Agent Hooks
Secure your agentic IDEs and agent frameworks with Guardrail Hooks — interception points that enforce security policy at every high-risk step
Lessons (7)
1. Introduction to Agent Hooks
Understand what makes agentic systems different from traditional AI and how Guardrail Hooks provide mechanical security enforcement at every critical step
2. The Threat Landscape for Agentic Systems
Learn the attack vectors unique to agentic IDEs and agent frameworks — and why traditional security tools are not designed to catch them
3. The Four Hook Points
Deep dive into the Pre-LLM, Post-LLM, Pre-Tool, and Post-Tool hooks — what each one inspects, what it can block, and how they provide layered defense
4. Installing Hooks on Your IDE
Step-by-step installation of Guardrail Hooks for Cursor, Claude Code, Kiro, GitHub Copilot, and Claude Desktop
5. Installing Hooks for Agent Frameworks
Add Guardrail Hooks to LangChain, LangGraph, CrewAI, OpenAI Agents SDK, and Strands with framework-specific integration snippets
6. Configuring Your Guardrail Policy
Master the guardrails_config.json file — hook events, detectors, sensitive tools, per-workspace vs global config, and common configuration patterns
7. Observability and Monitoring
Use hook event logs to investigate incidents, track security trends, and produce audit-ready evidence of agent security enforcement