fix(tui): plan mode reachable (/plan, PLAN badge, --plan-mode), @bare-name completion, matching question-box borders, bubble width/tabs (#1407)

[dennisonbertram]

Closes #1407

Summary

A twelve-scenario live walk of the TUI (fake provider with scripted streaming and tool turns, plus OpenRouter DeepSeek) surfaced several first-time-user problems. This PR fixes four with red-first tests and live tmux proof: plan mode is now reachable and visible, @name Tab completion works for bare file names, the question and plan-approval boxes have matching borders, and assistant markdown bubbles never exceed the terminal width or contain tabs. Two guards are added for behaviours that were already correct (streamed-transcript integrity; no duplicate tool card on ctrl+o after an interrupt).

The streamed-markdown truncation under a real color profile turned out to be a separate, deeper streaming re-render accounting defect (not the bubble width, which is fixed and unit-tested); it is filed as its own issue and not claimed fixed here — see the scope note on #1407.

Scope and issue reconciliation

Fixed: (1) /plan command toggles enforced plan mode, a PLAN badge shows in the status bar, and harnesscli --tui --plan-mode is honored (previously runTUI dropped the flag and ctrl+o only reached plan mode before any tool call ran); (2) @cal+Tab completes bare relative names; (3) AskUserQuestion / Plan-Approval box borders sized to content; (4) assistant bubbles render markdown at the indented width, expand tabs, trim padding. Out of scope and filed separately: streaming glamour reflow truncation, ctrl+o precedence redesign.

Impact analysis reconciliation

• cmd/harnesscli/tui/model.go: executePlanCommand, PLAN badge in statusBarModelLabel, cfg.PlanMode applied in New.
• cmd/harnesscli/tui/cmd_parser.go: /plan registered (registry-completeness test updated).
• cmd/harnesscli/main.go: runTUI gains a planMode arg (test updated).
• cmd/harnesscli/tui/config.go: TUIConfig.PlanMode.
• cmd/harnesscli/tui/filecomplete.go: bare relative names complete.
• cmd/harnesscli/tui/askuser.go: box borders sized to content.
• cmd/harnesscli/tui/components/messagebubble/assistant.go: fitLine, render at contentWidth.
• Server/API untouched.

Test-first evidence

Red commands (per slice, on the branch): TestPlanCommand_TogglesAndShowsInStatusBar / TestTUIConfig_PlanModeFlag; TestFilePathCompleter_BareRelativeName (want the two calc files, got []); TestAssistantBubble_FitsWidthAndHasNoTabs (line contains a tab / is N columns wide); TestAskUser_Overlay_BordersMatch (border widths differ: top 12 vs bottom 50).
Green: go test ./cmd/harnesscli/... -race → 30 packages ok.

Verification evidence

Live tmux 120x40, built binary:
Plan approval reached via /plan (PLAN badge in the status bar):

    • Created  calc.go  w
    • Created  calc_test.go
    •  go test -v ./...  — all four subt
      === RUN   TestAdd
      --- PASS: TestAdd (0.00s)
      PASS

@bare-name Tab completion:

⏺

Question box borders match:

❯ make calc
⏺
    Done. Summary:
    • Created  calc.go  w
    • Created  calc_test.go
    •  go test -v ./...  — all four subt
      === RUN   TestAdd
      --- PASS: TestAdd (0.00s)
      PASS

Rollout and rollback

TUI-only; rebuild with scripts/install.sh, restart the TUI. Rollback: revert.

Documentation

website/docs/cli/tui.md plan-mode note; engineering-log entry.

Contract checklist

• Linked issue follows the current structured contract and this PR closes it
• Issue acceptance criteria, impact map, and scope were updated when the design changed
• All callers, consumers, sources of truth, and similar abstractions were searched
• No unrelated cleanup, hidden scope growth, duplicated wiring, or parallel abstraction was introduced
• Tests were written first and the expected red failure was observed, or this is a strictly docs-only minor PR
• Targeted checks and the repository-required full regression are green
• Security, compatibility, lifecycle, deployment, observability, documentation, and rollback were reconciled
• Real mouse/keyboard/API/operator behavior was exercised when the change is interaction- or integration-heavy

🤖 Generated with Claude Code

https://claude.ai/code/session_01WJGxhoFhA8JjkwZFcLGdS5

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.