ai-agents
Safeguard articles tagged "ai-agents" — guides, analysis, and best practices for software supply chain and application security.
69 articles
AI Agent Tool Confused Deputy Problem in 2026
A senior engineer's take on the confused deputy problem in AI agent tool use, why it keeps reappearing in 2026, and the architectural patterns that actually fix it.
MCP Transport Layer Security Options
MCP supports stdio, streamable HTTP, and a handful of experimental transports. Each has distinct security properties, and the choice of transport constrains every other security decision you make about the deployment.
Securing MCP Servers in the Enterprise: A Practical Guide
MCP servers connect AI agents to your infrastructure. Here's how to secure them without killing the productivity gains.
Griffin Agent Loop: Design Decisions
The design rationale behind Griffin, Safeguard's triage agent — how the loop is structured, why we bounded reasoning depth, and how tool calls stay auditable.
MCP Server Discovery Protocol Security
MCP server discovery turns a client connection string into a live capability graph. The protocol mechanics that make this convenient also widen the blast radius when discovery is spoofed, tampered with, or silently reshaped mid-session.
Agent Security Buyer's Guide Overview
As AI agents gain production write-access across the enterprise, security teams need a rigorous buyer's guide to separate real agent security platforms from repackaged AppSec dashboards.
MCP vs Skills vs Hooks vs Rules Explained
MCP, Skills, hooks, and rules extend AI coding agents in very different ways — two execute code, two only steer behavior. Here's how each one breaks.
AI Agent Security Risks: Why Autonomous Systems Are the Next Supply Chain Frontier
AI agents are consuming APIs, installing packages, and executing code autonomously. The security implications are massive and largely unaddressed.
AI Agent Tool-Scope Enforcement Patterns
Agents get tool lists, not tool boundaries. We walk through scoping patterns that actually hold when Claude 4 or GPT-5 picks the wrong function at runtime.
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.