langchain-ai/langgraph
1
0
Fork 0
mirror of https://github.com/langchain-ai/langgraph.git synced 2026-07-15 12:35:43 -04:00
Code Issues 314 Packages Projects Releases 17 Wiki Activity

update type annotations in create_react_agents #868

New Issue
Closed
opened 2026-02-20 17:42:11 -05:00 by yindo · 1 comment
yindo commented 2026-02-20 17:42:11 -05:00
Owner
Copy Link
Copy Source

Originally created by @eyurtsev on GitHub (Jul 30, 2025).

Privileged issue

  • I am a LangGraph maintainer, or was asked directly by a LangGraph maintainer to create an issue here.

Issue Content

bind_tools return Runnable[LanguageModelInput, BaseMessage], not BaseChatModel and this is actually a bug in type annotations.

Please update the type annotations in create_react_agent to allow one to provide a callable for the model that uses bind_tools and returns a Runnable[LanguageModelInput, BaseMessage].

i.e., we want to widen the accepted types

Agent Context { "tasks": [ { "id": "9679a1a1-5ed2-4d57-81c5-7e0167d92d05", "taskIndex": 0, "request": "[original issue]\n**update type annotations in create_react_agents**\n### Privileged issue\n\n- [x] I am a LangGraph maintainer, or was asked directly by a LangGraph maintainer to create an issue here.\n\n### Issue Content\n\nbind_tools return Runnable[LanguageModelInput, BaseMessage], not BaseChatModel and this is actually a bug in type annotations.\n\nPlease update the type annotations in create_react_agent to allow one to provide a callable for the model that uses bind_tools and returns a Runnable[LanguageModelInput, BaseMessage].\n\ni.e., we want to widen the accepted types\n\n", "title": "Update type annotations in create_react_agent to allow bind_tools return type", "createdAt": 1753900472096, "completed": true, "planRevisions": [ { "revisionIndex": 0, "plans": [ { "index": 0, "plan": "**Update the type annotations for dynamic model callables in `create_react_agent`** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 255-256. Change `CallableStateSchema, Runtime[ContextT, BaseChatModel]` to `CallableStateSchema, Runtime[ContextT, LanguageModelLike]` and `CallableStateSchema, Runtime[ContextT, Awaitable[BaseChatModel]]` to `CallableStateSchema, Runtime[ContextT, Awaitable[LanguageModelLike]]`. This allows dynamic model functions to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`).", "completed": true, "summary": "Successfully updated the type annotations for dynamic model callables in `create_react_agent` function in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py`. \n\n**Changes made:**\n- Line 255: Changed `CallableStateSchema, Runtime[ContextT, BaseChatModel]` to `CallableStateSchema, Runtime[ContextT, LanguageModelLike]`\n- Line 256: Changed `CallableStateSchema, Runtime[ContextT, Awaitable[BaseChatModel]]` to `CallableStateSchema, Runtime[ContextT, Awaitable[LanguageModelLike]]`\n\n**Impact:**\nThis change widens the accepted return types for dynamic model callables, allowing them to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`). This fixes the type annotation bug where dynamic model functions that use `bind_tools()` were incorrectly constrained to return only `BaseChatModel` instead of the broader `LanguageModelLike` type that includes `RunnableBinding` objects returned by `bind_tools()`.\n\nThe change maintains backward compatibility since `BaseChatModel` is a subset of `LanguageModelLike`, while enabling the intended functionality of using models with bound tools in dynamic model callables." }, { "index": 1, "plan": "**Update the docstring for the model parameter** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 288 and 292-293. Change the text from `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike` and update \"Must return a `BaseChatModel` instance\" to \"Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)\" to reflect that dynamic models can return the wider type that includes bind_tools results.", "completed": true, "summary": "Successfully updated the docstring for the model parameter in `create_react_agent` function in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py`.