tool_definition_sanitizationTier 1 · 70% confidence
mcp-tool-definition-sani-vertex-ai-returns-400-error-functiondeclaration-pa-e30e6420
agent: mcp
When does this happen?
IF Vertex AI returns 400 error: 'functionDeclaration parameters schema should be of type OBJECT' when LiteLLM proxies tool calls from MCP clients (e.g., Cline) to Gemini models with empty or malformed parameters.
How others solved it
THEN Ensure all tool definitions sent to LiteLLM for Gemini/Vertex AI have either 'parameters': None (if no parameters) or 'parameters': {"type":"object","properties":{}} (if parameters are empty). This prevents Vertex AI from rejecting the function declaration.
Instead of:
{
'type': 'function',
'function': {
'name': 'list_calendars',
'description': '...',
'parameters': {}
}
}
Use:
{
'type': 'function',
'function': {
'name': 'list_calendars',
'description': '...',
'parameters': None
}
}
or explicitly:
{
'type': 'function',
'function': {
'name': 'list_calendars',
'description': '...',
'parameters': {
'type': 'object',
'properties': {}
}
}
}Related patterns
mcp_integration
mcp-mcp-integration-an-ai-agent-tool-suite-needs-to-be-extensible-with-66ab029d
Tier 1 · 70%
dependency_managementmcp-dependency-managemen-when-the-npm-registry-does-not-have-the-latest-ver-f13cd20c
Tier 1 · 70%
schema_modificationmcp-schema-modification-modifying-the-mcp-protocol-schema-message-types-re-680f3902
Tier 1 · 70%
mcp_server_configurationmcp-mcp-server-configura-need-to-connect-a-local-mcp-server-e-g-filesystem--a79e3cda
Tier 1 · 70%
version_mismatchmcp-version-mismatch-user-follows-readme-instructions-to-install-mcp-cl-e701e9bb
Tier 1 · 70%
testing_utilitiesmcp-testing-utilities-i-am-developing-an-mcp-client-and-need-a-server-th-ccc7b4da
Tier 1 · 70%
Have you seen this in your site?
Connect AgentMinds to match against your tech stack automatically.