agent-security
Safeguard articles tagged "agent-security" — guides, analysis, and best practices for software supply chain and application security.
33 articles
Out-Of-Band Confirmation For Irreversible Tool Calls
Some tool calls cannot be undone. Out-of-band confirmation is the cheapest defense for that small set, and the most expensive thing to skip.
How to Secure AI Agents on the MCP Protocol
MCP gives AI agents real tools, real credentials, and real blast radius. Here is a hardening guide for running MCP servers in production without torching your environment.
MCP Server Capability Drift Detection
MCP servers do not stay still. Tool surfaces drift, scopes expand, and the server you approved is not the server in production. Here is how to catch that.
MCP Server Authorization Patterns in 2026
The Model Context Protocol shifted agent integration from custom glue to a standard surface. Authorization patterns that work, and the ones that keep biting teams.
Prompt Injection Defence Stack 2026
No single control stops prompt injection. The current state of the art is a defence-in-depth stack with controls at five distinct layers. Here it is.
Griffin AI vs Claude Computer Use: Security
Claude's Computer Use lets an agent drive a GUI. For security, this is powerful and dangerous in equal measure. The architecture around it matters.
AI Agent Blast Radius Management
Every agent in production has a blast radius. Most teams have not measured theirs. Here is how to measure it and how to bring it under control.
Eval Harness As Release Gate For AI Features
Shipping AI features without an eval harness is shipping without tests. Here is how to build one that actually gates releases without becoming a bottleneck.
LLM Selection For Security Workflows
Picking a model for a security workflow is not the same as picking one for a chatbot. Here are the criteria that actually matter and how to weigh them.
ToxicSkills: When Claude Skills Become a Malware Distribution Channel
Snyk's ToxicSkills research found prompt injection in 36% of Claude skills tested and 1,467 malicious payloads. The SKILL.md trust model is the structural issue.
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.
Anthropic MCP Security Model: A Deep Dive
Anthropic's Model Context Protocol introduces a new trust boundary between agents and tools. Here is how the security model actually works in practice.