Software Architecture: Why a green linter check alone isn't enough

TVS News AI, editorially supervised, presents Software Architecture: Why a green linter check alone isn't enough.

You know the feeling: the CI pipeline is green, every linter rule is satisfied – and yet the API still doesn't feel right. A new piece on heise+ tackles exactly this tension and argues that a clean automated check is far from a quality verdict on software architecture.

What's at stake: Linters and other automated tools check syntax, style, and simple patterns. Whether an API is professionally convincing, whether interfaces are sensibly cut, and whether a system remains viable in the long term – these questions they don't answer. The piece formulates its core thesis as follows:

A green linter check doesn't mean an API is professionally convincing. Architecture requires more than automation – it needs context and responsibility.

With that, the text brings to the fore a point that gets lost in many development teams: Quality isn't produced by the tool, but by the judgment of the people using it. Automation can secure standards and ground discussions in facts – but it can't decide whether a domain is modeled correctly or whether a design will hold up over time.

What architecture must deliver beyond the linter

Whoever carries responsibility for software faces daily decisions that no rulebook can resolve: What assumptions about users are baked into the model? How does the system behave under load that hasn't been measured yet? Which trade-offs are documented – and which are only implicitly hidden in the code? These questions need deliberate answers, not checkmarks on a list.

At the same time, the piece warns against demonizing automation wholesale. Linters, type checkers, and architecture tests create a shared language within the team and catch gross errors. They relieve the burden – but they don't replace a substantive engagement with the problem the software is meant to solve.

Context beats rulebook

The message can be reduced to a simple denominator: Context beats rulebook. Whoever evaluates an API only by whether all linter rules light up overlooks the real architectural work – weighing alternatives, naming assumptions, and taking responsibility for the consequences of a design decision.

The full piece has appeared in the heise+ section on heise.de and is aimed at developers who want to sharpen their view beyond pure tool output.