JawakarAdith/ai-incident-response-platform

# 🤖 Multi-Agent AI Workflow Automation Platform Automatically analyzes failed deployment logs, identifies root causes, creates Jira tickets, and notifies Slack — in under 60 seconds. ## What it does | Manual Process | With This Tool | |---|---| | Read 500 lines of logs → 45 min | AI analysis → 10 seconds | | Identify root cause → 20 min | Auto detected | | Write Jira ticket → 10 min | Auto created | | Notify team → 5 min | Auto Slack message | | **Total: ~1 hour** | **Total: ~60 seconds** | ## Agent Pipeline User pastes logs ↓ 🧠 Planner → 🔍 RAG Search → 📋 Log Analysis ↓ 💡 Recommendation → ✅ Validation → 🔧 Tool Execution ↓ Jira Ticket + Slack Notification (auto) ## Tech Stack - **Backend** — FastAPI, Python, async - **AI Pipeline** — LangGraph, LangChain, Groq LLM (llama-3.3-70b) - **Database** — PostgreSQL, Alembic - **Vector Memory** — ChromaDB, sentence-transformers - **Integrations** — Jira API, Slack API - **Auth** — JWT - **Frontend** — Streamlit ## Setup 1. Clone the repo 2. Create virtual environment 3. Install dependencies 4. Configure `.env` 5. Run migrations 6. Start backend + frontend See SETUP.md for detailed instructions. ## Environment Variables Create `backend/.env`: DATABASE_URL=postgresql+asyncpg://user:password@localhost:5432/workflow_platform GROQ_API_KEY=your-key JIRA_BASE_URL=https://yourcompany.atlassian.net JIRA_PROJECT_KEY=SCRUM SLACK_BOT_TOKEN=your-token SLACK_DEFAULT_CHANNEL=#incidents SECRET_KEY=your-secret-key ## Features - 6-agent LangGraph pipeline - RAG memory — AI learns from past incidents - Confidence scoring — flags uncertain analyses - Auto Jira ticket creation with warning labels for low confidence - Slack notifications with severity context - Full audit trail in PostgreSQL - JWT authentication - Streamlit dashboard with history, RAG memory feed, workflow details