fix: validate backtest run_dir CLI input

[BigNounce90]

Summary

The standalone backtest validation entrypoint accepted sys.argv[1] without checking that it was a usable directory path, which could surface low-level path and file errors to operators. This change adds a small argument parser that rejects empty, malformed, missing, or non-directory run_dir values with clear CLI messages before running validation.

Why it matters

Without this guard, a malformed path such as an embedded-NUL argument or an accidental empty string can produce a raw exception instead of a clear operator-facing error. That makes the failure harder to diagnose and turns a narrow input mistake into an avoidable crash.

Testing

Added focused regression tests for blank, malformed, missing, and valid run_dir arguments. The existing successful path is preserved by asserting that a real directory still parses to the same Path value passed into main before.

[BigNounce90]

PR #60 direction good. Scope narrow. Failure mode concrete.

Fix before sending:

• Move new CLI tests under agent/tests/. CI runs pytest from repo root and existing suite lives there.
• Use from backtest import validation or from backtest.validation import _parse_run_dir. Root-level from agent.backtest import validation breaks repo conventions.
• Add test for missing argument. Current code changes CLI behavior here but does not lock it down.
• Add test for non-directory path. PR body claims this case, current tests do not prove it.

Suggested patch file:

• artifacts/vibe-trading-pr60-improvement.patch

[BigNounce90]

Updated this PR to address the earlier layout/test concerns:

• moved the new CLI regression test under agent/tests/
• aligned imports with the existing test suite (from backtest import validation)
• added coverage for missing-argument and non-directory path cases
• hardened embedded-NUL handling so malformed input fails with a clear operator-facing message

Local verification run:

• python -m py_compile agent\\backtest\\validation.py agent\\tests\\test_validation_cli.py
• PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 pytest agent\\tests\\test_validation_cli.py -q

[warren618]

Thanks for the update 🙂 I re-checked the latest PR head.

The validation CLI now rejects missing, blank, malformed, missing-path, and non-directory run_dir inputs before calling the validation runner. The new regression tests are in the project test tree and match the existing import style.

Verified locally with:

• pytest agent/tests/test_validation_cli.py -q
• pytest agent/tests/test_validation.py agent/tests/test_validation_cli.py -q
• python -m py_compile agent/backtest/validation.py agent/tests/test_validation_cli.py
• ruff check agent/backtest/validation.py agent/tests/test_validation_cli.py
• CLI smoke checks for missing args and non-directory paths

No blockers from my side. Approved, thanks for the fix 🙌