AgentIQ – Transparency & Explainability in Agentic AI

This repository contains prototypes for AgentIQ, a next-generation framework focused on ensuring transparency, explainability, and auditability in agentic AI systems. The apps are built using Streamlit and demonstrate features such as agent action simulation, logging, explanations, audit dashboards, and human-in-the-loop feedback.

Project Structure

• traceagent_app:
  The main prototype app. Provides navigation for simulating agent actions, viewing logs, generating explanations, visualizing audit dashboards, and collecting feedback. Uses Streamlit session state to manage logs and feedback.

• traceagent_dashboard.py:
  The interactive audit dashboard for AgentIQ, including performance metrics, error detection, SLA breach analysis, and export options (CSV, PDF).

• pdf_utils.py:
  Utility for exporting filtered data to PDF.

• Other supporting modules for remediation policy, environment variables, and sample agent logic.

Features

• Agent Action Simulation:
  Simulate tasks such as document classification, risk scoring, fraud detection, and workflow automation.

• Logging:
  View and manage logs of agent actions with timestamps, details, and confidence scores.

• Explanations:
  Generate natural language justifications for agent decisions using LLMs.

• Audit Dashboard:
  Visualize agent workflows, decision distributions, SLA breaches, and error metrics. Includes clickable links for remediation tickets (JIRA integration).

• Data Export:
  Export filtered agent action data as CSV or PDF.

• Human-in-the-Loop Feedback:
  Submit feedback to improve agent behavior and auditability.

Deployment on Streamlit Cloud

To deploy this app on Streamlit Cloud:

1. Push your code (including requirements.txt) to a public GitHub repository.
2. Go to Streamlit Cloud and sign in.
3. Click "New app" and connect your GitHub repo.
4. Set the main file path (e.g., AgentIQ).
5. Click "Deploy".

Streamlit Cloud will automatically install dependencies from requirements.txt and launch your app.

Getting Started

1. Install dependencies:

   pip install streamlit pandas matplotlib fpdf

2. Run the main app:

   streamlit run traceagent_app

Notes

• These are prototype/demo apps using mock data for demonstration purposes.
• You can extend the apps to connect to real agentic AI systems or datasets.
• For PDF export, ensure you have the fpdf package installed.

License

This project is provided for demonstration and prototyping purposes.