Skip to content

petter2025us/arf-hackathon-agent

Repository files navigation

ARF Access Control Agent

One‑liner: AI agent that makes provably governed access decisions using Bayesian risk scoring.

Demo Video

Watch the demo

“The ARF legacy API endpoint is currently unavailable. The agent is designed to call it via OpenAPI, and the mock responses used in the demo exactly match the real API’s output schema. In production, you would replace the mock with the live ARF API.”

Problem

Enterprise agents often act without governance – they are non‑deterministic, unauditable, and unsafe for high‑stakes workflows like access control.

Solution

This agent uses the Agentic Reliability Framework (ARF) to evaluate every access request with:

  • Bayesian risk scoring
  • Expected loss minimisation
  • Clear approve / deny / escalate decisions
  • Full audit trail

Flow

Input (role, resource) → ARF API → Decision (approve/deny/escalate) + risk_score + audit_log

Demo Scenarios

Role Expected Decision Risk Level
admin approve low (<0.3)
intern deny high (>0.7)
contractor escalate medium (0.3–0.7)

Stack

  • ARF Legacy API (Bayesian risk engine)
  • JSON interface (no UI required)
  • Optional: FastAPI wrapper

Repository Files

  • openapi.yaml – ARF API specification for watsonx Orchestrate
  • agent-instructions.md – deterministic instructions for the agent
  • test_cases.json – structured test cases
  • evaluate.py – simple Python script to call the API (optional)

How to Use

  1. Import openapi.yaml into watsonx Orchestrate as a tool.
  2. Create an agent with the instructions from agent-instructions.md.
  3. Send a JSON payload: {"incident": {"type": "access_request", "user_role": "admin"}}
  4. Receive decision, risk_score, and audit log.

Governance Guarantee

Every decision is auditable, risk‑quantified, and derived from ARF’s Bayesian inference – no black boxes.

About

Enterprise access control agent using ARF Bayesian risk scoring for IBM watsonx Orchestrate hackathon.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages