fix(core): handle Pydantic-incompatible types in injected tool args

[Alon Naor (AlonNaor22)]

Summary

• Fixes tool creation crash when a parameter is annotated with InjectedToolArg and typed as a Pydantic-incompatible type (e.g. a Protocol union like StateLike)
• Catches PydanticSchemaError during validate_arguments and retries with injected arg annotations replaced by Annotated[Any, InjectedToolArg()], preserving the injected marker while using a Pydantic-safe type
• Original annotations are restored after schema creation

Closes #32067

Files changed

• libs/core/langchain_core/tools/base.py — try/except in create_schema_from_function to handle SchemaError for injected args
• libs/core/tests/unit_tests/test_tools.py — 2 new tests: Protocol union type (reproduces Potential validation issue when using StateLike in Annotation for a tool #32067) and multiple injected arg types

Test plan

• test_injected_arg_with_protocol_union_type — verifies tool with Annotated[StateLike, InjectedToolArg] can be created and invoked
• test_injected_arg_filtered_from_schema_at_creation — verifies both InjectedToolArg and _DirectlyInjectedToolArg types work correctly
• All 163 existing tests in test_tools.py pass (no regressions)

🤖 Generated with Claude Code

[codspeed-hq]

Merging this PR will not alter performance

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

✅ 13 untouched benchmarks
⏩ 23 skipped benchmarks¹

---

_{Comparing AlonNaor22:fix/filter-injected-args-from-schema (f51f100) with master (29134dc)}

Footnotes

1. 23 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