Buckets:
| import sys | |
| import os | |
| import time | |
| # Ensure src is in path | |
| sys.path.append(os.path.join(os.path.dirname(__file__), "src")) | |
| from ai_sudo import Agent, sudo, PermissionDeniedError | |
| # Try importing ScopedModel, handle if pydantic not installed | |
| try: | |
| from ai_sudo.integrations import ScopedModel | |
| HAS_PYDANTIC = True | |
| except ImportError: | |
| HAS_PYDANTIC = False | |
| # --- 1. Define Safe Functions & Models --- | |
| def fetch_user_data(user_id): | |
| print(f" ๐พ [DB] Fetching data for user {user_id}...") | |
| return {"id": user_id, "name": "Alice"} | |
| # Audit Mode: Will LOG but NOT BLOCK | |
| def delete_user_audit_only(user_id): | |
| print(f" ๐ฅ [DB] (Audit Mode) DELETING user {user_id}...") | |
| return True | |
| # Callback Mode: Ask for permission | |
| def slack_approval_callback(agent, scope, func, args, kwargs): | |
| print(f" ๐ฑ [Slack] Requesting approval for {agent.name} to use {scope}...") | |
| # Simulate human approval logic | |
| if "sudo" in agent.name.lower(): # Auto-approve if name contains "sudo" | |
| return True | |
| return False | |
| def delete_user_with_approval(user_id): | |
| print(f" ๐ฅ [DB] DELETING user {user_id}...") | |
| return True | |
| if HAS_PYDANTIC: | |
| class RefundRequest(ScopedModel): | |
| _required_scope = "write:refunds" | |
| order_id: str | |
| amount: float | |
| # --- 2. Create Agents --- | |
| analyst_agent = Agent(name="AnalystBot", scopes=["read:database"]) | |
| admin_agent = Agent(name="AdminBot", scopes=["read:database", "write:database"]) | |
| sudo_agent = Agent(name="SuperSudoBot", scopes=["read:database"]) # Has specific name for callback | |
| # --- 3. Run Simulation --- | |
| def run_demo(): | |
| print("============================================") | |
| print("๐ก๏ธ AI-SUDO / AGENT SCOPE DEMO") | |
| print("============================================") | |
| # 1. Basic Blocking | |
| print("\n--- 1. Basic Permission Check (Analyst tries DELETE) ---") | |
| try: | |
| with analyst_agent.start_session(): | |
| # This function uses default on_deny="raise" | |
| # But wait, I didn't define a default delete_user in this file yet. | |
| pass | |
| # Let's define one locally or just skip. | |
| # I'll use fetch (success) and approval (fail) to show basic flow. | |
| fetch_user_data(101) | |
| except PermissionDeniedError as e: | |
| print(f"๐ BLOCKED: {e}") | |
| # 2. Audit Mode | |
| print("\n--- 2. Audit Mode (Analyst tries DELETE - Log Only) ---") | |
| with analyst_agent.start_session(): | |
| # Should succeed but print a warning log | |
| delete_user_audit_only(999) | |
| # 3. Callback / Approval Mode | |
| print("\n--- 3. Approval Mode (Analyst tries DELETE - Needs Approval) ---") | |
| try: | |
| with analyst_agent.start_session(): | |
| delete_user_with_approval(555) | |
| except PermissionDeniedError as e: | |
| print(f"๐ BLOCKED by Callback: {e}") | |
| print("\n--- 4. Approval Mode (SuperSudoBot tries DELETE - Approved) ---") | |
| try: | |
| with sudo_agent.start_session(): | |
| delete_user_with_approval(777) | |
| except PermissionDeniedError as e: | |
| print(f"๐ BLOCKED: {e}") | |
| # 4. Pydantic Integration | |
| if HAS_PYDANTIC: | |
| print("\n--- 5. Pydantic Model Scope (Analyst tries Refund) ---") | |
| try: | |
| with analyst_agent.start_session(): | |
| # Analyst lacks "write:refunds" | |
| req = RefundRequest(order_id="abc", amount=100.0) | |
| print("Refund created!") | |
| except PermissionDeniedError as e: | |
| print(f"๐ BLOCKED Model Init: {e}") | |
| print("\n--- 6. Pydantic Model Scope (Admin w/ Scope) ---") | |
| # Give Admin the scope temporarily for demo | |
| admin_agent.scopes.add("write:refunds") | |
| try: | |
| with admin_agent.start_session(): | |
| req = RefundRequest(order_id="abc", amount=100.0) | |
| print(f"โ Refund Request Validated: {req}") | |
| except PermissionDeniedError as e: | |
| print(f"๐ BLOCKED: {e}") | |
| if __name__ == "__main__": | |
| run_demo() | |
Xet Storage Details
- Size:
- 4.16 kB
- Xet hash:
- ea381a0a17e350572a5471548c6fd2e1a32308e3f5cf7d815450880203a799fe
ยท
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.