Summary

create_model_from_json_schema() in src/mcpadapt/utils/modeling.py passes invalid parameters (anyOf, items, enum, properties) directly to pydantic's Field() constructor. This causes:

1. Malformed JSON schemas with empty/null values
2. Deprecation warnings in Pydantic v2
3. Will break in Pydantic v3 when **extra support is removed

Steps to Reproduce

1. Create a simple JSON schema without anyOf, items, enum, or properties
2. Observe deprecation warning:

PydanticDeprecatedSince20: Using extra keyword arguments on `Field` is deprecated 
and will be removed. Use `json_schema_extra` instead. (Extra keys: 'anyOf', 'items', 
'enum', 'properties'). Deprecated in Pydantic V2.0 to be removed in V3.0.

3. Check generated schema - contains malformed properties

Expected Behavior

Generated schema should only contain relevant JSON schema properties:

Impact

• Polluted JSON schemas with unnecessary empty/null properties
• Deprecation warnings in Pydantic v2.0+
  ++
• When using mcpadapt to wrap MCP tools for frameworks like CrewAI
• Multi-integration scenarios where schemas are cached and regenerated
• Type information gets lost in malformed schemas (anyOf: [] with no type)
• Causes validation failures when boolean values are coerced to strings
  ++
• Will break in Pydantic v3 when **extra parameter support is removed from Field()

Root Cause

Pydantic's Field() function does NOT accept:

• anyOf - JSON schema construct, not a Field parameter
• items - JSON schema construct for arrays
• enum - JSON schema construct (should use Literal/Enum types instead)
• properties - JSON schema construct for nested objects

Environment

• mcpadapt version: v0.1.13 to v0.1.20
• pydantic version: 2.x
• Python version: 3.10+

Additional Context

This issue was discovered while debugging a validation failure where boolean parameters were being coerced to strings in a multi-integration setup. The malformed schemas with anyOf: [] lacked type information, causing downstream validation issues.