Code smells for AI agents: Q&A with Eno Reyes of Factory

2026-02-04 · Source: Stack Overflow Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

Factory, co-founded by Eno Reyes, offers a platform for large engineering organizations to achieve fully autonomous software development. Their frontier coding agent, Droid, is designed to perform tasks across the software development lifecycle and is model-agnostic, deployable across any environment, OS, or IDE. Factory emphasizes "harness engineering," a process involving hundreds of small optimizations to manage context, inject environment information, and handle tool calls for LLMs over long tasks. The platform also provides tooling to analyze codebase quality and the impact of agents, identifying hundreds of validation signals like compilation, linting, test passing, and documentation. This approach aims to prevent "slop code" by optimizing agents for high-quality signals, as research indicates that higher baseline code quality correlates with increased productivity when AI agents are introduced.

Key takeaway

For CTOs and VPs of Engineering aiming to scale software development with AI agents, your organization's existing code quality is the primary determinant of success. Prioritize instrumenting and enforcing robust code quality signals (e.g., linters, static analysis, comprehensive testing) before widespread agent adoption. This foundational work will enable agents like Factory's Droid to accelerate productivity rather than generate "slop code" requiring manual fixes, transforming your development workflow.

Key insights

High-quality codebases are crucial for AI agent productivity, as agents excel at pattern recognition.

Principles

• Agent effectiveness requires model and vendor agnosticism.
• Harness engineering is key to robust agent deployment.
• Hundreds of validation signals define code quality.

Method

Factory's Droid agent uses harness engineering to integrate with diverse environments and leverages extensive validation signals (linters, tests, security scanners) to autonomously improve code quality and automate software development tasks.

In practice

• Implement comprehensive validation signals in codebases.
• Use AI agents for automated code review and incident response.
• Deploy agents to fix missing quality signals.

Topics

• AI Coding Agents
• Code Quality Analysis
• Software Development Automation
• Harness Engineering
• Autonomous Software Development

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Software Engineer, AI Product Manager, MLOps Engineer

Related on AIssential

• Harness engineering and agent feedback: Exploring AI coding sensors — Computational sensors provide deterministic feedback, enabling AI coding agents to self-correct and improve code quality autonomously.
• PR reviews were already broken. AI made it worse — AI-coding agents exacerbate PR review issues by generating high-volume, low-quality code from insufficient data, demanding a layered verification approach.
• Building Pi, and what makes self-modifying software so fascinating — Human judgment and deliberate friction are crucial for maintaining software quality amidst rapid AI-driven code generation.
• addyosmani / agent-skills — AI coding agents can achieve production-grade software quality by following structured, verifiable engineering workflows and best practices.
• Vibe coding and agentic engineering are getting closer than I'd like — The increasing reliability of AI coding agents blurs the distinction between casual "vibe coding" and professional "agentic engineering."
• All Smoke, No Alarm: Oracle Signals in Agent-Authored Test Code — Agent-authored test files frequently lack meaningful verification logic, making test file presence an unreliable indicator of code quality.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Stack Overflow Blog.