The pitfalls of coding agents: why backpressure is the missing ingredient
By
Lucas F. Costa
A baker's-dozen of insight crammed into one ring.
Summary
The article critiques the two common approaches to using coding agents (LLMs for code generation). The first approach—letting the agent run unattended—is fast but leads to bugs, chaotic changes, and unmanageable PRs that erode code review standards. The second—treating the agent as glorified autocomplete with human review of every step—is safer but defeats the purpose of using an agent by being too slow. The article argues for a middle ground: applying "backpressure" mechanisms to balance speed and safety when using AI coding agents.
Key quotes
· 3 pulledThe first is to let the LLM run unattended and hope the repository survives. This is fast, exciting, and stupid.
The second approach is to treat the agent like glorified autocomplete and force a human to review every tiny step. This is safer, but slow enough to partially defeat the purpose of using an agent in the first place.
It leads to bugs, confused changes, and a flood of PRs that humans cannot review quickly enough, at least not without eventually lowering their standards and merging things they do not really understand.
You might also wanna read
How I Used Coding Agents to Automate My AI Research Work in Copilot Applied Science
An AI researcher shares their experience using coding agents to automate intellectual work, specifically building agents that automate parts
GitHub·2mo ago
How AI coding agents are reshaping social science research: Opportunities and concerns
This article examines how AI coding agents are transforming social science research by automating core research tasks traditionally performe
share.google·4d ago
Scorecard CEO warns of AI agent dangers in high-stakes domains, offers evaluation platform
Darius, CEO of Scorecard, shares a cautionary tale about building AI agents in high-stakes domains. He describes how his EMR agent for docto
Product Hunt·7mo ago
Scorecard: Platform for Evaluating and Optimizing AI Agents in High-Stakes Applications
The CEO of Scorecard shares a cautionary tale about nearly shipping a dangerous AI agent for doctors that confused pediatric and adult dosin
Product Hunt·7mo ago
Practical Guide to Using AI Coding Tools for Responsible Development
The article provides practical guidance for developers on responsibly integrating AI coding tools into their workflow. Based on two years of
Smashing Magazine·4mo ago
Claude Code Launches Multi-Agent AI Code Review System for Bug Detection
Anthropic's Claude Code now offers a multi-agent AI code review system that analyzes pull requests to catch bugs, security issues, and logic
Product Hunt·3mo ago