求助,openclaw使用 cpa反代的gpt-5.3-codexgpt-5.4,提示错误(invalid_function_parameters)
- 内容介绍
- 文章标签
- 相关推荐
{ "error": { "message": "Invalid schema for function 'agents_list': None is not of type 'array'.", "type": "invalid_request_error", "param": "tools[10].parameters", "code": "invalid_function_parameters" } }
我的模型provider配置如下:
"models": {
"providers": {
"cpa": {
"baseUrl": "http://ai.xxxx.xyz/v1",
"apiKey": "sk-xxxx",
"api": "openai-completions",
"headers": {
"User-Agent": "curl/8.0"
},
"authHeader": true,
"models": [
{
"id": "gpt-5.3-codex",
"name": "gpt-5.3-codex",
"reasoning": false,
"input": [
"text",
"image"
],
"cost": {
"input": 0,
"output": 0,
"cacheRead": 0,
"cacheWrite": 0
},
"contextWindow": 1000000,
"maxTokens": 8192
}
]
}
}
},
相同模型参数配置下,gemini-3-flash能正常工作。在openclaw 中一个个关闭出问题的tool,才能正常对话
问了下ai说是oai检查比较严格,请教下各位佬是否有其他解决办法?
一个一摸一样的issue(close但没提供解决方案):
cli-proxy-api | [2026-02-10 15:08:42] [b77a317b] [debug] [conductor.go:2126] Use OAuth provider=codex auth_file=codex-edaezo7576@virgilian.com.json for model gpt-5.2-codex
cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:156] thinking: original config from request | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium
cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:197] thinking: processed config to apply | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': None is not of type 'array'.
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [warn ] [gin_logger.go:91] 400 | 3.427s | 172.28.0.1 | POST "/v1/messages"
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list':
openclaw调用CPA 中的codex5.2 报错。
已打开 07:13AM - 10 Feb 26 UTC 已关闭 07:43AM - 10 Feb 26 UTC nsyncxy1cli-proxy-api | [2026-02-10 15:08:42] [b77a317b] [debug] [conductor.go:2126] Us…e OAuth provider=codex auth_file=codex-edaezo7576@virgilian.com.json for model gpt-5.2-codex cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:156] thinking: original config from request | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:197] thinking: processed config to apply | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': None is not of type 'array'. cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [warn ] [gin_logger.go:91] 400 | 3.427s | 172.28.0.1 | POST "/v1/messages" cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': 用的codex普号
网友解答:--【壹】--:
弄好了吗。兄弟如果弄好了请教一下怎么使用反代的
--【贰】--:
还没有,有解决办法我在这儿更新
--【叁】--:
佬,真的非常非常感谢!
--【肆】--:
“api”: “openai-completions”, 这个错了换成"api": “openai-responses”, 因为codex的请求用的不是openai的通用协议
--【伍】--:
请教一下,CPA部署好之后,是不是两种协议都支持?
我本地试了下都成功的:
curl -X POST "http://cpaip:18317/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-" \
-d '{
"model": "gpt-5.4",
"messages": [
{
"role": "user",
"content": "hi"
}
],
"temperature": 0.7
}'
curl http://cpaip:18317/v1/responses \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-" \
-d '{
"model": "gpt-5.4",
"input": "hi"
}'
--【陆】--:
你可以换成gemini模型,让龙虾自己适配试试
--【柒】--:
不是说 free 账号的 5.4 不让用了吗?你是 free 还是 plus
--【捌】--:
我马上试试
--【玖】--:
思路不错哦,我试试
--【拾】--:
team
{ "error": { "message": "Invalid schema for function 'agents_list': None is not of type 'array'.", "type": "invalid_request_error", "param": "tools[10].parameters", "code": "invalid_function_parameters" } }
我的模型provider配置如下:
"models": {
"providers": {
"cpa": {
"baseUrl": "http://ai.xxxx.xyz/v1",
"apiKey": "sk-xxxx",
"api": "openai-completions",
"headers": {
"User-Agent": "curl/8.0"
},
"authHeader": true,
"models": [
{
"id": "gpt-5.3-codex",
"name": "gpt-5.3-codex",
"reasoning": false,
"input": [
"text",
"image"
],
"cost": {
"input": 0,
"output": 0,
"cacheRead": 0,
"cacheWrite": 0
},
"contextWindow": 1000000,
"maxTokens": 8192
}
]
}
}
},
相同模型参数配置下,gemini-3-flash能正常工作。在openclaw 中一个个关闭出问题的tool,才能正常对话
问了下ai说是oai检查比较严格,请教下各位佬是否有其他解决办法?
一个一摸一样的issue(close但没提供解决方案):
cli-proxy-api | [2026-02-10 15:08:42] [b77a317b] [debug] [conductor.go:2126] Use OAuth provider=codex auth_file=codex-edaezo7576@virgilian.com.json for model gpt-5.2-codex
cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:156] thinking: original config from request | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium
cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:197] thinking: processed config to apply | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': None is not of type 'array'.
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [warn ] [gin_logger.go:91] 400 | 3.427s | 172.28.0.1 | POST "/v1/messages"
cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list':
openclaw调用CPA 中的codex5.2 报错。
已打开 07:13AM - 10 Feb 26 UTC 已关闭 07:43AM - 10 Feb 26 UTC nsyncxy1cli-proxy-api | [2026-02-10 15:08:42] [b77a317b] [debug] [conductor.go:2126] Us…e OAuth provider=codex auth_file=codex-edaezo7576@virgilian.com.json for model gpt-5.2-codex cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:156] thinking: original config from request | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium cli-proxy-api | [2026-02-10 15:08:42] [--------] [debug] [apply.go:197] thinking: processed config to apply | provider=codex model=gpt-5.2-codex mode=level budget=0 level=medium cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': None is not of type 'array'. cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [warn ] [gin_logger.go:91] 400 | 3.427s | 172.28.0.1 | POST "/v1/messages" cli-proxy-api | [2026-02-10 15:08:45] [b77a317b] [debug] [codex_executor.go:360] request error, error status: 400, error message: Invalid schema for function 'agents_list': 用的codex普号
网友解答:--【壹】--:
弄好了吗。兄弟如果弄好了请教一下怎么使用反代的
--【贰】--:
还没有,有解决办法我在这儿更新
--【叁】--:
佬,真的非常非常感谢!
--【肆】--:
“api”: “openai-completions”, 这个错了换成"api": “openai-responses”, 因为codex的请求用的不是openai的通用协议
--【伍】--:
请教一下,CPA部署好之后,是不是两种协议都支持?
我本地试了下都成功的:
curl -X POST "http://cpaip:18317/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-" \
-d '{
"model": "gpt-5.4",
"messages": [
{
"role": "user",
"content": "hi"
}
],
"temperature": 0.7
}'
curl http://cpaip:18317/v1/responses \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-" \
-d '{
"model": "gpt-5.4",
"input": "hi"
}'
--【陆】--:
你可以换成gemini模型,让龙虾自己适配试试
--【柒】--:
不是说 free 账号的 5.4 不让用了吗?你是 free 还是 plus
--【捌】--:
我马上试试
--【玖】--:
思路不错哦,我试试
--【拾】--:
team

