BunkerVM – Secure runtime for AI agents using microVM sandboxes
Firecracker microVM sandbox for agents in 5 seconds, Claude Desktop ready.

Pause running agents in under a second while preserving memory and file handles.
AI engineers and agent framework developers
Modal · E2B · Replicate
Python SDK + demo: https://github.com/tinyhq/anyframe-python
AnyFrame has three key features
- define an agent once (repo, install cmd, skills, MCPs) and it bakes a cached image
- boot a session and chat with it from the web or from Python on the same channel
- plug in MCP connectors (Linear, Sentry, …) once and toggle them per-agent.
Let me know if you have any questions/comments/feature requests!
Firecracker microVM sandbox for agents in 5 seconds, Claude Desktop ready.
Hardware-isolated VM sandbox for Claude, 2-second boot, no Docker complexity.
Hypervisor isolation for AI agents beats containers when running untrusted code locally.
MicroVM sandboxes keep agents off your host, auth gateway hides API keys from model.
Runs agents in sandboxed microVMs to contain execution; promising but docs barely explain why this matters.
Custom agent framework in 2300 lines beats 400K-line bloatware; auditable and runs fully local.