fix: refactor bash validation

[akihikokuroda]

Pull Request

Issue

Fix #1398

Description

1. Bug Fixes (Phase 1):

• Fixed --login typo in interactive shell detection (both files)
• Improved git pattern to catch --force-with-lease variants
• Added apt/yum dangerous flag detection to pattern framework

2. Audit Trail Consolidation (Phase 2):

• Refactored validation to make check_all_patterns() canonical entry point
• All violations now recorded with audit trail (pattern, severity, category, reason)
• Nested command validation extracted into dedicated function
• Improved DestructiveGitPattern for --force-* and --directory flags

Testing

• Tests added to the respective file if code was changed
• New code has 100% coverage if code was added
• Ensure existing tests and github automation passes (a maintainer will kick off the github automation when the rest of the PR is populated)

Attribution

• AI coding assistants used

---

Adding a new component, requirement, sampling strategy, or tool?

If your PR adds or modifies one of the types below, check the matching box. A checklist of type-specific review items will be posted as a comment.

• Component
• Requirement
• Sampling Strategy
• Tool

NOTE: Please ensure you have an issue that has been acknowledged by a core contributor and routed you to open a pull request against this repository. Otherwise, please open an issue before continuing with this pull request.

[akihikokuroda]

Some tests are added suggested in #1398 (comment)

[AngeloDanducci]

One additional comment not inlined, looking at the issue it seems we want to allow single semicolon terminators.

However this sample snippet seems to fail

pattern = ShellOperatorPattern()
is_dangerous, _ = pattern.check(["find", "logs", "-exec", "cat", "{}", ";"])
assert is_dangerous is False

[akihikokuroda]

pattern = ShellOperatorPattern()
is_dangerous, _ = pattern.check(["find", "logs", "-exec", "cat", "{}", ";"])
assert is_dangerous is False

is working now.

Thanks!

[AngeloDanducci]

A few more nits on second read through.

[AngeloDanducci]

Suggested change

	dangerous_flags = {"-f", "--force", "-r", "--recursive", "--force-all"}
	dangerous_flags = {"-rf", "-f", "--force", "-r", "--recursive", "--force-all"}

[akihikokuroda]

The "-rf" and tests for it added.

[AngeloDanducci]

One coverage gap I think - there's no test that a nested-command rejection records a violation with pattern_name="NestedDangerousCommandPattern". The audit-trail tests all go through check_all_patterns, not _check_nested_dangerous_commands.

test/stdlib/tools/test_shell.py (append to TestBashValidationBugFixes)

Suggestion deletes the class line but it's just meant as a test to be added under this class.

Suggested change

	class TestBashValidationBugFixes:
	def test_nested_dangerous_command_records_violation(self) -> None:
	"""Nested-command rejections (e.g. env sudo) record an audit violation."""
	from mellea.stdlib.tools._bash_audit import get_bash_violations
	initial_count = len(get_bash_violations())
	env = StaticBashEnvironment()
	result = env.execute("env sudo whoami")
	assert result.skipped is True
	assert result.success is False
	violations = get_bash_violations()
	assert len(violations) > initial_count
	assert violations[-1].pattern == "NestedDangerousCommandPattern"
	assert violations[-1].category == "nested_command"

[akihikokuroda]

This test is added.

[AngeloDanducci]

Everything I spotted looks good, thanks for the followups Aki.