Anti-patterns: things to avoid

2026-03-04 · Source: Simon Willison's Weblog · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

This content identifies a critical anti-pattern in "agentic engineering": submitting unreviewed, agent-generated code to collaborators via pull requests. The author emphasizes that developers are responsible for ensuring agent-produced code is functional and ready for review, rather than delegating this initial validation to others. A good agentic engineering pull request should feature working, small, well-contextualized code, with human-reviewed descriptions. To demonstrate due diligence, developers are advised to include evidence of their self-review, such as testing notes or screenshots, ensuring reviewers' time is respected. This discussion is a chapter from the comprehensive guide, "Agentic Engineering Patterns."

Key takeaway

A critical anti-pattern in agentic engineering is submitting unreviewed, agent-generated code in pull requests, effectively delegating your initial validation work to collaborators. To mitigate this, developers must self-review agent-produced code, ensuring it's functional, small, well-contextualized, and accompanied by validated descriptions and evidence of manual testing or specific implementation choices. This discipline significantly improves team collaboration, code quality, and overall efficiency in agent-assisted development workflows.

Topics

• Agentic Engineering
• Code Review
• Pull Requests
• Software Development Anti-patterns
• AI Code Generation

Best for: Software Engineer, AI Engineer, Machine Learning Engineer

Related on AIssential

• Agentic Engineering — Effective agentic engineering requires structured workflows to plan, steer, and verify AI-generated code.
• Choosing the Right Agentic Design Pattern: A Decision-Tree Approach — Agentic design pattern selection requires a structured decision process based on task properties.
• Before You Hit Push, Review Your Own Code — Proactive self-review, augmented by AI, elevates engineering quality and team collaboration.
• Introducing Google Antigravity 2.0 — Agentic coding is a necessary but insufficient step towards general model intelligence, requiring co-optimized layers and user-friendly interfaces.
• Anthropic Introduces Agent-Based Code Review for Claude Code — Anthropic's agent-based code review system uses multiple AI agents for deeper, parallel analysis of pull requests.
• Agentic Engineering Is Not Vibe Coding: The Patterns That Actually Work — Distinguishing "vibe coding" from "agentic engineering" is crucial for effective AI-driven software development.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.