Theo Browne examines coding-agent features he finds useful, including scripts invoked from skillsAn AI agent skill is a reusable package of instructions, resources, and tool guidance for performing a bounded kind of work., imported instruction files and separate local preferences. He discusses the security implications of executing code from installed skills.
Browne demonstrates a staged pull-request auditCode review examines proposed source changes to assess their correctness, clarity and suitability before or around integration. using parallel agentsA multi-agent system contains multiple AI agents that interact, coordinate, divide work, or influence one another while pursuing tasks. and programmatically generated promptsPrompt engineering is the practice of designing and refining instructions, context, examples, and constraints to obtain useful AI outputs.. He explains how the workflow organizes review and verification, while warning that it can consume substantial usage.
The presentation also covers side questions, branching, rewind and remote access. His feature comparisons and cost observations reflect his own experience rather than independent benchmarks.
Watch on YouTube
