BerriAI/litellm · error · ValueError

CodeInterpreterInterception: no sandbox available to run cod

Error message

CodeInterpreterInterception: no sandbox available to run code.

What it means

ValueError from CodeInterpreterInterception._run_code: after a container exists, code execution requires either a sandbox_config or the SandboxToolParams captured when the container was created. If params is None and sandbox_config is None, the handler cannot determine which provider API to call and raises. This indicates an inconsistent state — usually containers created via sandbox_config that was later cleared, or a container handle passed in from elsewhere without its params.

Source

Thrown at litellm/integrations/code_interpreter_interception/handler.py:758

            raise ValueError(
                "CodeInterpreterInterception: no sandbox available. Provide a "
                "sandbox_config or configure a sandbox tool resolvable via "
                "sandbox_tool_name."
            )
        container: Final = await litellm.acreate_sandbox(
            provider=params["sandbox_provider"],
            api_key=params.get("api_key"),
            api_base=params.get("api_base"),
        )
        return container, params

    async def _run_code(
        self, container: ContainerHandle, params: SandboxToolParams | None, code: str
    ) -> CodeExecutionResult:
        if self.sandbox_config is not None:
            return await self.sandbox_config.arun_code(container=container, code=code)
        if params is None:
            raise ValueError("CodeInterpreterInterception: no sandbox available to run code.")
        return await litellm.arun_code(
            provider=params["sandbox_provider"],
            container=container,
            code=code,
            api_key=params.get("api_key"),
        )

    async def _delete_container(self, container: ContainerHandle, params: SandboxToolParams | None) -> None:
        try:
            if self.sandbox_config is not None:
                await self.sandbox_config.adelete_sandbox(container=container)
                return
            if params is None:
                return
            await litellm.adelete_sandbox(
                provider=params["sandbox_provider"],
                container=container,
                api_key=params.get("api_key"),

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Always obtain (container, params) as a pair from _create_container and pass both to _run_code
2. Set sandbox_config on the handler so it is the single source of truth for run/delete operations
3. If you provision containers yourself, also construct/pass matching SandboxToolParams
4. Don't clear or reassign sandbox_config while cached containers are still in use

Example fix

# before
container = await handler._create_container()[0]
result = await handler._run_code(container, None, code)  # ValueError

# after
container, params = await handler._create_container()
result = await handler._run_code(container, params, code)

def can_run_code(handler, params) -> bool:
    return handler.sandbox_config is not None or params is not None

try:
    result = await handler._run_code(container, params, code)
except ValueError as e:
    if "no sandbox available to run code" in str(e):
        raise RuntimeError("Container/params mismatch — always pair them from _create_container") from e
    raise

When it happens

Trigger: Calling _run_code with a ContainerHandle obtained from outside _create_container (so no SandboxToolParams accompany it) while handler.sandbox_config is None; mutating handler.sandbox_config to None after containers were cached; subclass overriding _create_container to return (container, None).

Common situations: Custom integrations reusing the interception handler with externally provisioned containers; tests constructing fake ContainerHandles directly; refactors that move sandbox_config initialization after container creation.

Related errors

• CodeInterpreterInterception: no sandbox available. Provide a
• validate_environment must be implemented by provider
• acreate_sandbox must be implemented by provider
• arun_code must be implemented by provider
• adelete_sandbox must be implemented by provider

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/7fc82da0f5316464. Report an issue: GitHub.