AI Security
Model Context Protocol Permissions Model Explained
MCP's permissions model is subtle. Here is a careful walkthrough of how tool scoping, sampling, and resource access actually work in production.
Apr 12, 20266 min read
Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
MCP's permissions model is subtle. Here is a careful walkthrough of how tool scoping, sampling, and resource access actually work in production.
Patterns for managing MCP servers through development, staging, rollout, and deprecation — with an eye on the security gaps that appear at each transition.
The MCP client surface is often overlooked. We examine trust boundaries, schema handling, credential storage, and safe defaults for the agent side of the protocol.
Enterprise MCP deployments need more than a static API key. The protocol is evolving toward OAuth 2.1 and dynamic client registration, and understanding which pattern fits which workload decides whether your rollout survives the first audit.
A side-by-side security comparison of Windsurf and Sourcegraph Cody: data handling, agent scope, deployment models, and enterprise controls.
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.
A practical look at rate-limiting patterns for Model Context Protocol servers, covering per-tool quotas, token budgets, burst control, and abuse-resistant designs.
Cursor's 2026 security model introduces privacy modes, indexing controls, and agent sandboxes. Here is the enterprise-ready view of what works.
Practical guidance on isolating tenants on shared Model Context Protocol servers, covering identity, data, compute, and observability boundaries at production scale.
Weekly insights on software supply chain security, delivered to your inbox.