Awesome AI Agent Attacks maps 90 incidents

This is less a “product” than a reference asset, but it fills a real gap: agent security coverage is scattered across blog posts, advisories, and news, and this pulls it into one usable index.

• –The incident mix is broad enough to be practical, spanning supply chain compromise, framework vulns, enterprise breaches, and AI coding tool issues
• –The per-entry structure makes it more valuable than a loose link dump because it supports trend analysis and pattern matching
• –Weekly updates and PR-friendly contribution flow suggest it can stay current if maintainers keep the sourcing discipline tight
• –The main risk is curation quality: a list this wide lives or dies on source quality, de-duplication, and whether each event is actually comparable
• –For security teams, the taxonomy and stats section are likely the most useful parts because they turn anecdotes into patterns