Brian Turcotte

A productive coding agent needs more than a good prompt. It needs the right context, tools, instructions, and boundaries. This workshop explores how to set up your IDE or coding agent before you start writing code, so it can work more consistently across tasks, projects, and teams.

Learn how to organize reusable skills, connect tools and information through Model Context Protocol (MCP) servers, define custom agents for specific responsibilities, and establish configurations that keep everyone working from the same expectations. We’ll also cover practical guardrails: what agents can change, when they should ask for approval, and how to check their work.

The focus is building a workspace you can maintain and expand without accumulating conflicting instructions, unnecessary tools, or agents with unclear roles.

What you’ll come away with

• A repeatable setup checklist for preparing an agentic workspace before development begins.
• A framework for choosing the right approach: when to use a skill, an MCP server, a custom agent, or a simple instruction.
• A structure for reusable skills and configurations that can grow across projects and teams.
• Clear agent responsibilities that make delegation and handoffs easier to manage.
• Practical guardrails for tool access, permissions, code changes, testing, and human review.
• A way to evaluate your setup using representative tasks, identify friction, and improve it over time.