fix(langchain): use positional args for wrap_function_wrapper for wrapt v1/v2 compat, fix BaseChatOpenAI uninstrument

[dvirski]

Fixes #4009 — LangChain init crash: wrap_function_wrapper() got unexpected keyword argument 'module' when using wrapt v2

Also fixes a pre-existing bug where _uninstrument() was unwrapping BaseOpenAI instead of BaseChatOpenAI in langchain_openai.chat_models.base, meaning those wrappers were never actually removed on uninstrument.

Summary by CodeRabbit

• New Features

  • Expanded OpenTelemetry instrumentation for Agno framework (Agent, Team, function execution)
  • Extended LangChain tracing for additional OpenAI LLM/chat backends and LangGraph components

• Bug Fixes

  • Corrected uninstrumentation behavior for improved removal of OpenAI-related wrappers

• Improvements

  • Standardized wrapper handling to make tracing more consistent across instrumentations

[CLAassistant]

All committers have signed the CLA.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: e3f6b206-df37-4138-9216-9b909c3354fe

📥 Commits

Reviewing files that changed from the base of the PR and between 90b1e29 and f935608.

📒 Files selected for processing (2)

• packages/opentelemetry-instrumentation-agno/opentelemetry/instrumentation/agno/__init__.py
• packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py

---

📝 Walkthrough

Walkthrough

Standardize calls to wrap_function_wrapper to use positional arguments across Agno and Langchain instrumentations; also change langchain_openai uninstrument targets to BaseChatOpenAI._generate/_agenerate.

Changes

Instrumentation wrap_function_wrapper refactor (Agno & Langchain)

Layer / File(s)	Summary
Agno wrapper refactor packages/opentelemetry-instrumentation-agno/opentelemetry/instrumentation/agno/__init__.py	Six wrap_function_wrapper calls (Agent.run/arun, Team.run/arun, FunctionCall.execute/aexecute) switched from keyword to positional arguments; no functional change.
BaseCallbackManager wrapper packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	BaseCallbackManager.__init__ wrap invocation converted from wrapper= kw to positional argument.
OpenAI backend wrappers packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	OpenAI-related wrap_function_wrapper calls (langchain_community / langchain_openai backends) updated to positional-argument form for LLM/chat methods.
LangGraph Pregel & Command packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	Pregel.stream/astream and Command.__init__ wrap calls switched to positional-argument usage.
Agent factory wrappers packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	create_react_agent and create_agent wrappers (including re-export modules) changed to positional-argument form.
AgentMiddleware hooks packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	Sync and async AgentMiddleware.<hook> wrap_function_wrapper calls updated to use positional arguments.
Uninstrument adjustment packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/__init__.py	Uninstrument for langchain_openai now unwraps BaseChatOpenAI._generate/_agenerate instead of the prior BaseOpenAI targets.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Suggested reviewers

• galzilber
• nina-kollman
• max-deygin-traceloop
• netanel-tl

Poem

🐰 I hopped through code to change a call,

Positional args now stand tall,
Agno runs and agents hum,
Langchain wrappers updated, done!
A tiny tweak, a tidy prance — tracing gets its chance ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main fix: converting wrap_function_wrapper calls to use positional arguments for wrapt v1/v2 compatibility and fixing BaseChatOpenAI uninstrumentation.
Linked Issues check	✅ Passed	The PR directly addresses issue #4009 by converting wrap_function_wrapper calls from keyword arguments to positional arguments, fixing the TypeError that occurred with wrapt v2. The uninstrumentation bug fix for BaseChatOpenAI also aligns with the mentioned pre-existing bug.
Out of Scope Changes check	✅ Passed	All changes are scoped to fixing wrap_function_wrapper call syntax and the BaseChatOpenAI uninstrumentation bug; no unrelated modifications were introduced.

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

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix(langchain)-use-positional-args-for-wrap_function_wrapper-for-wrapt-v1/v2-compat,-fix-BaseChatOpenAI-uninstrument

---

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.}

[kamilamin123]

thank you I hope this will be released soon I will cancel my pr 👍

[GuilhermePSC]

Hey, thx for the fix! Do you have any schedule to deploy this?