Descriptions:
John Linquist of egghead.io joins Claire Vo on How I AI to share the advanced Claude Code and Cursor techniques he uses daily as a senior software engineer — content explicitly aimed at experienced developers and engineering leaders rather than beginners.
Linquist’s centerpiece technique is using Mermaid diagrams as compressed, pre-loaded context for AI coding tools. Instead of letting the model explore files on every request, he generates diagrams that map authentication flows, database operations, and event channels across the codebase, then injects them into Claude Code’s context window upfront. This cuts response latency and significantly improves accuracy on multi-file tasks. He notes that Windsurf has shipped a similar concept under the name “code maps,” signaling broader industry movement toward structured context pre-loading as a standard practice.
The episode also covers building custom CLI tools with Claude Code — small, purpose-built command-line interfaces that expose specific workflows without the distraction of building a full web UI. Linquist demonstrates a CLI that uses Gemini models to generate themed image variations from structured prompts, walking through how the constrained terminal environment actually accelerates prototyping by eliminating UI scope creep. Custom hooks and automation patterns for composable AI-powered workflows round out the session. For senior engineers, VPs of engineering, and CTOs looking to move beyond basic autocomplete and get structurally better output from tools like Claude Code and Cursor, Linquist offers a concrete and immediately applicable technical playbook.
📺 Source: How I AI · Published January 26, 2026
🏷️ Format: Tutorial Demo
