Add Trustabl security scanning to CI

[joshua-trustabl]

I came across your repo and I like the way you demonstrate practical LLM agent patterns using modern SDKs like OpenAI Agents and CrewAI, making it easy for developers to get started with agentic workflows. I scanned the repo, and noticed some agent runtime reliability findings that might be worth reviewing.

1. Add input_guardrails for WebSearchTool usage
   In Agentic AI/02-OpenAI Agents SDK/code/search_agent.py, the agent uses WebSearchTool without input_guardrails configured. This could allow malicious or unexpected user input to trigger unsafe web searches or expose the agent to harmful content during tool execution.

2. Configure agent safety_settings for code execution
   In Agentic AI/03-CrewAI/code/coder/src/coder/crew.py, the agent enables built-in code execution via allow_code_execution=True without explicit safety_settings. This increases the risk of unintended or unsafe code interpretation during agent execution if no additional safeguards are in place.

3. Configure agent safety_settings for code execution
   In Agentic AI/03-CrewAI/code/engineering_team/src/engineering_team/crew.py, the agent enables built-in code execution via allow_code_execution=True without explicit safety_settings. This increases the risk of unintended or unsafe code interpretation during agent execution if no additional safeguards are in place.

Recommendations are based on our understanding of agent runtime reliability, some findings may be intentional. Please let us know if this was intentional or if our findings are helpful so we can improve the accuracy of the scanner.

Best,
Trustabl.ai
Open-source AI agent reliability scanner (runs locally, GitHub Action)