[HamzaKhanBUIC]

Summary

Fixes #70.

Preserves $defs and �dditionalProperties in create_tool_schema, preventing unresolvable $ref dangling pointers when tools use Enum or nested BaseModel parameters.

Problem

When constructing tool parameters, create_tool_schema extracted only properties and
equired from Pydantic's model_json_schema(). Nested models and enums are emitted by Pydantic as $ref: #//..., but the $defs block was dropped. This left emitted schemas with unresolvable references, causing LLM providers (e.g. OpenAI, Gemini) to reject or fail to interpret tool definitions.

Solution

• In �gentkit/tools/schema.py:
  • Retain
    aw_json_schema[$defs] in parameters[$defs] when present.
  • Retain
    aw_json_schema[additionalProperties] in parameters[additionalProperties] when present.
  • Built-in tools without $defs continue producing identical schemas without extra keys.
• In ests/unit/test_tool_schema.py:
  • Added test validating tools with Enum and nested BaseModel parameters.
  • Added recursive JSON Pointer resolution test asserting that every $ref resolves cleanly inside the emitted schema.
  • Added regression test asserting that built-in tools (calculator,
    ead_file, http_fetch, sql_readonly, web_search) produce expected schemas with no unnecessary keys.

Verification

• uv run pytest tests/unit/test_tool_schema.py -> 6/6 passed
• uv run ruff check . -> All checks passed
• uv run ruff format --check . -> 110 files already formatted
• uv run mypy agentkit tests/unit/test_tool_schema.py -> Success: no issues found in 43 source files
• Structural AST lint & Semgrep security check -> Clean

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.