[terminal-stylist] Terminal Stylist: Console Output Analysis (Lipgloss/Huh)

[github-actions[bot]]

Terminal Stylist Report — Console Output Consistency Analysis

Scanned all non-test .go files under pkg/. Overall the codebase is highly consistent: the pkg/console package (Lipgloss-based, 551 LOC) is used broadly, with 1,828 console.Format* calls across 20+ CLI files, and 20 files leverage console.RenderTable/RenderStruct. pkg/styles centralizes a Dracula-inspired adaptive color palette, and pkg/tty centralizes TTY detection (no scattered term.IsTerminal/isatty calls found outside it).

Key metrics

• fmt.Print* in non-test/non-testdata code: only 2 occurrences, both intentional pass-throughs of pre-rendered console-formatted strings (status_command.go:288, view_command.go:168).
• console.* usage: 248 files.
• lipgloss.* direct usage: 6 files (mostly inside pkg/styles/pkg/console, as intended — good encapsulation).
• huh (interactive forms) usage: 0 files import charmbracelet/huh for actual forms, despite a fully-built pkg/styles/huh_theme.go theme existing unused.
• Manual %-*s padding for column alignment: 3 call sites, none using lipgloss/table or console.RenderTable.

Detailed findings

1. Console package usage — good

• pkg/console/console.go provides FormatError, FormatSuccessMessage, FormatInfoMessage, FormatWarningMessage, FormatErrorChain, RenderTable, etc., each with TTY-aware stdout/stderr variants (e.g. FormatInfoMessage vs FormatInfoMessageStderr). This dual-variant pattern is applied consistently.
• The two remaining fmt.Print/fmt.Fprintln calls in status_command.go and view_command.go already wrap console.RenderStruct(...)/pre-rendered console output, so they are not violations — just the terminal Print of an already-styled string.

2. Lipgloss usage — well encapsulated but a few manual-formatting gaps

• pkg/styles/theme.go defines lipgloss.AdaptiveColor{Light:..., Dark:...} values used everywhere via lipgloss.LightDark(isDark) (see huh_theme.go), which is the correct adaptive-color pattern.
• Three call sites hand-roll column alignment with fmt.Sprintf("%-*s", width, ...) instead of lipgloss.Style.Width()/lipgloss/table:
  • pkg/cli/engine_secrets.go:788 (secret name padding)
  • pkg/cli/compile_schedule_calendar.go:269 (day-label padding)
  • pkg/console/render.go:174 (struct field name padding — inside console pkg itself, lower priority since it's the shared formatter)
    These aren't bugs, but migrating to lipgloss.Style{}.Width(n) would remove manual rune-width arithmetic and handle wide/unicode names more robustly.
• compile_schedule_calendar.go builds a calendar heatmap manually with strings.Builder + per-cell fmt.Fprintf("%02d "); this is a reasonable custom visualization, but lipgloss/table could reduce manual width bookkeeping if revisited.

3. Huh forms — theme exists, unused

• pkg/styles/huh_theme.go defines a complete HuhTheme huh.ThemeFunc mapping the CLI's color palette to focused/blurred field styles, select indicators, buttons, and text inputs — clearly intended for future interactive prompts.
• No command currently imports charmbracelet/huh to build an actual form. Any interactive Y/N or manual bufio.Scanner prompt in the CLI should be checked against this theme when confirmation/selection UX is added, so the investment in HuhTheme isn't stranded.

4. TTY detection — centralized correctly

• All raw term.IsTerminal calls live in pkg/tty/tty.go (IsStdoutTTY, IsStderrTTY); pkg/console builds its isTTY/isStderrTTY helpers on top of it consistently, and format functions take an injectable ttyCheck func() bool for testability — a solid pattern for degrade-to-plain-text behavior in pipes/redirects.

Recommendations

1. Low priority / cosmetic: Replace the 3 manual %-*s padding call sites (engine_secrets.go:788, compile_schedule_calendar.go:269, console/render.go:174) with lipgloss.Style{}.Width(n) for consistent unicode-width handling.
2. Opportunity: Wire up the existing unused styles.HuhTheme the next time an interactive prompt/confirmation is added to the CLI (e.g., destructive-action confirmations), rather than hand-rolling a scanner-based prompt.
3. No action needed for the 2 remaining fmt.Print* sites — they're intentional final emission of already-console-formatted output.
4. Overall, no widespread inconsistency was found; the console/styles/tty package boundaries are well respected throughout pkg/.

Generated by 💄 Terminal Stylist · auto · 21.5 AIC · ⌖ 5.12 AIC · ⊞ 8.1K · ◷

• expires on Aug 8, 2026, 1:02 AM UTC-08:00