Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

Thanks @victorhooi , and your instinct is right on both counts: the README was wrong, and the default is more confusing than it should be. I traced it through the code:

Your correction is accurate. The effective default really is ~/.cache/all-smi/records/all-smi-record.ndjson.zst. It comes from Settings::default() always setting record.output_dir = "~/.cache/all-smi/records" (src/common/config_file.rs), which RecorderOptions::from_args_with_settings joins with the default basename even when no config file is present (src/record/mod.rs).

Why it feels confusing- the default is surfaced inconsistently across three places:

• all-smi record --help still advertises the --output default as all-smi-record.ndjson.zst (src/cli.rs), i.e. the current directory;
• the source doc comment says the compiled default is ./all-smi-record.ndjson.zst (src/record/mod.rs);
• only the README (after your fix) says ~/.cache/....

So this PR fixes the README, but --help- where most people look first- would still point at the working directory. On top of that, the effective path is computed by string-matching --output against the magic basename to detect "operator didn't pass -o", which is brittle (the code comment itself says "if that ever changes, update this guard") and has a latent wart: passing -o all-smi-record.ndjson.zst explicitly gets silently redirected into the cache dir.

Cross-platform caveat: the ~/.cache/... default is a hard-coded string expanded via dirs::home_dir() (src/common/paths.rs), not the platform-aware dirs::cache_dir(). So on Windows it lands at C:\Users\<you>\.cache\all-smi\records\... (a Unix-style dotdir in the profile) and on macOS at ~/.cache/... rather than ~/Library/Caches/...- functional, but not where users on those platforms expect cache data. The config path, by contrast, already uses dirs::config_dir() (%APPDATA%\all-smi on Windows), so there's an internal inconsistency worth reconciling too.

Because of that, we'd like to fix this a little more thoroughly: align --help, the doc comment, and the README on one default, make the path resolution explicit rather than sentinel-based, and consider deriving the cache location from dirs::cache_dir() so it's platform-correct. We'll fold your README correction into that change, with credit to you.

One process note: we can't merge this PR as-is because the CLA check is still unsigned. We'll track the fuller fix as a separate issue/PR. Thank you again for catching this and for the clear write-up- the "I find this default a bit confusing" remark is exactly what prompted the deeper look. 🙏

Tracking the fuller fix in #223 : it covers the doc alignment (--help / doc comment / README), de-brittling the default-path resolution, and the cross-platform cache location. Your README correction will be incorporated there with credit. Thanks again @victorhooi! 🙏