.NET Agent framework: tool approval in harness with Ollama

Authorised Territory demonstrates how a .NET AI agent running in a harness can request human approval before executing a tool, using a locally hosted LLM via Ollama.

Overview

The video shows an AI agent implemented with the .NET Agent framework running inside a harness, where the agent can pause and ask for an explicit approval step before it proceeds to run a tool (a human-in-the-loop safety/control pattern).

Key points called out in the video description:

• The tutorial focuses on tool approval: the agent requests approval before executing a tool.
• The LLM runs locally using Ollama.
• A starting solution for the tutorial is provided in a previous video: https://youtu.be/AZNavfhjgSs