sandflare-llamaindex 0.1.5

LlamaIndex tool for Sandflare Firecracker microVM sandboxes — run agent-generated code safely

sandflare-llamaindex

LlamaIndex code interpreter tool for Sandflare — execute agent-generated code in isolated Firecracker microVM sandboxes.

Installation

pip install sandflare-llamaindex

Usage

from llama_index.core.agent import ReActAgent
from llama_index.llms.openai import OpenAI
from sandflare_llamaindex import SandflareCodeInterpreterTool

tool = SandflareCodeInterpreterTool(api_key="sf-...")

agent = ReActAgent.from_tools(
    [tool],
    llm=OpenAI(model="gpt-4o"),
    verbose=True,
)

response = agent.chat("Compute the first 20 Fibonacci numbers and plot them")
print(response)

How it works

Each tool call creates a fresh Sandflare microVM sandbox, runs the code, and returns stdout/stderr to the LlamaIndex agent as a tool response.

Links

• Sandflare docs
• sandflare Python SDK