On demand
Claude Code Advanced Patterns: Subagents, MCP, and Scaling to Real Codebases
You’ve completed your first tasks with Claude Code. You know how to give it context, review its output, and land a commit. This session is about what comes next — the patterns that separate teams using Claude Code occasionally from teams that have made it part of how they ship every day.
We’ll go deep on subagents and hooks for orchestrating multi-step work, MCP integrations that connect Claude Code to your internal tools and services, and the context strategies that keep it effective in large, messy, real-world codebases.
Featured speakers
Lizzie Alvarado Ford
Applied AI @ Anthropic
Alon Krifcher
Applied AI @ Anthropic
Kacie Jenkins
Claude Code Marketing @ Anthropic
What you’ll learn
How subagents and hooks turn Claude Code into an orchestrator — running parallel tasks, enforcing guardrails, and chaining tool calls without babysitting every step
Wiring Claude Code to your internal services with MCP, and knowing when a custom tool is worth building versus when a script will do
Strategies for large repos: structuring CLAUDE.md for monorepos, managing context windows across hundreds of thousands of lines, and keeping Claude focused on the task at hand
Where teams are putting Claude Code in their CI pipelines — automated PR review, test generation, and the patterns that catch regressions before a human sees them