> For clean Markdown of this page, append .md to its URL. For the complete documentation index, see https://www.superagent.sh/llms.txt.


Check whether an agent that runs code stays within its permitted workspace and resources.

# AI agent execution isolation


[← Back to attack scenarios](https://www.superagent.sh/docs/products/red-teaming/attack-scenarios)

Agents that run code need clear limits on where their work can go and what it can reach. This scenario checks whether the execution environment enforces those limits.

It applies to agents with execution capabilities. A finding identifies which workspace or execution boundary failed and what became accessible.

The app lists this scenario as **Escape the agent's permitted workspace or execution boundary** (`AG-15`) in [Agent Red Teaming](https://www.superagent.sh/docs/products/red-teaming/agents).

## Standards and guidance

[OWASP Agentic Top 10 (2026), ASI05 Unexpected Code Execution](https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/)

---
Source: https://www.superagent.sh/docs/products/red-teaming/attack-scenarios/agent-execution-isolation
Index: https://www.superagent.sh/llms.txt
