What is egress control?
Answer in brief
Rules that limit which network destinations code inside a sandbox can reach.
Egress control in agent sandboxes
Egress control is one of the most important defenses against prompt-injection-driven data exfiltration: if an agent is tricked into running malicious code, an allowlist (for example only your package registry and API) keeps it from sending secrets to an attacker’s server. Look for per-sandbox allow/deny lists, the ability to disable internet entirely, and logging of outbound connections.
Cite this page
Agent Sandboxes. (2026). What is egress control?. agentsandboxes.org. Last reviewed November 8, 2026. https://agentsandboxes.org/glossary/egress-control/
Found an error or a missing sandbox? Submit a correction — every entry links to the sources it relies on.