superagent-ai/brin

Scans packages, repositories, MCP servers, web pages, and GitHub contributions to detect malware, prompt injection, phishing, and supply chain attacks using a four-dimensional scoring system (identity, behavior, content, graph). Provides open-source threat intelligence as JSONL datasets organized by source type, with confidence-weighted verdicts (safe/caution/suspicious/malicious) and granular threat signal breakdowns. Integrates with agent frameworks through a CLI tool and Cursor plugin for real-time safety assessment before context consumption.