Using BoxLite as a sandbox for AI agent workflows.
| File | Description |
|---|---|
drive_box_with_llm.py |
Let an LLM drive a SimpleBox via tool-use loop (OpenAI) |
drive_box_with_minimax.py |
Let MiniMax M3 drive a SimpleBox via tool-use loop |
use_skillbox.py |
Run Claude Code CLI with skills inside a box |
chat_with_claude.py |
Multi-turn Claude conversation via stdin JSON protocol |
order_starbucks.py |
End-to-end agent: order Starbucks via browser automation |
run_openclaw.py |
Run OpenClaw (ClawdBot) AI gateway in a container |
Most examples require CLAUDE_CODE_OAUTH_TOKEN to be set.
Recommended first example: drive_box_with_llm.py
BoxLite works with any LLM provider to create secure sandboxed environments for AI agents.
The examples in this directory include ready-to-run integrations for
OpenAI and MiniMax (MiniMax-M3, MiniMax-M2.7, MiniMax-M2.7-highspeed).