Vibe Coding to Agentic Engineering with Claude Code
The article explores how Claude Code enables a shift from "vibe coding" (prompt-driven, loosely guided development) to agentic engineering, where AI assists in structured, autonomous software development tasks like debugging, refactoring, and testing through natural language commands and multi-step reasoning.
Background
- "Vibe coding" is a term popularized by AI researcher Andrej Karpathy for a highly informal style of programming where the developer gives high-level prompts to an AI coding assistant (like Cursor, GitHub Copilot, or Claude) and blindly accepts its output without reading or understanding the code.
- "Agentic engineering" is a newer, more disciplined approach where AI is still the primary coder, but the human acts as a reviewer, architect, and quality gate — combining AI speed with human judgment.
- Claude Code is Anthropic's terminal-based coding agent that can write, edit, and test code autonomously across a full project, not just in isolated snippets.
- Apimatic is a company that builds API tooling and developer experience products; this post reflects their practical experience moving from casual AI-assisted coding to a structured workflow they call "agentic engineering."
- The shift matters because as AI coding tools become more powerful and autonomous, the human role is evolving from writing code line-by-line to directing, reviewing, and orchestrating AI agents — a change that affects how software teams are organized and what skills developers need.
Newer Claude models sometimes invent extra keys in tool call arguments, breaking validation in Pi's edit tool. The author suspects post-training for Claude Code's forgiving harness makes alternative schemas fail. This suggests closed RL training can degrade general tool-use reliability.
The article argues that Claude for Mac is an Electron app because its lead developer, Felix Rieseberg, is a former Microsoft engineer known for porting Slack to Electron and has a professional history strongly tied to the Electron framework.
Simon Willison released llm-coding-agent 0.1a0, an experimental coding agent built on his LLM library. The agent provides tools for reading, editing, searching files and executing commands, and ships with a Python API and CLI. It was developed using Claude Code (Fable 5) via TDD with a spec-first approach, and is available as a slop-alpha on PyPI.
Truth Social remains an outlet primarily for Trump's own posts, while other administration officials continue using X. The platform functions as a blog-like channel for Trump's messages rather than a genuine social network competitor.