Greg Pstrucha

Greg Pstrucha is a Staff Engineer on Sentry's AI/ML team focused on the security of AI coding agents. He researches malicious agent 'skills' — showing how a skill can silently take over a machine or leak credentials — and builds practical tooling for teams running agents at scale, most notably dotagents (declare skills, MCP servers, hooks, and subagents in agents.toml across Claude, Cursor, Codex, and more) and the dangerous-skills security-research repo.