Coding Agents

What are coding agent integrations

Coding agent integrations allow Agent Identity Security to connect to the AI coding tools that developers use on their machines. When a developer opens Cursor, runs Claude Code in a terminal, or invokes GitHub Copilot CLI, the Guardian that you installed detects the session and reports it to Agent Identity Security. Agent Identity Security then applies whatever security policies you have configured, for example blocking access to sensitive files, requiring approval before certain operations, or logging all activity for compliance purposes.

Installation modes

The Guardian can be deployed to a developer's machine in one of two ways:

Manual

Download the Guardian package from Agent Identity Security and run the installer directly on the developer's workstation. Best for pilots, single machines, or environments without a managed device fleet.

Via MDM

Push the Guardian bundle silently, at scale, through your existing device management platform, with no per-developer action required.

• MDM Intune setup — deploy the Guardian to managed Windows machines through Microsoft Intune.
• Jamf Pro — deploy the Guardian to managed macOS machines through Jamf Pro.

Both MDM platforms also power Shadow Agent Discovery, surfacing coding tools that are installed but not yet governed by a Guardian.

How it works

The process has four stages:

Install

Download the Guardian package from Agent Identity Security and run the installer on the developer's machine. The installer takes about two to five minutes.

Discover

Agent Identity Security automatically detects the next time a coding agent session starts on that machine. The agent appears in Agent Identity Security under Registry > Agents.

Govern

Agent Identity Security evaluates every tool call, file access, or command the agent makes against your configured policies.

Audit

All activity is logged to Agent Identity Security Event Logs, giving you a full audit trail.

Before you begin

Complete these one-time preparation steps before installing any Guardian:

1. Sign in to Agent Identity Security. Go to your Agent Identity Security URL (for example, https://acme.stage.eoscyber.ai, replacing acme with your company's subdomain) and sign in with your Agent Identity Security credentials.
2. Generate an API key. In Agent Identity Security, go to Settings > API Keys and generate a key with the scopes audit_logs:write, agent_discovery:write, and policy:read. Copy the key and store it safely; it is only shown once.
3. Note your Tenant ID. For SaaS deployments, your Tenant ID is the company-specific part of your Agent Identity Security URL. For example, if your URL is https://acme.stage.eoscyber.ai, your Tenant ID is acme.
4. Confirm Python is installed. All Guardian packages require Python 3.9 or later. Run python --version or python3 --version in a terminal to check.

Vendor integrations

Agent Identity Security ships a Guardian for each of the following coding agent platforms:

Each vendor topic follows the same structure documented at Coding Agents: prerequisites, a step-by-step configuration walkthrough, and a confirmation of what Agent Identity Security discovers after the Guardian reports its first session.

Related topics

• Agent Discovery
• Non-Coding Agents
• Shadow Agent Discovery
• Managing Integrations