Fix MCP tool validation for nullable optional fields

[AnuradhaKaruppiah]

• Make MCP client tool calls schema‑aware by dropping None values only when the tool’s input schema does not allow nulls.
• Prevents validation failures from tools that advertise default: null on string fields (e.g., cursor, space) while preserving explicit‑null semantics when allowed.

Description

Closes

By Submitting this PR I confirm:

• I am familiar with the Contributing Guidelines.
• We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license.
  • Any contribution which contains commits that are not Signed-Off will not be accepted.
• When the PR is ready for review, new or existing tests cover these changes.
• When the PR is ready for review, the documentation is up to date with these changes.

Summary by CodeRabbit

• Bug Fixes
  • Improved input validation and compatibility with MCP servers by filtering invalid null fields during request processing.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Walkthrough

Adds internal utility functions to filter invalid None values from Pydantic model inputs before validation. The utilities detect whether type annotations permit None and remove corresponding None-valued keys from input payloads. Changes are applied in two tool response handler functions without altering public APIs.

Changes

Cohort / File(s)	Summary
Internal validation utilities packages/nvidia_nat_mcp/src/nat/plugins/mcp/client/client_impl.py	Added _annotation_allows_none() to check if a type annotation allows None values. Added _drop_invalid_none_values() to filter keys with None values when the field annotation does not permit None. Applied filtering before Pydantic model_validate() calls in mcp_per_user_tool_function() and mcp_session_tool_function() tool response handlers.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: fixing MCP tool validation by handling nullable optional fields, which aligns with the core implementation of dropping invalid None values from Pydantic payloads.
Docstring Coverage	✅ Passed	Docstring coverage is 83.33% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[AnuradhaKaruppiah]

/merge