Agentic Engineering Patterns
simonwillison.net
March 4, 2026
1 min read
🔥🔥🔥🔥🔥
68/100
Summary
Agentic engineering patterns provide strategies for optimizing interactions with coding agents such as Claude Code and OpenAI Codex. Key areas include principles for effective use, testing and quality assurance, code comprehension, and annotated prompts.
Key Takeaways
- Agentic engineering patterns are designed to optimize the performance of coding agents like Claude Code and OpenAI Codex.
- The patterns include principles, testing and QA methods, understanding code, and annotated prompts.
- The guide provides a structured approach to achieve better results when working with AI coding tools.
Community Sentiment
Positives
- The marginal cost of code is collapsing, which could revolutionize software development and lead to more efficient engineering practices.
- AI's ability to generate functioning output, even if not perfect, is impressive and signifies a shift in how developers approach coding.
- Agentic coding is particularly effective for easily testable software, allowing for rapid iterations and improvements in projects like compression algorithms.
- Red/green TDD with AI tools like Claude helps manage complexity in larger projects, ensuring that changes do not inadvertently break existing functionality.
Concerns
- Despite the potential of agentic work, many users remain unimpressed with the quality of the output, raising concerns about reliability.
- The limited context windows of AI models like Claude can hinder their understanding of complex systems, leading to potential integration issues.
