP tools directly when writing their own custom Agent Skills, for example to combine Google Cloud's IAM capabilities with internal ticketing systems, custom compliance checks and existing deployment pipelines.

For organizations adopting agents, the pitch is that routine permission blockers can be resolved by the agent itself, while policy changes still pass through human approval. Google Cloud points developers to its Agent Skills Catalog and the official IAM documentation to get started.