Test Suite

The test suite covers all five major components: cortex, immune system, brain, sandbox, and the Sovereign Script compiler. 35+ test cases in tests/test_starter.py (324 LOC).

Running Tests

python -m pytest tests/ -v

Test Coverage

TestCortex — 7 Tests

The persistent memory engine:

Test	What It Verifies
test_remember_and_recall	Store memory → recall by keyword → correct content and tags
test_recall_by_tags	Tag-based memory retrieval works correctly
test_count	Internal memory count tracking
test_stats	Memory statistics (total, db size, types)
test_decay	Ebbinghaus decay reduces importance over iterations
test_identity_tags_immune_to_decay	Memories tagged identity survive decay

Each test creates a temporary SQLite DB that is deleted after the test.

TestImmune — 5 Tests

Pipeline health monitoring and input sanitization:

Test	What It Verifies
test_track_success	Success recording increases score
test_quarantine_after_failures	3 consecutive failures trigger quarantine
test_release_from_quarantine	Quarantined component can be released
test_reflex_rate_limiting	Rate limiter blocks after threshold
test_summary	Health summary returns correct structure

TestBrain — 3 Tests

The reasoning engine:

Test	What It Verifies
test_rule_based_analysis	Rules fire correctly for given vitals
test_confidence_threshold	Decisions below 0.6 confidence are rejected
test_stats	Brain statistics track calls and decisions

TestSandbox — 5 Tests

Pipeline execution safety:

Test	What It Verifies
test_blocks_eval	eval() is caught by static analysis
test_blocks_os	os.system() is caught
test_allows_safe_code	Legitimate code passes all checks
test_sandbox_code_injection	Sandbox header is correctly injected
test_blocks_class_escape	__class__.__bases__ escape vector blocked

TestSovereignScript — 8 Tests

The full compiler pipeline:

Test	What It Verifies
test_tokenize_pipeline	Lexer produces correct token types
test_parse_pipeline	Parser generates correct AST structure
test_codegen_pipeline	Code generator outputs valid Python
test_full_roundtrip	Source → compile → execute produces expected output
test_pipe_operator	|> desugars to nested function calls
test_if_else	Conditional compilation works
test_for_loop	For loop compilation works
test_function_def	Function definition + call roundtrip

Writing New Tests

Follow the existing pattern:

class TestNewOrgan:
    """Your new organ tests."""

    def setup_method(self):
        """Create fresh instances for each test."""
        self.organ = NewOrgan()

    def teardown_method(self):
        """Clean up any resources."""
        self.organ.close()

    def test_basic_functionality(self):
        result = self.organ.do_something()
        assert result is not None
        assert result.status == "success"

Test Fixtures

• Cortex tests use a temp file path (/tmp/test_cortex_*.db) and delete it on teardown
• Brain tests run with LLM_PROVIDER=none to test rule-based mode only
• Sandbox tests use inline code strings — no file I/O needed

---

Next: Advanced →