\n\n**Changes made:**\n- Line 288: Changed `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike`\n- Lines 292-293: Updated \"Must return a `BaseChatModel` instance\" to \"Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)\"\n- Added clarification that the return type can be either a `BaseChatModel` or the result of `model.bind_tools()`\n\n**Impact:**\nThe docstring now accurately reflects the updated type annotations from the previous task. It clearly communicates to users that dynamic model callables can return either a `BaseChatModel` directly or the result of calling `bind_tools()` on a model, which returns a `LanguageModelLike` type. This provides better guidance for developers using dynamic models with bound tools in their `create_react_agent` implementations." }, { "index": 2, "plan": "**Run the formatting, linting, and testing commands** in the `/home/daytona/langgraph/libs/prebuilt` directory: `make format`, `make lint`, and `make test` to ensure the changes don't break existing functionality and follow the project's code standards.", "completed": true, "summary": "Successfully ran the formatting, linting, and testing commands in the `/home/daytona/langgraph/libs/prebuilt` directory to ensure the changes don't break existing functionality and follow the project's code standards.\n\n**Commands executed and results:**\n\n1. **`make format`**: ✅ Completed successfully\n - Reformatted 1 file, 17 files left unchanged\n - All formatting checks passed\n\n2. **`make lint`**: ✅ Completed successfully\n - All ruff checks passed\n - All import sorting checks passed\n - All mypy type checks passed with no issues found in 6 source files\n\n3. **`make test`**: ⚠️ Docker unavailable, ran `uv run pytest . -v` instead\n - All 115 tests passed successfully\n - No test failures related to the type annotation changes\n - Only minor deprecation warnings from dependencies (unrelated to our changes)\n\n**Validation results:**\n- The type annotation changes from `BaseChatModel` to `LanguageModelLike` for dynamic model callables are working correctly\n- All existing functionality remains intact\n- The code follows the project's formatting and linting standards\n- Tests confirm that the wider type acceptance doesn't break any existing use cases\n\nThe changes successfully allow dynamic model functions to return the result of `bind_tools()` while maintaining backward compatibility with existing code that returns `BaseChatModel` directly." } ], "createdAt": 1753900472096, "createdBy": "agent" } ], "activeRevisionIndex": 0, "pullRequestNumber": 5740, "completedAt": 1753900653582, "summary": "Updated `create_react_agent` in `chat_agent_executor.py` to widen dynamic model callable return types from `BaseChatModel` to `LanguageModelLike`, allowing `bind_tools()` results. Also revised related docstrings accordingly. Ran formatting, linting, and tests—115 tests passed with no issues. Changes maintain backward compatibility and fix the type annotation bug as requested." } ], "activeTaskIndex": 0 }
Originally created by @eyurtsev on GitHub (Jul 30, 2025). ### Privileged issue - [x] I am a LangGraph maintainer, or was asked directly by a LangGraph maintainer to create an issue here. ### Issue Content bind_tools return Runnable[LanguageModelInput, BaseMessage], not BaseChatModel and this is actually a bug in type annotations. Please update the type annotations in create_react_agent to allow one to provide a callable for the model that uses bind_tools and returns a Runnable[LanguageModelInput, BaseMessage]. i.e., we want to widen the accepted types <details> <summary>Agent Context</summary> <open-swe-do-not-edit-task-plan> { "tasks": [ { "id": "9679a1a1-5ed2-4d57-81c5-7e0167d92d05", "taskIndex": 0, "request": "[original issue]\n**update type annotations in create_react_agents**\n### Privileged issue\n\n- [x] I am a LangGraph maintainer, or was asked directly by a LangGraph maintainer to create an issue here.\n\n### Issue Content\n\nbind_tools return Runnable[LanguageModelInput, BaseMessage], not BaseChatModel and this is actually a bug in type annotations.\n\nPlease update the type annotations in create_react_agent to allow one to provide a callable for the model that uses bind_tools and returns a Runnable[LanguageModelInput, BaseMessage].\n\ni.e., we want to widen the accepted types\n\n", "title": "Update type annotations in create_react_agent to allow bind_tools return type", "createdAt": 1753900472096, "completed": true, "planRevisions": [ { "revisionIndex": 0, "plans": [ { "index": 0, "plan": "**Update the type annotations for dynamic model callables in `create_react_agent`** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 255-256. Change `Callable[[StateSchema, Runtime[ContextT]], BaseChatModel]` to `Callable[[StateSchema, Runtime[ContextT]], LanguageModelLike]` and `Callable[[StateSchema, Runtime[ContextT]], Awaitable[BaseChatModel]]` to `Callable[[StateSchema, Runtime[ContextT]], Awaitable[LanguageModelLike]]`. This allows dynamic model functions to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`).", "completed": true, "summary": "Successfully updated the type annotations for dynamic model callables in `create_react_agent` function in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py`. \n\n**Changes made:**\n- Line 255: Changed `Callable[[StateSchema, Runtime[ContextT]], BaseChatModel]` to `Callable[[StateSchema, Runtime[ContextT]], LanguageModelLike]`\n- Line 256: Changed `Callable[[StateSchema, Runtime[ContextT]], Awaitable[BaseChatModel]]` to `Callable[[StateSchema, Runtime[ContextT]], Awaitable[LanguageModelLike]]`\n\n**Impact:**\nThis change widens the accepted return types for dynamic model callables, allowing them to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`). This fixes the type annotation bug where dynamic model functions that use `bind_tools()` were incorrectly constrained to return only `BaseChatModel` instead of the broader `LanguageModelLike` type that includes `RunnableBinding` objects returned by `bind_tools()`.