P2-02: Evaluate typed CLI inputs after canonical schemas converge

[Dan-StrategicAutomation]

Backlog metadata

• Priority: P2
• Backlog order: P2-02
• Type: Research / refactor
• Area: CLI ergonomics
• Status: Open
• Depends on: P0-06: Consolidate shared guard results, CLI setup and review helpers #132, P0-08: Introduce a canonical Pydantic Scope model and shared scope loader #134, P0-09: Model persisted receipts, pending batches and skill state with Pydantic #135
• Related work: Related to Make the guard's own command diagnostics reachable from the raw terminal #82

---

Summary

From the 20 September 2026 Violin dev duplication/library review, baseline 5ca0d9d6767b91fe809b2e4996c85ac65ba9e962. Priority: P2. Source locations and measurements describe the review baseline.

Scope

Compare retaining argparse with a Typer/Click approach backed by canonical application request models. Map scripts/violin_guard.py arguments to those models and measure duplicate validation reduction, help/completion benefit, compatibility and dependency cost.

Acceptance criteria

• Document an explicit adopt/defer decision using a representative prototype.
• Preserve flags, help expectations, exit codes and administration-only CLI security boundary.
• If adopting Click/Typer, declare the direct dependency and provide incremental migration criteria; do not rely on transitive Click.
• A documented defer decision can close this evaluation; a broad CLI rewrite is not required.

Dependencies / related work

Sequence after #132, #134, #135.

#82 diagnostic access and closed #86 virtualenv handling remain separate and must be preserved.

Verification

Working branch: codex/review-cli-evaluation, based on dev. Future PRs target dev; follow AGENTS.md and preserve fail-closed policy. Rebase dependent branches onto updated dev before implementation.

Implementation must pass uv run pytest, uv run ruff check ., uv run ruff format --check . and uv run python scripts/violin_guard.py check-release, plus the focused checks above. No implementation or test rerun is claimed by this tracking issue.