[drakulavich]

Closes #147.

#153 (fixed-width rows at 60 columns, the sync grid reflowing) was planned for this PR and is left out: the issue names no target width, and nothing in the README or CHANGELOG promises 60 columns, so picking one is a product decision.

What

• --help goes through a showUsage of our own, passed to runMain. It renders with citty's renderUsage, then strips trailing whitespace from every line, including whitespace sitting before the colour codes that close a line.
• A db subcommand's screen names it in full: USAGE oura-cli db rows [OPTIONS] <COLLECTION>, banner (oura-cli db rows v0.9.1). citty names a nested command after its immediate parent only, so showUsage passes a stand-in parent called oura-cli db.
• db stats and db rows have shorter descriptions, so the longer banner still fits 80 columns. db rows keeps "twin of fetch", which the manifest test pins. The describe snapshot moves with them.

Why

Reproduced on main on an 80-column pty (script with stty cols 80):

• Every OPTIONS row was 87 characters, padded with spaces, because citty's column formatter counts colour codes as width. Each row wrapped although its text fitted.
• db today --help printed USAGE db today [OPTIONS] without oura-cli.
• The db rows banner was 77 characters as (db rows). With the full name it would have been 93, hence the shorter descriptions.
• The root USAGE line is still 88 (it lists every command), the known exception that the Prose lines overhang an 80-column terminal: hints, notes, help descriptions #130 test already allows.

After the change, the widest line of every help screen except the root is at most 80 on that pty.

How checked

• New tests in src/index.test.ts: no line of any help screen ends in spaces; the db rows USAGE line is exact. The existing Prose lines overhang an 80-column terminal: hints, notes, help descriptions #130 test, which holds every help line to 80 columns except the root USAGE line, now covers the longer banners too. Both new tests fail on main's sources.
• Mutation, reverted after one run of src/index.test.ts: const usage = (await renderUsage(cmd, named)).split('\n').map(l => l.replace(TRAILING_SPACE, '$1')).join('\n'); → const usage = await renderUsage(cmd, named);: 1 of 42 fails (the padding test).
• bun run check: typecheck, Biome, Knip clean; 713 tests pass. The first push had one failure, manifest.test.ts expecting "twin of fetch"; the second commit restores the phrase.
• Codex review: Approve, no findings (report in the first comment).

Overlap: src/index.test.ts is also touched by #173 (different blocks), and src/commands/db.ts descriptions may meet the db description change planned for #152.

🤖 Generated with Claude Code

https://claude.ai/code/session_018qh9xqviBHYmzVtmQWSKqA

[drakulavich]

Codex review (gpt-6-luna, final, ebcf1cc..409db7f):

Simplification: not needed

No Critical, Required, Optional, Nit, or FYI findings.

Ran bun test src/index.test.ts: 42 passed, 0 failed.

Verdict: Approve

[greptile-apps]

Confidence Score: 5/5

[Low impact] The PR appears safe to merge; the remaining concern is a non-blocking gap in the colored-help test.

Findings

1.  Colored padding goes untested ▶

Summary

The PR removes trailing padding from help output and gives nested db commands their full names.

• CLI help lines fit better, and database commands show their full names.

_{Reviews (2) · Last reviewed commit: "Merge remote-tracking branch 'origin/mai..." · Reviewed by Greptile}

[greptile-apps]

Colored padding goes untested

The new test pipes help output without forcing color, so it normally checks only plain output. Removing ANSI codes in the assertion does not make the child emit them. A change that breaks TRAILING_SPACE for spaces before closing color codes could therefore pass.

Add a forced-color case that clears NO_COLOR, CI, and TEST, confirms ANSI codes are present, and checks for trailing spaces after removing them.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!