Coding agents have replaced every framework I used
A provocative blog post suggests that coding agents have fundamentally reshaped software engineering, eliminating the need for traditional frameworks and boilerplate code. The author argues this shift frees developers to focus on core architectural problems, returning the craft to its 'true' engineering roots. This perspective has ignited a fiery debate on Hacker News about the future of development, the value of frameworks, and the evolving role of the human engineer.
The Lowdown
The author, Alain, contends that coding agents powered by frontier models have revolutionized his development process since late 2025, enabling him to build complex products from the ground up without relying on traditional frameworks. He likens this to 'automated programming,' where the tedious manual labor of typing every line of code is eliminated, allowing him to act as an architect focusing on high-level design and critical decisions.
- Framework Critique: Alain argues that frameworks represent 'intellectual surrender,' solving problems that shouldn't exist and introducing unnecessary complexity. He dismisses their purported benefits of 'simplification,' 'automation' (boilerplate), and 'labor cost' (creating 'React Developers' instead of true software engineers).
- Return to Engineering: With agents handling boilerplate, he can now create purpose-built tools instantly, using simple, foundational technologies like Bash for 99% of his needs. He asserts this enables engineers to solve current problems, not hypothetical future ones, and build systems that are truly 'theirs.'
- Revolutionary Shift: The author believes this change is a profound opportunity, akin to historical industrial revolutions, allowing developers to shed useless complexity and embrace the meaningful challenges of product development.
In essence, Alain believes the revolution is already here, with tools and models available to empower developers to build directly and intentionally, rather than being constrained by the abstractions and lock-in of third-party frameworks.
The Gossip
Framework Fanfare & Fears
Many commenters vigorously defend frameworks, highlighting their role in providing battle-tested solutions, security, and a common structure for team collaboration and onboarding. They argue that frameworks embody collective knowledge and engineering best practices developed over years, cautioning that AI-generated bespoke code might reintroduce silent bugs, security vulnerabilities, and maintenance headaches due to a lack of community support and rigorous testing. The consensus among these commenters is that abandoning frameworks entirely for AI-generated code is a risky proposition that overlooks critical benefits.
Agentic Agitation: Empowerment or Erosion?
The discussion delves into whether coding agents truly empower developers or merely shift the burden, potentially leading to a shallower understanding of the underlying code. Some argue that AI frees them from 'code monkey work' and tedious tasks, allowing more focus on design and problem-solving, making the development process faster and more enjoyable. Conversely, others express concern that relying on AI might hinder the development of deep intuition and mastery, turning engineers into 'operators' of AI-generated black boxes and increasing the risk of technical debt and incomprehensible systems.
The Evolving Engineer's Embrace
Commenters reflect on the changing nature of software engineering itself. Some embrace AI as a tool that enhances optionality and allows for building highly customized, efficient solutions, even sharing personal anecdotes of using AI for embedded development to replace extensive libraries with purpose-built functions. Others emphasize that the real challenge remains understanding the problem domain and architecting maintainable solutions, suggesting that AI doesn't solve these hard problems. There's also debate on whether the 'pain of writing code' is real (tedium vs. intellectual challenge) and if AI is truly saving time or just reallocating effort.
Skill Shift & Semantic Scrutiny
The impact of AI on developer skills and learning is a key point of contention. Some believe AI can rapidly accelerate learning and fluency in new languages or domains by reducing friction. However, many worry that relying too heavily on AI might prevent new engineers from developing fundamental skills and a deep understanding of code, akin to skipping basic math for a calculator. There's also a strong sentiment that while AI can generate code, the human role of critical review, understanding ecosystem changes, and ensuring code quality remains paramount, especially given AI's non-deterministic nature.