SeaSnoke

Docs

Core Concepts

MicroVM Isolation

Secure, lightweight execution environments for every agent.

Security Architecture

Every agent runs inside a Firecracker microVM with no network access to your production systems. Each VM is provisioned with only the dependencies needed for the specific task, and destroyed immediately after the agent completes.

Resource Limits

Agents are constrained by CPU, memory, disk, and network quotas. Default limits:

cpu: 2 cores
memory: 4GB
disk: 10GB
network: outbound only, no LAN access