macOS xterm-ghostty fish 4 views

A step-by-step walkthrough demonstrating the capabilities and workflow of the AI-powered command-line coding assistant, mlc (via uvx). This recording showcases how to iteratively prompt an AI agent to build, modify, translate, and refactor code, while heavily leveraging advanced session management features like branching, multiple tabs, session resuming, and terminal file inspection:

  • Initial Setup: Launching mlc using uvx to generate and immediately execute a standard fizzbuzz.py script.
  • Iterative Prompting: Prompting the agent to modify the logic (adding “woof” for multiples of 7) and translating the entire script into Rust.
  • Agent Control & Navigation: Utilizing native slash-commands (/diff, /branch, and /tab 1) to track file changes and manage parallel contexts.
  • Session Persistence: Demonstrating how to completely exit the agent (/exit --all) and seamlessly pick up exactly where you left off using mlc --resume <hash>.
  • Refactoring & Verification: Asking the agent to completely rewrite the logic using recursion, verifying the file outputs via cat, and using the yazi terminal file manager for final directory inspection.

Perfect for developers interested in terminal-based AI agents, LLM tool-calling workflows, and advanced CLI session management.

More recordings by albersj66