pawel-cell/micky-podcast-agentic-engineering

micky-podcast-agentic-engineering collects four packaged skills drawn from David Ondrej's interview with Micky, also known as Michael Shimeles, about building serious AI systems and what the author calls agentic engineering. The repo bundles these as proper SKILL.md files rather than loose prompts pasted into a chat. With 43 stars and no listed language, it sits in the category of agent-instruction kits rather than runnable code. The first skill is the Agentic Engineering Workflow. The README describes it as the full loop covering the agent setup, the context an agent needs, how to structure the work, a review cycle, launching, and security. This is the central skill, and the other three feed into it. The second skill, Source Code as Agent Context, addresses a specific failure mode: the agent guessing from documentation rather than reading the actual package or repository source. The skill is meant to push the model toward referencing real code when it answers questions or makes changes. The third skill is Code Structure Cleanup. It is a refactoring routine that takes mechanics which have been repeated across a feature and pulls them out into reusable service-layer modules. The README is explicit that this runs after a feature already works, not before. The fourth skill is the Grep Loop Review Workflow. This is the small-PR review pattern: open a small pull request, get AI and human review on it, apply the feedback, run the tests, and repeat the cycle until the change is ready to merge. The README also points to Michael Shimeles' own public skills repository on GitHub as an external resource, so anyone who finds this bundle useful can trace the material back to its source. Beyond these four files and the link, the README does not describe install steps, licence, or any usage tooling.