Descriptions:
Matthew Berman, who uses every major AI coding agent professionally, breaks down the advanced techniques separating casual AI coders from true power users in this comprehensive tutorial. The video covers the full stack of expert vibe-coding practices across tools including Cursor, Codex, Claude Code, Devin, and Factory — with particular focus on configuration, automation, and workflow design rather than basic prompting.
Berman walks through setting up agents.md and Claude.md rule files to define model personality, commit message style, and project-specific behavior, then moves into building reusable skills (slash-command shortcuts for repeated prompts) and configuring automations in both Cursor and Codex. Concrete examples include a GitHub automation that comments on pull requests and a nightly documentation sweep that scans the codebase and opens a PR with updates — all running without manual intervention. He also covers Greptile, a sponsored code-review tool that integrates directly into these agentic workflows.
The centerpiece announcement is Berman’s newly launched Loop Library — a free, growing collection of practical agent loops hosted at signals.fordfuture.ai/loop-library. A loop is defined as any repeating agent process with a trigger, a recurring action, and a defined end goal. The broader thesis is that AI coding efficiency is less about model selection and more about systematic infrastructure: reusable skills, automated triggers, and well-designed loops that let agents handle orchestration while the developer focuses on higher-level decisions.
📺 Source: Matthew Berman · Published June 18, 2026
🏷️ Format: Tutorial Demo
