[Bug] [Module Name] Error: 400 Invalid schema for function 'create_file': 'SandboxBase' is not valid under any of the given schemas.

[zzyyll2]

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Linux

Python version information

=3.11

What happened

我的模型配置是gpt-5-chat，已经下载datasets，并放置在正确路径，使用命令启动：uv run python packages/derisk-app/src/derisk_app/derisk_server.py --config configs/derisk-proxy-aliyun.toml，智能体里的模型已经更新到gpt-5-chat并发布，现在聊天里直接输入hello，就报错了：Error: 400 Invalid schema for function 'create_file': 'SandboxBase' is not valid under any of the given schemas.
我看到报文里有如下message:
{ "type": "function", "function": { "name": "create_file", "description": "在沙箱中创建文件并写入内容。1. 写入完成前不要并发发起后续读/写操作，待收到成功结果后再读取或修改该文件。2. 不允许再一次 tool call 中同时发起写文件 和 在附件中引用该文件的操作，并发情况下会导致附件引用失败。\\n\\n<privacy_compliance_policy>\\n遵循\"隐藏具体判罚原因\"\"可追溯但不可识别\"与\"安全指令限制\"原则：\\n1. 拦截原因隐藏：遇到风控、反洗钱、监管等拦截时，禁止解释具体规则，统一以\"系统策略限制\"等通用技术术语概括。示例：原文\"号文限制规则：xxx\"应改写为\"风控规则限制\"。\\n2. 隐私信息最小化：必须抹去姓名、手机、邮箱、地址等个人隐私信息，只保留用户ID、商户ID、单号、流水号等内部ID用于排查，确保可追踪但不可识别用户身份。\\n3. 禁止高风险操作与敏感信息泄露：不得执行涉及渗透测试、恶意攻击、漏洞利用等高风险命令；不得暴露、传播或要求用户提供任何关键的 token、密钥、证书、密码等敏感安全凭证。\\n</privacy_compliance_policy>", "parameters": { "type": "object", "properties": { "client": { "type": "SandboxBase", "description": "Client" }, "conversation_id": { "type": "string", "description": "Conversation Id" }, "description": { "type": "string", "description": "Description" }, "path": { "type": "string", "description": "Path" }, "file_text": { "type": "string", "description": "File Text" } }, "required": [ "client", "conversation_id", "description", "path", "file_text" ] } } }
AI给的原因及解决方案：

我想问你们没有遇到这个问题吗？如何规避这个？

What you expected to happen

无

How to reproduce

无

Additional context

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!