Generated by the
tui-refactorskill. No source files were modified — this is an actionable redesign + migration plan. To implement, execute the milestones below; to verify afterwards, runtui-validatorand diff against the captured baseline.
- TUI: shotgun-cli · Framework: Bubble Tea v1.3.5 + lipgloss + bubbles (Go)
- Ambition chosen: conservative cleanup (IA preserved; both wizard + config TUI in scope)
- State of the app: already mature — centralized Nord theme, non-blocking async coordinators, context-grouped help, per-step validation. Gaps are refinement-level.
- Findings: 0 blocker · 5 major · 5 minor · 0 cosmetic
- Plan: 4 milestones, 10 items, mostly additive. F-keys retained; 2 approved behavior changes (Esc, template Enter).
| Top issue | Severity | Fix |
|---|---|---|
Dot band ..... across the bottom of every screen |
major | M1 · P010 |
| Footer key hints truncate on narrow terminals | major | M2 · P006 |
| Function-key-only nav (F5/F7/F8/F9) | major | M3 · P007 (add Tab/letter aliases) |
| Tree focus highlight obscures the label | major | M1 · P004 |
| Esc repurposed as focus toggle in inputs | major | M3 · P008 |
- Conservative cleanup — keep the linear wizard + separate config TUI.
- F-keys stay; add
Tab/Shift+Tab/r/s/galiases (additive). - Esc = cancel/leave,
Tab= textarea focus toggle. - Whole-app scope; align shared chrome across both programs.
5 major + 5 minor findings — full detail in the workspace:
~/.cache/tui-refactor/shotgun-cli/20260520T165320Z/02-gap-analysis.md
(structured: …/findings.json).
…/03-target-design/ — overview, the File-Selection before→after wireframe, and
the target keybinding map with change log.
- M1 — Design-system foundation (low risk, invisible): P004 color tiers + readable selection; P010 remove dot band.
- M2 — Shared chrome: P005 anchor header/body/status-bar layout; P006 responsive shared status bar; P003 min-size guard.
- M3 — Keybinding ergonomics: P007 F-key aliases; P008 Esc=cancel / Tab=focus / unified selection; P009 unified quit semantics.
- M4 — Per-screen polish: P011 roll the new primitives across all screens, File Selection → Review → Template → Task → Rules → Config.
Full file-level detail (what/where/how/preserves/effort/validation per item):
…/04-refactor-plan.md.
~/.cache/tui-refactor/shotgun-cli/20260520T165320Z/before/01-file-selection.png~/.cache/tui-refactor/shotgun-cli/20260520T165320Z/before/help-overlay.png~/.cache/tui-refactor/shotgun-cli/20260520T165320Z/before/02-template-selection.png~/.cache/tui-refactor/shotgun-cli/20260520T165320Z/before/wide-160x40.png
Implement M1→M4 (each as a small, buildable PR). Then run tui-validator on
the result and diff against the before/ gallery to prove the fixes landed with
no regression.
🤖 Generated with Claude Code — tui-refactor skill