Managed Agents Configure Runtime Environment-Alibaba Cloud Model Studio(Model Studio)-阿里云帮助中心

A runtime environment defines the execution sandbox for tool calls. Create and manage environments on the Environments page in the console. An environment can be reused across multiple sessions.

Types

Type	Description
Cloud	A sandbox container managed by Model Studio, ready to use out of the box. Once created, it can be bound to sessions

Configuration Fields

On the Environments page, click Add Environment and fill in the following fields:

Field	Required	Mutable	Description
Name	Yes	Yes	Unique identifier within the workspace
Description	No	Yes	Description of the environment purpose
Hosting type	No	No	`cloud`, managed by Model Studio. Cannot be changed after creation
Packages	No	Yes	Declared by package manager (`apt` / `pip` / `npm`), automatically installed when the environment is created
Network policy	No	Yes	`unrestricted` allows all outbound access (API-only setting, not displayed in the console)
Scope	No	Yes	`organization` (default), available to all workspace members (API-only setting, not displayed in the console)
Metadata	No	Yes	Custom key-value pairs (corresponds to the `metadata` field in the API), does not affect runtime behavior

When creating an environment via the API, specify the sandbox type and packages. For complete parameters and response fields, see Create an environment.

Bash

curl -X POST "https://{workspace_id}.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio/environments" \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "data-sandbox",
    "description": "Data analysis sandbox",
    "config": {
      "type": "cloud",
      "packages": {
        "apt": ["ffmpeg"],
        "pip": ["pandas", "numpy", "matplotlib"]
      },
      "networking": {"type": "unrestricted"}
    }
  }'

Python

env = client.environments.create(
    name="data-sandbox",
    config={
        "type": "cloud",
        "networking": {"type": "unrestricted"},
        "packages": {
            "apt": ["ffmpeg"],
            "pip": ["pandas", "numpy", "matplotlib"],
        },
    },
    description="Data analysis sandbox",
)

Java

Environment env = client.environments().create(EnvironmentCreateParam.builder()
    .name("data-sandbox")
    .description("Data analysis sandbox")
    .build());

Supported Package Managers

Packages are declared using the following package managers:

apt: System packages, for example ffmpeg
pip: Python packages, for example pandas, numpy
npm: Node.js packages

Archive and Delete

Environments support two operations: archive and delete.

Archive: The environment is retained but hidden from the list by default. Sessions already bound to it remain functional.
Delete: Hard delete. The environment configuration is permanently removed and cannot be recovered. Use archive instead if you want to preserve the configuration.

To archive an environment via the API, see Archive Environment.

Bash

curl -X POST "https://{workspace_id}.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio/environments/env_xxx/archive" \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY"

Python

client.environments.archive("env_xxx")

Java

client.environments().archive("env_xxx");

To delete an environment via the API, see Delete Environment.

Bash

curl -X DELETE "https://{workspace_id}.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio/environments/env_xxx" \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY"

Python

client.environments.delete("env_xxx")

Java

client.environments().delete("env_xxx");