Category: Agentic AI

2 posts found

Jan 21, 2026

Least-Privilege MCP Agents: Capability Tokens, Tool Scopes, and a Policy Gate That Stops Bad Tool Calls

A concrete security design for tool-using agents: every tool call goes through a policy gate, with scoped permissions and auditable decisions.

By Ali Elborey

Jan 20, 2026

Agentic AI

Replayable Agent Workflows: Checkpoints, Time-Travel, and Regression Tests for Tool-Using Agents

A practical pattern to make agents reproducible: replay failed runs, fork from checkpoints, and turn real incidents into tests. Learn how to debug non-deterministic agent behavior with durable execution and time-travel debugging.

By Yusuf Elborey

Sign In

Category: Agentic AI

Least-Privilege MCP Agents: Capability Tokens, Tool Scopes, and a Policy Gate That Stops Bad Tool Calls

Replayable Agent Workflows: Checkpoints, Time-Travel, and Regression Tests for Tool-Using Agents