IT Support Agent 🤖

IT Support Agent 🤖

A modular, agentic AI system for automating IT support requests. This repository provides a FastAPI-based API, multiple cooperating agents (coordinator, diagnostic, automation, writer) and approval workflows for safe execution of sensitive actions.

Key capabilities:

• Automated diagnosis and remediation
• Script generation (PowerShell/Bash) and validation
• Approval and pause/resume workflow for human-in-the-loop operations
• Extensible agent architecture for adding new skills and tools

---

Project Snapshot

Below is the workflow used to orchestrate requests and approvals. The image is embedded from the repository root so it will render on GitHub and local markdown viewers.

---

Quick Start (Windows with bash.exe)

1. Create and activate a virtual environment (bash.exe compatible):

python -m venv .venv
source .venv/Scripts/activate

2. Install dependencies:

pip install -r requirements.txt

3. Run the API locally (uses uvicorn):

uvicorn app.main:app --reload --host 127.0.0.1 --port 8000

4. Open the interactive API docs at: http://127.0.0.1:8000/docs

5. Run tests:

pytest -q

---

Architecture

• app/main.py — FastAPI application entrypoint
• app/api.py — API routes and request handlers
• app/models.py — Pydantic models and type definitions
• app/agents/ — Agent implementations (coordinator, diagnostic, automation, writer)
• workflows/ — Orchestration logic and graphs

The coordinator agent receives requests, dispatches work to the diagnostic and automation agents, and routes approvals to human reviewers when necessary.

Environment & Configuration

• Python 3.9+
• Dependencies are listed in requirements.txt.
• LLM provider keys (OpenAI or other) should be supplied via environment variables. Example (bash):

export OPENAI_API_KEY="sk-..."

Adjust configuration in the code or add a .env loader as needed.

Usage Examples

• Submit a new IT request via the API route implemented in app/api.py (see the route docstrings and OpenAPI docs for concrete schemas).
• The system may pause for approval during automation steps; approved or rejected outcomes are recorded by the coordinator.

Development Notes

• Add new agents under app/agents/ and register them in the coordinator.
• Keep unit tests in test/ alongside example workflows and artifacts.

Contributing

1. Fork the repo
2. Create a branch: git checkout -b feat/your-feature
3. Make changes and add tests
4. Run tests locally: pytest -q
5. Open a pull request with a clear description