Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 16 additions & 6 deletions agentshield/eval_gym.py
Original file line number Diff line number Diff line change
Expand Up @@ -405,25 +405,35 @@ def _prior(agent_id, amount, timestamp):
"expected": "APPROVED",
"description": "Prior transaction in different session, not counted"},

{"id": 54, "category": "session_budget",
"transaction": {**_txn("t054", amount=100.00), "session_id": None},
"rules": [{"id": "sb4", "type": "session_budget", "priority": 1,
"params": {"max_session": 500}, "action": "BLOCK"}],
"prior_transactions": [
{**_txn("t054a", amount=450.00), "session_id": None},
],
"expected": "BLOCKED",
"description": "Missing session_id cannot bypass session budget evaluation"},

# ─── Cascade Cost (inspired by HeartFlow / @yun520-1) ───
{"id": 54, "category": "cascade_cost",
"transaction": {**_txn("t054", amount=50.00), "fail_probability": 0.3, "reversal_cost": 200},
{"id": 55, "category": "cascade_cost",
"transaction": {**_txn("t055", amount=50.00), "fail_probability": 0.3, "reversal_cost": 200},
"rules": [{"id": "cc1", "type": "cascade_cost", "priority": 1,
"params": {"max_cascade_cost": 100}, "action": "BLOCK"}],
"prior_transactions": [],
"expected": "BLOCKED",
"description": "Cascade cost $110 ($50 + 30% × $200) exceeds $100 limit"},

{"id": 55, "category": "cascade_cost",
"transaction": {**_txn("t055", amount=10.00), "fail_probability": 0.1, "reversal_cost": 50},
{"id": 56, "category": "cascade_cost",
"transaction": {**_txn("t056", amount=10.00), "fail_probability": 0.1, "reversal_cost": 50},
"rules": [{"id": "cc2", "type": "cascade_cost", "priority": 1,
"params": {"max_cascade_cost": 100}, "action": "BLOCK"}],
"prior_transactions": [],
"expected": "APPROVED",
"description": "Cascade cost $15 ($10 + 10% × $50) under $100 limit"},

{"id": 56, "category": "cascade_cost",
"transaction": {**_txn("t056", amount=10.00), "estimated_cascade_cost": 150},
{"id": 57, "category": "cascade_cost",
"transaction": {**_txn("t057", amount=10.00), "estimated_cascade_cost": 150},
"rules": [{"id": "cc3", "type": "cascade_cost", "priority": 1,
"params": {"max_cascade_cost": 100}, "action": "BLOCK"}],
"prior_transactions": [],
Expand Down
22 changes: 16 additions & 6 deletions tests/eval_gym.py
Original file line number Diff line number Diff line change
Expand Up @@ -449,24 +449,34 @@ def _prior(agent_id, amount, timestamp):
"expected": "BLOCKED",
"description": "Negative prior session amount ($-95) should not reduce session total, $190 > $100 → BLOCKED"},

{"id": 59, "category": "transaction_limit_block",
"transaction": _txn("t059", amount=-1000000.00),
{"id": 59, "category": "session_budget",
"transaction": {**_txn("t059", amount=100.00), "session_id": None},
"rules": [{"id": "sb4", "type": "session_budget", "priority": 1,
"params": {"max_session": 500}, "action": "BLOCK"}],
"prior_transactions": [
{**_txn("t059a", amount=450.00), "session_id": None},
],
"expected": "BLOCKED",
"description": "Missing session_id cannot bypass session budget evaluation"},

{"id": 60, "category": "transaction_limit_block",
"transaction": _txn("t060", amount=-1000000.00),
"rules": [{"id": "r1", "type": "transaction_limit", "priority": 1,
"params": {"max_amount": 500}, "action": "BLOCK"}],
"prior_transactions": [],
"expected": "BLOCKED",
"description": "Negative amount should be rejected by transaction_limit"},

{"id": 60, "category": "transaction_limit_block",
"transaction": _txn("t060", amount=0.00),
{"id": 61, "category": "transaction_limit_block",
"transaction": _txn("t061", amount=0.00),
"rules": [{"id": "r1", "type": "transaction_limit", "priority": 1,
"params": {"max_amount": 500}, "action": "BLOCK"}],
"prior_transactions": [],
"expected": "BLOCKED",
"description": "Zero amount transaction should be rejected"},

{"id": 61, "category": "cascade_cost",
"transaction": {**_txn("t061", amount=50.00), "estimated_cascade_cost": -10},
{"id": 62, "category": "cascade_cost",
"transaction": {**_txn("t062", amount=50.00), "estimated_cascade_cost": -10},
"rules": [{"id": "cc4", "type": "cascade_cost", "priority": 1,
"params": {"max_cascade_cost": 100}, "action": "BLOCK"}],
"prior_transactions": [],
Expand Down