Agentic learning via skills

See how an agent learns new information from external data, solves problems, and updates itself, enabling faster experimentation and self-learning.

Overview

I will be demoing the ability to teach a Claude Code agent new information from an external data source (Header). The agent will describe the problems it’s currently facing, ask the external data source for potential solutions to those problems, then implement them locally within it’s Claude.md file.

The goal of this pattern is to promote experimentation around agent setups and begin the path towards self-learning agents.

Tech stack