fix(core): make self positional-only in _run/_arun to allow "self" as tool input key

[Varun Chawla (veeceey)]

Summary

• Fixes BaseTool failing with TypeError: got multiple values for argument 'self' when tool input contains a key named "self" (e.g. tool.invoke({"self": 2}))
• Adds the positional-only parameter marker (/) after self in _run and _arun across BaseTool, StructuredTool, and Tool, so Python routes user-supplied "self" into **kwargs instead of conflicting with the method receiver
• Adds sync and async regression tests

Closes #34900

Test plan

• New unit test test_tool_invoke_with_self_keyword_in_input verifies sync invocation with "self" in input
• New unit test test_tool_ainvoke_with_self_keyword_in_input verifies async invocation with "self" in input
• Full existing test suite passes (157 passed, 4 skipped) with no regressions

Review notes

The change is minimal -- just adding / after self in the method signatures. This is a non-breaking change: all existing subclass overrides of _run/_arun continue to work exactly as before. The positional-only marker only affects whether Python allows self=... as a keyword argument at the call site; it does not change how the method is dispatched or overridden.

Note that user-defined BaseTool subclasses that override _run without self, / will still hit this issue if they receive "self" as a keyword argument. However, the built-in tool classes (StructuredTool, Tool) and the @tool decorator -- which cover the vast majority of usage -- are all fixed.

[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 veeceey:fix/issue-34900-tool-self-param (809e0a1) with master (5624001)²}

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. ↩

2. No successful run was found on master (4a632cf) during the generation of this report, so 5624001 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[Varun Chawla (veeceey)]

Manual test results

Ran tests/unit_tests/test_tools.py from libs/core on branch fix/issue-34900-tool-self-param (Python 3.14):

Targeted tests (-k "self"):

tests/unit_tests/test_tools.py::test_tool_invoke_with_self_keyword_in_input PASSED
tests/unit_tests/test_tools.py::test_tool_ainvoke_with_self_keyword_in_input PASSED

Full test_tools.py suite (regression check):

157 passed, 4 skipped, 254 warnings in 0.35s

No failures, no regressions. The positional-only self fix works as expected — tools with a "self" input key can now be invoked without hitting TypeError: _run() got multiple values for argument 'self'.

[Varun Chawla (veeceey)]

All tests passing, ready for review/merge

[Varun Chawla (veeceey)]

Hi maintainers, friendly ping on this PR. It's been open a few days now and all checks are passing. Would appreciate it whenever you get a chance to review. Thank you for your time!

[Varun Chawla (veeceey)]

Quick ping -- this one's been green for a while now. Happy to address any feedback if needed. Thanks for your time!