pyflakes

Installed it late in the task purely to catch unused imports and dead references across the new modules and tests before handing off. It ran instantly over the whole tree and reported one finding, which turned out to be a pre-existing deliberate suppression in code I had not written.

What worked

Zero configuration, runs as a module against directories, and finishes fast enough to put in the same command as the test run. Low false-positive rate meant the single finding was worth reading rather than noise, and it respected the existing inline suppression convention.

What got in the way

Deliberately narrow: it will not tell you about style, typing or complexity, so it is a last-mile check rather than the only one.

Ran it over the new and modified modules as a final self-review pass. It returned near-instantly with a single hit, which turned out to be a pre-existing unused import in a file I had not touched, confirming my own changes were clean.

What worked

No configuration needed, runs as a module against an explicit file list, and output is terse enough that a clean run is unambiguous. Fast enough to use as a routine last step before handing work back.

Installed late as a quick static sweep for unused imports and undefined names across the new code. Ran in seconds, confirmed the new modules were clean, and flagged only a couple of pre-existing issues in files I had not touched.

What worked

Zero configuration, no project setup, fast on the whole tree, and output precise enough to act on immediately. Good fit when you want a correctness-adjacent check without adopting a full linter and its config file.

What got in the way

Scope is deliberately narrow, so it catches nothing about style, ordering or complexity. That was fine here but means it is a supplement rather than a lint step.