Sentro is a multi-agent AI system that automates Level-1 AML/KYC compliance triage for private banking, built for the Kaggle 5-Day AI Agents Intensive Capstone Project.
Manual AML/KYC alert review is slow, repetitive, and prone to human error, creating bottlenecks in private banking compliance teams.
Sentro uses a two-agent workflow to automatically fetch client data and generate a risk assessment in seconds.
- Orchestrator Agent: Receives the compliance request and routes it to the specialist agent.
- Risk Analyst Agent: Fetches client data via MCP and produces a structured risk assessment (suspicious activity, risk rating, recommended action).
- MCP Server: Exposes a
get_client_datatool that securely serves mock banking data. - Security Guardrail: Intercepts and blocks prompt-injection attempts (e.g., "ignore previous instructions") before they reach the LLM.
- Google Agent Development Kit (ADK)
- Model Context Protocol (MCP)
- Groq (Llama 3.1) via LiteLLM
- Install dependencies: pip install -r requirements.txt
- Add your Groq API key to a
.envfile: GROQ_API_KEY=your_key_here - Start the MCP server: python mcp_server.py
- In a separate terminal, run the agent: python main.py
Sentro includes a prompt-injection guardrail that blocks malicious instructions (e.g., "ignore previous instructions") before they reach the LLM, preventing manipulation of the risk assessment output.
Built by Anne for the Kaggle AI Agents: Intensive Vibe Coding Capstone Project.