Agentic Engineering

The skills, prompts and checks I use when working with coding agents.

I keep these in my agentic-engineering repository. Below are the workflows I reuse and what I've written about working with agents.

Skills I've written

Reusable workflows for understanding code, checking changes and testing applications. Each link opens the skill's instructions, references and scripts on GitHub.

  • Finds hotspots where frequent changes meet complex code, checks a quality baseline in CI, and explains the factors behind a pull request's risk score.

  • Builds a headless tool that drives an application through its real entry points and records sessions as replayable action logs.

  • Quizzes you on your codebase and its language, one question at a time, with code you write by hand and a journal of weak spots.

Prompts

Prompts I reuse to keep agent instructions and skills manageable.

  • Splits a long AGENTS.md into a short root index and task-specific files that agents load when needed.

  • Restructures a skill into a lean SKILL.md, reference files and scripts while preserving its instructions.

Requirements and installation steps live in the README. The repository also includes a skill validator that checks structure, links and script syntax.

Skills I use from others

  • Engineering skills for investigation, implementation and review, with an emphasis on verifiable changes and keeping code simple.

  • Matt Pocock's engineering skills. I use grilling, domain-modeling, codebase-design, retro and wayfinder almost daily.

Writing on agents