Home

SupportAgent Wiki

An AI-powered autonomous support system that diagnoses, remediates, and resolves failures in data platforms and applications. Built with LangGraph, LangChain, and Azure services.

Repository: arun-8687/SupportAgent

---

Quick links

Page	What it covers
Architecture	System design, data flow, component map
Workflow-Pipeline	LangGraph state machine, the 8-stage pipeline
Agents	The 7 agent nodes (triage, diagnose, proposal, execution, verification, resolution)
Intelligence-Layer	Deduplication, correlation, error classification, knowledge retrieval, guardrails
Tools	Platform-specific diagnostic and remediation tools
Providers	SQL Server, Databricks, Azure SQL integrations
Storage	PostgreSQL, pgvector, Azure AI Search, caching
API-Endpoints	FastAPI routes, Azure Functions triggers
Observability	LangSmith tracing, Prometheus metrics, audit logging
Configuration	Environment variables, error patterns YAML, Docker Compose
Deployment	Docker, Azure Container Apps, local development
SDK	Python client library for submitting incidents
Security	API keys, guardrails, production hardening
Testing	Unit tests, integration tests, test scripts
Runbooks	Remediation procedures for common failure patterns
Pilot-User-Stories	Requirements and user stories for the pilot phase

---

How this wiki is structured

This wiki follows the Karpathy knowledge-base pattern. Pages are interlinked. The Index page catalogs every page with a one-line summary. The Log tracks what changed and when. The wiki is the compiled, cross-referenced layer that sits between you and the raw source code.

Getting started

# Clone and install
git clone https://github.com/arun-8687/SupportAgent.git
cd SupportAgent
pip install -r requirements.txt

# Start local stack (PostgreSQL + Prometheus + Grafana)
docker compose up -d

# Run the API server
uvicorn src.api.main:app --host 0.0.0.0 --port 8000

# Send a test incident
python scripts/send_test_incident.py

See Configuration for environment variables and Deployment for production setup.