commixproject/commix: technical due diligence, read from its repository

36% of commit subjects are exact repeats of another subject, and subject length varies by only 38.3 characters. A log written by someone explaining their reasoning to a future reader does not look like this. The log is therefore weak evidence of intent, whatever the code is like.

The median source file here is 78 lines. 9 files exceed 1,000 lines, the largest being 7,257. Size is not a defect on its own — some problems genuinely live in one place — but these are where merge conflicts, review fatigue and single-owner knowledge concentrate, and they are the first thing that slows an inheriting engineer down.

80 places catch an error and do nothing with it, across 26 files (0.9 per thousand lines). Each is a failure the system will not report. They surface later as problems nobody can explain, and they are cheap to fix now.

Of five ordinary markers — CI workflows, CODEOWNERS, a pull-request template, pre-commit hooks, contribution guidance — 1 is present: CI workflows. Branch protection itself is a GitHub setting rather than a file, so this is a reading of proxies and not of the rules themselves. What it does establish is that the process is not written down anywhere a new engineer would find it.

Five files out of 490 absorb 29% of all file-level change. Concentration like this is where debt is usually being serviced rather than repaid: the same code is reopened because it is load-bearing, fragile, or both. A repository cannot tell you which — that is a conversation with the team.

12 places run a command through a shell (`shell=True`, `os.system`, `exec` with a template) rather than passing arguments. Where any part comes from outside, it is command injection. Ask where each one's input comes from.