\n\nThe change maintains backward compatibility since `BaseChatModel` is a subset of `LanguageModelLike`, while enabling the intended functionality of using models with bound tools in dynamic model callables." }, { "index": 1, "plan": "**Update the docstring for the model parameter** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 288 and 292-293. Change the text from `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike` and update \"Must return a `BaseChatModel` instance\" to \"Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)\" to reflect that dynamic models can return the wider type that includes bind_tools results.", "completed": true, "summary": "Successfully updated the docstring for the model parameter in `create_react_agent` function in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py`.\n\n**Changes made:**\n- Line 288: Changed `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike`\n- Lines 292-293: Updated \"Must return a `BaseChatModel` instance\" to \"Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)\"\n- Added clarification that the return type can be either a `BaseChatModel` or the result of `model.bind_tools()`\n\n**Impact:**\nThe docstring now accurately reflects the updated type annotations from the previous task. It clearly communicates to users that dynamic model callables can return either a `BaseChatModel` directly or the result of calling `bind_tools()` on a model, which returns a `LanguageModelLike` type. This provides better guidance for developers using dynamic models with bound tools in their `create_react_agent` implementations." }, { "index": 2, "plan": "**Run the formatting, linting, and testing commands** in the `/home/daytona/langgraph/libs/prebuilt` directory: `make format`, `make lint`, and `make test` to ensure the changes don't break existing functionality and follow the project's code standards.", "completed": true, "summary": "Successfully ran the formatting, linting, and testing commands in the `/home/daytona/langgraph/libs/prebuilt` directory to ensure the changes don't break existing functionality and follow the project's code standards.\n\n**Commands executed and results:**\n\n1. **`make format`**: ✅ Completed successfully\n - Reformatted 1 file, 17 files left unchanged\n - All formatting checks passed\n\n2. **`make lint`**: ✅ Completed successfully\n - All ruff checks passed\n - All import sorting checks passed\n - All mypy type checks passed with no issues found in 6 source files\n\n3. **`make test`**: ⚠️ Docker unavailable, ran `uv run pytest . -v` instead\n - All 115 tests passed successfully\n - No test failures related to the type annotation changes\n - Only minor deprecation warnings from dependencies (unrelated to our changes)\n\n**Validation results:**\n- The type annotation changes from `BaseChatModel` to `LanguageModelLike` for dynamic model callables are working correctly\n- All existing functionality remains intact\n- The code follows the project's formatting and linting standards\n- Tests confirm that the wider type acceptance doesn't break any existing use cases\n\nThe changes successfully allow dynamic model functions to return the result of `bind_tools()` while maintaining backward compatibility with existing code that returns `BaseChatModel` directly." } ], "createdAt": 1753900472096, "createdBy": "agent" } ], "activeRevisionIndex": 0, "pullRequestNumber": 5740, "completedAt": 1753900653582, "summary": "Updated `create_react_agent` in `chat_agent_executor.py` to widen dynamic model callable return types from `BaseChatModel` to `LanguageModelLike`, allowing `bind_tools()` results. Also revised related docstrings accordingly. Ran formatting, linting, and tests—115 tests passed with no issues. Changes maintain backward compatibility and fix the type annotation bug as requested." } ], "activeTaskIndex": 0 } </open-swe-do-not-edit-task-plan> </details>
yindo added the open-swe-auto label 2026-02-20 17:42:11 -05:00
yindo closed this issue 2026-02-20 17:42:11 -05:00
yindo commented 2026-02-20 17:42:11 -05:00
Author
Owner
Copy Link
Copy Source

