psagun/sentinelwatch

# SentinelWatch **AI-Powered Security & Compliance Monitoring Platform** SentinelWatch continuously monitors the open web for threats, regulatory changes, third-party risks, and brand exposure — delivering structured, actionable intelligence directly to security and compliance teams. Built for the **Web Data UNLOCKED Hackathon** — Track 3: Security & Compliance. ## What It Does - **Threat Intelligence Pipeline** — Monitors open web sources (CVE feeds, security blogs, dark web forums, paste sites) for org-specific risk indicators - **Regulatory Compliance Monitor** — Tracks GDPR, CCPA, SEC, and other regulatory changes, parsing updates into structured alerts - **Third-Party Risk Assessment** — Continuously assesses supplier and vendor exposure across the web - **Brand & Data Exposure Detection** — Detects leaked credentials, exposed data, and reputational threats early - **Autonomous Investigation Agents** — AI agents that investigate threat indicators and return structured risk assessments without human intervention ## Architecture Overview ┌─────────────────────────────────────────────────────────────┐ │ Data Collection Layer │ │ Bright Data Web Unlocker | SERP API | Scraping Browser │ └──────────────────────────┬──────────────────────────────────┘ │ ┌──────────────────────────▼──────────────────────────────────┐ │ Analysis & Intelligence Layer │ │ Threat Analyzer | Compliance Analyzer | Risk Scorer │ │ AI/ML API-powered investigation agents │ └──────────────────────────┬──────────────────────────────────┘ │ ┌──────────────────────────▼──────────────────────────────────┐ │ Memory & Storage Layer │ │ Cognee (persistent agent memory) | PostgreSQL (findings) │ └──────────────────────────┬──────────────────────────────────┘ │ ┌──────────────────────────▼──────────────────────────────────┐ │ Alerting & Reporting Layer │ │ Alert Manager | Slack/Email/PagerDuty | Webhook exports │ └──────────────────────────┬──────────────────────────────────┘ │ ┌──────────────────────────▼──────────────────────────────────┐ │ API & Integration Layer │ │ REST API | Webhook receiver | Dashboard feed │ └─────────────────────────────────────────────────────────────┘ ## Bright Data Tools Used | Tool | Purpose | |---|---| | **Web Unlocker** | Bypass bot detection on security forums, paste sites, regulatory pages | | **SERP API** | Real-time search across Google, Bing for threat indicators | | **MCP Server** | Connect AI agents directly to live web context | | **Scraping Browser** | Full browser automation for JS-heavy compliance and risk pages | | **Web Scraper API** | Structured data from pre-built site scrapers | ## Tech Stack - **Python 3.11+** — Core application - **FastAPI** — REST API - **Bright Data SDK** — Web data collection - **AI/ML API** — AI reasoning and analysis - **Cognee** — Agent memory and context - **PostgreSQL** — Findings and alert storage - **Docker** — Containerized deployment ## Getting Started See [SETUP.md](SETUP.md) for full setup instructions. Quick start: cp .env.example .env # Fill in your API keys docker-compose up -d ## Project Structure sentinelwatch/ ├── src/ # Core application source │ ├── main.py # Entry point │ ├── config.py # Configuration management │ ├── data_collection/ # Bright Data integration layer │ ├── intelligence/ # AI analysis and risk scoring │ ├── memory/ # Cognee agent memory │ ├── alerting/ # Alert management and notifications │ └── api/ # REST API ├── agents/ # Autonomous AI agent definitions ├── config/ # YAML configuration files ├── tests/ # Test suite ├── docs/ # Documentation └── docker-compose.yml # Deployment ## Partner Technologies - **AI/ML API** — Powers the intelligence layer (analysis, reasoning, extraction) - **Cognee** — Provides persistent agent memory across investigation workflows - **TriggerWare.ai** — Event-driven workflows triggered by web data changes - **Speechmatics** — Optional voice-enabled alerting and reporting ## Demo Watch the product demo or flip through the presentation: - **📹 Demo Script** — [`docs/demo-script.md`](docs/demo-script.md) — A ~3-minute narrated walkthrough covering entity monitoring, findings, alerts, compliance, and AI reports - **📊 Presentation** — Open the slide deck at [`/presentation/`](https://frontend-lilac-eta-55.vercel.app/presentation/index.html) (10 slides with screenshots) - **🌐 Live App** — [frontend-lilac-eta-55.vercel.app](https://frontend-lilac-eta-55.vercel.app) - **📄 PDF** — Download the presentation as PDF from the slide deck (Ctrl+P → Save as PDF, Landscape, A3) ### Demo Flow Entities → Add Website → Configure Scans → Scan Runs → Dashboard → Findings → Alerts → Compliance → AI Reports ## License Hackathon project — see hackathon terms.