Codebase review: 7 failing tests in unit/integration suite

[neuromechanist]

Summary

After a full codebase audit, 7 of 1866 tests fail. One is tracked (#271), 6 are untracked bugs.

Failing Tests

Test	File	Likely Cause
test_no_platform_key_configured_raises_500	test_api/test_authorization.py	Tracked: #271 (.env present)
test_returns_none_when_no_key_available	test_api/test_security.py	get_llm_api_key() assertion mismatch
test_standalone_server_starts	test_integration/test_llm.py	uvicorn/standalone server startup failure
test_cli_hed_ask_standalone	test_integration/test_llm.py	CLI standalone mode with BYOK
test_cli_shows_available_assistants	test_integration/test_llm.py	CLI assistant listing
test_cli_hed_help_shows_commands	test_integration/test_llm.py	CLI help output
test_cli_unavailable_assistant_rejects	test_integration/test_llm.py	CLI unavailable assistant handling

Scope

• 2 unit test failures in auth/security (test_authorization, test_security)
• 5 integration test failures in LLM/CLI path (test_llm.py)

Acceptance Criteria

• All 1866 tests pass locally (uv run pytest -q)
• Root cause documented for each failure
• Fixes committed on develop branch