@open-swe[bot] commented on GitHub (Jul 30, 2025):

🤖 Open SWE has been triggered for this issue. Processing...

View run in Open SWE here (this URL will only work for @eyurtsev)

Dev Metadata { "runId": "1f06d733-3009-6347-b28f-57687edd2809", "threadId": "1d37d949-de60-4999-85ec-b835dd3fa7f8" }

🤖 Plan Generated

I've generated a plan for this issue and will proceed to implement it since auto-accept is enabled.

Plan: Update type annotations in create_react_agent to allow bind_tools return type

  • Task 1:
**Update the type annotations for dynamic model callables in `create_react_agent`** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 255-256. Change `Callable[[StateSchema, Runtime[ContextT]], BaseChatModel]` to `Callable[[StateSchema, Runtime[ContextT]], LanguageModelLike]` and `Callable[[StateSchema, Runtime[ContextT]], Awaitable[BaseChatModel]]` to `Callable[[StateSchema, Runtime[ContextT]], Awaitable[LanguageModelLike]]`. This allows dynamic model functions to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`).
  • Task 2:
**Update the docstring for the model parameter** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 288 and 292-293. Change the text from `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike` and update "Must return a `BaseChatModel` instance" to "Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)" to reflect that dynamic models can return the wider type that includes bind_tools results.
  • Task 3:
**Run the formatting, linting, and testing commands** in the `/home/daytona/langgraph/libs/prebuilt` directory: `make format`, `make lint`, and `make test` to ensure the changes don't break existing functionality and follow the project's code standards.

Proceeding to implementation...

@open-swe[bot] commented on GitHub (Jul 30, 2025): 🤖 Open SWE has been triggered for this issue. Processing... View run in Open SWE [here](https://swe.langchain.com/chat/1d37d949-de60-4999-85ec-b835dd3fa7f8) (this URL will only work for @eyurtsev) <details> <summary>Dev Metadata</summary> { "runId": "1f06d733-3009-6347-b28f-57687edd2809", "threadId": "1d37d949-de60-4999-85ec-b835dd3fa7f8" } </details> <open-swe-plan-message> ### 🤖 Plan Generated I've generated a plan for this issue and will proceed to implement it since auto-accept is enabled. **Plan: Update type annotations in create_react_agent to allow bind_tools return type** - Task 1: ``` **Update the type annotations for dynamic model callables in `create_react_agent`** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 255-256. Change `Callable[[StateSchema, Runtime[ContextT]], BaseChatModel]` to `Callable[[StateSchema, Runtime[ContextT]], LanguageModelLike]` and `Callable[[StateSchema, Runtime[ContextT]], Awaitable[BaseChatModel]]` to `Callable[[StateSchema, Runtime[ContextT]], Awaitable[LanguageModelLike]]`. This allows dynamic model functions to return the result of `bind_tools()` which is `Runnable[LanguageModelInput, BaseMessage]` (aliased as `LanguageModelLike`). ``` - Task 2: ``` **Update the docstring for the model parameter** in `/home/daytona/langgraph/libs/prebuilt/langgraph/prebuilt/chat_agent_executor.py` at lines 288 and 292-293. Change the text from `(state, runtime) -> BaseChatModel` to `(state, runtime) -> LanguageModelLike` and update "Must return a `BaseChatModel` instance" to "Must return a `LanguageModelLike` instance (e.g., `BaseChatModel` or result of `model.bind_tools()`)" to reflect that dynamic models can return the wider type that includes bind_tools results. ``` - Task 3: ``` **Run the formatting, linting, and testing commands** in the `/home/daytona/langgraph/libs/prebuilt` directory: `make format`, `make lint`, and `make test` to ensure the changes don't break existing functionality and follow the project's code standards. ``` Proceeding to implementation... </open-swe-plan-message>
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
agents
awaiting author revision
bug
change
chat models
checkpointer
ci
code quality
code quality
dependencies
documentation
drawing
duplicate
enhancement
good first issue
help wanted
invalid
maintainer
meta
missing information
not maintained by langchain
open-swe-auto
open-swe-max
open-swe-max-auto
pending
performance
pull-request
Mirrored from GitHub Pull Request
 question
requires design
sdk
server
stale
studio
tests
type checking
usage question
v0.5
v1.1
No Label open-swe-auto
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/langgraph#868
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?