Agent Security

8 pages in this section.

The Principle of Least Privilege Applied to Agent Tools

Learn how to apply the principle of least privilege to agent tools. Grant tools only necessary data, hosts, and permissions to prevent exploits.

Python Agent Security Explainer Least Privilege

Agent Security Basics

Learn agent security basics with 8 examples. Define a sandboxed code-execution tool with no network or filesystem access and strict limits.

Python Agent Security Basics Sandbox

Containerized Sandboxes for Code-Executing Agent Tools

Run untrusted agent code in isolated containers. Learn to configure read-only filesystems, network restrictions, resource limits, and secure execution.

Python Agent Security Containers Sandbox

Scoping API Keys and Credentials Passed to an Agent

Python Agent Security Advanced Credentials

Never Grant Blanket Shell Access in Production: What to Do Instead

Avoid unrestricted shell access for production agents. Learn safer alternatives like structured tools, allowlisted binaries, and sandboxes to prevent security risks.

Python Agent Security Explainer Shell

Network Egress Controls for Agents That Call External Tools

Python Agent Security Advanced Egress

Security Checklist Before Shipping an Agent to Production

Review this checklist before deploying an agent to production. Learn to assess scope, threat models, autonomy, and apply least privilege to tools.

Python Agent Security List Checklist

Agent Security Best Practices

Implement ten agent security best practices for sandboxing, credential scoping, and containment. Learn to shrink action spaces and isolate execution.

Python Agent Security Best Practices List

Agent Security

Search Documentation

Agent Security