By Vero SecurityBuilt and funded by Vero Security, the AI Security Engineering platform.Read the charter

Scoping an agent incident

Scoping asks which sessions, users, repositories, connectors and credentials the compromised agent could reach, not only what it did. Trace from the untrusted input that started it, then walk every tool the agent could call and every token those tools held. Public case studies of hijacked coding agents and MCP data leaks give a template for that walk.

Read in this order

  1. OfficialOWASP Top 10 for Agentic Applications for 2026Start here: the agent risks that define what to check during scoping.
  2. ArticleSecurity incident disclosure, July 2026 (Hugging Face)A victim's own account of what the intruding agents reached and how they scoped it.
  3. ArticleDisrupting an AI-orchestrated cyber espionage campaign (Anthropic)An incident where agents ran most of an attack, with scoping lessons.

Save your progress

We email you a sign-in link. No password.

By signing in you agree to the terms. The privacy policy explains how we use your details.

Join the AI security engineering community

Free. Roadmap updates, chapters near you, and first word when credentials open.

We’ll email you community updates. Unsubscribe anytime. By sending this you agree to the terms. The privacy policy explains how we use your details.