[BUG] litellm.BadRequestError: OpenAIException - Invalid schema for function 'series': In context=('properties', 'matches'), array schema missing items.

[zebesh]

📋 Prerequisites

• I have searched the existing issues to avoid creating a duplicate
• By submitting this issue, you agree to follow our Code of Conduct
• I am using the latest version of the software
• I have tried to clear cache/cookies or used incognito mode (if ui-related)
• I can consistently reproduce this issue

🎯 Affected Service(s)

App Service

🚦 Impact/Severity

No impact (Default)

🐛 Bug Description

i install this mcp server https://github.com/tjhop/prometheus-mcp-server/tree/main like this:
`apiVersion: apps/v1
kind: Deployment
metadata:
name: prometheus-mcp-server
namespace: kagent
labels:
app: prometheus-mcp-server
spec:
replicas: 1
selector:
matchLabels:
app: prometheus-mcp-server
template:
metadata:
labels:
app: prometheus-mcp-server
spec:
containers:
- name: prometheus-mcp-server
image: ghcr.io/tjhop/prometheus-mcp-server:v0.3.0
imagePullPolicy: IfNotPresent
args:
[
"--log.file",
"/dev/stdout",
"--log.level",
"debug",
"--mcp.transport",
"http",
"--prometheus.url",
"http://kube-prometheus-stack-prometheus.kube-prometheus-stack.svc.cluster.local:9090",
]
ports:
- containerPort: 8080
resources:
limits:
cpu: "500m"
memory: "256Mi"
requests:
cpu: "10m"
memory: "64Mi"

apiVersion: v1
kind: Service
metadata:
name: prometheus-mcp-server
spec:
selector:
app: prometheus-mcp-server
ports:
- port: 80
targetPort: 8080
type: ClusterIP
`

after add toolserver
apiVersion: kagent.dev/v1alpha1 kind: ToolServer metadata: name: prometheus spec: config: streamableHttp: sseReadTimeout: 5m0s timeout: 5s url: http://prometheus-mcp-server:80/mcp type: streamableHttp description: 'Prometheus MCP Server for metrics and monitoring data'

and after create agent
`apiVersion: kagent.dev/v1alpha1
kind: Agent
metadata:
name: prometheus
spec:
description: агент для работы с прометеем
memory:
- kagent/kagent-memory
modelConfig: default-model-config
systemMessage: |-
You're a helpful agent, made by the kagent team.

# Instructions
    - If user question is unclear, ask for clarification before running any tools
    - Always be helpful and friendly
    - If you don't know how to answer the question DO NOT make things up, tell the user "Sorry, I don't know how to answer that" and ask them to clarify the question further
    - If you are unable to help, or something goes wrong, refer the user to https://kagent.dev for more information or support.

# Response format:
    - ALWAYS format your response as Markdown
    - Your response will include a summary of actions you took and an explanation of the result
    - If you created any artifacts such as files or resources, you will include those in your response as well

tools:
- mcpServer:
toolNames:
- exemplar_query
- list_alerts
- list_targets
- metric_metadata
- query
- range_query
- series
- targets_metadata
- alertmanagers
toolServer: kagent/prometheus
type: McpServer
`

and then i ask anything i get error
litellm.BadRequestError: OpenAIException - Invalid schema for function 'series': In context=('properties', 'matches'), array schema missing items.

🔄 Steps To Reproduce

1. install mcp
2. create tools server
3. create agent
4. ask somthing

🤔 Expected Behavior

No response

📱 Actual Behavior

No response

💻 Environment

No response

🔧 CLI Bug Report

No response

🔍 Additional Context

No response

📋 Logs

litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'message': "Invalid schema for function 'series': In context=('properties', 'matches'), array schema missing items.", 'type': 'invalid_request_error', 'param': 'tools[7].function.parameters', 'code': 'invalid_function_parameters'}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/app/python/packages/kagent-adk/src/kagent_adk/_agent_executor.py", line 124, in execute
    await self._handle_request(context, event_queue, runner)
  File "/app/python/packages/kagent-adk/src/kagent_adk/_agent_executor.py", line 188, in _handle_request
    async for adk_event in runner.run_async(**run_args):
    ...<4 lines>...
            await event_queue.enqueue_event(a2a_event)
  File "/app/python/.venv/lib/python3.13/site-packages/google/adk/runners.py", line 233, in run_async
    async for event in self._exec_with_plugin(
    ...<2 lines>...
      yield event
  File "/app/python/.venv/lib/python3.13/site-packages/google/adk/runners.py", line 274, in _exec_with_plugin
    async for event in execute_fn(invocation_context):
    ...<6 lines>...
      yield (modified_event if modified_event else event)
  File "/app/python/.venv/lib/python3.13/site-packages/google/adk/runners.py", line 230, in execute

📷 Screenshots

No response

🙋 Are you willing to contribute?

• I am willing to submit a PR to fix this issue

[EItanya]

Very strange, have you tried these tools with other frameworks, or maybe other models? It seems like openai thinks the tool schema is invalid

[zebesh]

I connected it to a local cursor and it really only works with cloude