DeepTeam is a framework to red team LLMs and AI agents.
import pytest
from deepteam.vulnerabilities import RBAC
from deepteam.vulnerabilities.rbac import RBACType
from deepteam.test_case import RTTestCase
class TestRBAC:
def test_rbac_all_types(self):
types = [
"role_bypass",
"privilege_escalation",
"unauthorized_role_assumption",
]
rbac = RBAC(types=types)
assert sorted(type.value for type in rbac.types) == sorted(types)
def test_rbac_all_types_default(self):
rbac = RBAC()
assert sorted(type.value for type in rbac.types) == sorted(
type.value for type in RBACType
)
def test_rbac_role_bypass(self):
types = ["role_bypass"]
rbac = RBAC(types=types)
assert sorted(type.value for type in rbac.types) == sorted(types)
def test_rbac_privilege_escalation(self):
types = ["privilege_escalation"]
rbac = RBAC(types=types)
assert sorted(type.value for type in rbac.types) == sorted(types)
def test_rbac_unauthorized_role_assumption(self):
types = ["unauthorized_role_assumption"]
rbac = RBAC(types=types)
assert sorted(type.value for type in rbac.types) == sorted(types)
def test_rbac_all_types_invalid(self):
types = [
"role_bypass",
"privilege_escalation",
"unauthorized_role_assumption",
"invalid",
]
with pytest.raises(ValueError):
RBAC(types=types)
def test_simulate_attacks_returns_expected_cases(self):
rbac = RBAC(types=["role_bypass"])
test_cases = rbac.simulate_attacks(attacks_per_vulnerability_type=2)
assert len(test_cases) == 2
assert all(isinstance(tc, RTTestCase) for tc in test_cases)
assert all(tc.vulnerability == "RBAC" for tc in test_cases)
assert all(
tc.vulnerability_type == RBACType.ROLE_BYPASS for tc in test_cases
)
def test_assess_returns_results(self):
rbac = RBAC(types=["role_bypass"], async_mode=False)
def dummy_model_callback(prompt):
# Provide a simple pass-through or minimal callback if required by your real env
return prompt
results = rbac.assess(
model_callback=dummy_model_callback,
)
assert rbac.is_vulnerable() is not None
assert rbac.simulated_attacks is not None and isinstance(
rbac.simulated_attacks, dict
)
assert rbac.res is not None and isinstance(rbac.res, dict)
assert RBACType.ROLE_BYPASS in results
assert len(results[RBACType.ROLE_BYPASS]) == 1
test_case = results[RBACType.ROLE_BYPASS][0]
# The score and reason will depend on the real metric evaluation
assert hasattr(test_case, "score")
assert hasattr(test_case, "reason")
def test_get_metric_returns_RBAC_metric(self):
from deepteam.metrics import RBACMetric
rbac = RBAC(
async_mode=True, verbose_mode=True, evaluation_model="gpt-4o"
)
metric = rbac._get_metric(RBACType.ROLE_BYPASS)
assert isinstance(metric, RBACMetric)
assert metric.async_mode is True
assert metric.verbose_mode is True
@pytest.mark.asyncio
async def test_a_assess_returns_async_results(self):
rbac = RBAC(types=["role_bypass"], async_mode=True)
async def dummy_model_callback(prompt):
return prompt
results = await rbac.a_assess(
model_callback=dummy_model_callback,
)
assert rbac.is_vulnerable() is not None
assert rbac.simulated_attacks is not None and isinstance(
rbac.simulated_attacks, dict
)
assert rbac.res is not None and isinstance(rbac.res, dict)
assert RBACType.ROLE_BYPASS in results
assert len(results[RBACType.ROLE_BYPASS]) == 1
test_case = results[RBACType.ROLE_BYPASS][0]
assert hasattr(test_case, "score")
assert hasattr(test_case, "reason")