Buckets:

download
raw
1.89 kB
"""
Example: Using Safety System
"""
from safety import SafetyLevel, SafetyPipeline
def example_basic_usage():
# Initialize safety pipeline
safety = SafetyPipeline(safety_level=SafetyLevel.STRICT, auto_correct=True)
# Check an output
output = """
Found vulnerability CVE-2023-1234 on target 192.168.1.1.
Port 80 is maybe open. This is probably critical.
"""
result = safety.check_output(
output=output,
context={"target": "192.168.1.1"},
schema_name="vulnerability_report",
)
print(f"Passed: {result['passed']}")
print(f"Confidence: {result['confidence'].score} ({result['confidence'].level})")
print(f"Issues: {result['issues']}")
print(f"Should retry: {result['should_retry']}")
if result["should_retry"]:
retry_prompt = safety.get_retry_prompt("Analyze target", result)
print(f"\nRetry prompt: {retry_prompt}")
return result
def example_with_agent():
"""Example integration with agent"""
from memory import MemoryManager
# Setup
memory = MemoryManager().create_conversation_memory("test")
safety = SafetyPipeline(safety_level=SafetyLevel.STANDARD)
# Agent generates output
agent_output = "Port scan complete. Port 22 is closed."
# Add context from memory
context = {
"scan_results": {"ports": {"22": {"state": "open"}}},
"memory_context": memory.get_recent(5),
}
# Safety check
result = safety.check_output(agent_output, context)
if not result["passed"]:
print("Output failed safety check!")
print(f"Issues: {result['issues']}")
print(f"Corrected: {result.get('corrected_output', agent_output)}")
return result
if __name__ == "__main__":
print("=== Basic Safety Check ===")
example_basic_usage()
print("\n=== Agent Integration ===")
example_with_agent()

Xet Storage Details

Size:
1.89 kB
·
Xet hash:
7fcb7769386c96ff917b8f1a5cce0556ff070a0ba9cc03b12284e1cab4483230

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.