Skip to main content

​
security

AI Agents Framework praisonai.security — Agent security utilities for PraisonAI. Provides:
  • Prompt injection defense (6-check pipeline)
  • Append-only audit log for tool calls
  • Protected paths guard for code tools
All features are opt-in with zero overhead when not enabled. Usage (simplest): from praisonai.security import enable_security enable_security() from praisonaiagents import Agent agent = Agent(instructions=“You are a researcher”) agent.start(“Research the latest AI news”) Usage (selective): from praisonai.security import enable_injection_defense, enable_audit_log enable_injection_defense() enable_audit_log()

​
Import

​
Functions

enable_injection_defense()

Enable prompt injection defense globally for all agents.

enable_audit_log()

Enable an append-only JSONL audit log for all agent tool calls.

enable_security()

Enable all security features in one call.