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

[PR #5243] feat(langgraph): new context api (replacing config['configurable'] and config_schema) #4327

New Issue
Closed
opened 2026-02-20 17:50:01 -05:00 by yindo · 0 comments
yindo commented 2026-02-20 17:50:01 -05:00
Owner
Copy Link
Copy Source

Original Pull Request: https://github.com/langchain-ai/langgraph/pull/5243

State: closed
Merged: Yes

Overview

This PR introduces a new API that provides a cleaner, more type-safe way to pass runtime context to LangGraph nodes/tasks. It replaces the current pattern of using config['configurable'] and config_schema with a dedicated context parameter and wrapper Runtime object.

What's Changed

Before/After: Basic Context Usage

Before (Old Pattern)

from langchain_core.runnables import RunnableConfig

def node(state: State, config: RunnableConfig):
    user_id = config.get("configurable", {}).get("user_id")
    return {"result": f"Hello {user_id}"}

graph.invoke(input_data, config={"configurable": {"user_id": "123"}})

After (New Pattern)

from dataclasses import dataclass
from langgraph.runtime import Runtime

@dataclass
class ContextSchema:
    user_id: str

def node(state: State, runtime: Runtime[ContextSchema]):
    user_id = runtime.context.user_id
    return {"result": f"Hello {user_id}"}

graph.invoke(input_data, context={"user_id": "123"})

OR, you can use the get_runtime method:

from langgraph.runtime import get_runtime

def node(state: State):
    user_id = get_runtime(ContextSchema).context.user_id
    return {"result": f"Hello {user_id}"}
Before/After: Store and Stream Writer Access

Before (Old pattern)

from langgraph.store.base import BaseStore
from langchain_core.runnables import RunnableConfig

def update_memory(state: MessagesState, config: RunnableConfig, *, store: BaseStore):
    user_id = config.get("configurable", {}).get("user_id")
    namespace = (user_id, "memories")
    memory_id = str(uuid.uuid4())
    store.put(namespace, memory_id, {"memory": memory})

After (new pattern)

from langgraph.runtime import Runtime

def update_memory(state: MessagesState, runtime: Runtime[ContextSchema]):
    user_id = runtime.context.user_id
    namespace = (user_id, "memories")
    memory_id = str(uuid.uuid4())
    runtime.store.put(namespace, memory_id, {"memory": memory})

Key Benefits

  • Type Safety: Type checked context input to invoke / stream, plus typed access to Runtime attributes
  • Cleaner API: Direct context parameter instead of nested config['configurable']
  • Better DX: IDE autocomplete for context fields and Runtime attributes
  • Unified Runtime: Single Runtime object provides access to context, store, and stream writer, with room for expanding to streamlined config/checkpoint information in the near future.

Breaking Changes & Migration

Deprecated APIs

  • StateGraph(..., config_schema=X) -> StateGraph(..., context_schema=X)
  • Pregel.config_schema → Pregel.get_context_jsonschema() - this is largely meant to be external and we don't anticipate this affecting many users

Migration Details

  • Maintains backward compatibility with existing config['configurable'] usage
  • Deprecation warnings guide users to the new API

Future Work

  • Deprecate injection pattern for store, stream_writer, maybe previous, update docs to recommend popping from runtime.
  • Support Runtime injection for tools, right now only get_runtime is supported
  • LangGraph style guide with recommended best practices

Eventually, I think we should move away from storing and popping things from config["configurable"], which can be a fully internal refactor. Though I would love to do this pre v1, it should largely be internal, so can be done afterwards. Config changes should be in a different PR than this one to keep things reasonably scoped. This PR is already pretty big. Lots of plumbing.

Related Issues

Closes #5023

**Original Pull Request:** https://github.com/langchain-ai/langgraph/pull/5243 **State:** closed **Merged:** Yes --- ## Overview This PR introduces a new API that provides a cleaner, more type-safe way to pass runtime context to LangGraph nodes/tasks. It replaces the current pattern of using `config['configurable']` and `config_schema` with a dedicated `context` parameter and wrapper `Runtime` object. ## What's Changed ### Before/After: Basic Context Usage ### Before (Old Pattern) ```python from langchain_core.runnables import RunnableConfig def node(state: State, config: RunnableConfig): user_id = config.get("configurable", {}).get("user_id") return {"result": f"Hello {user_id}"} graph.invoke(input_data, config={"configurable": {"user_id": "123"}}) ``` ### After (New Pattern) ```python from dataclasses import dataclass from langgraph.runtime import Runtime @dataclass class ContextSchema: user_id: str def node(state: State, runtime: Runtime[ContextSchema]): user_id = runtime.context.user_id return {"result": f"Hello {user_id}"} graph.invoke(input_data, context={"user_id": "123"}) ``` OR, you can use the `get_runtime` method: ```python from langgraph.runtime import get_runtime def node(state: State): user_id = get_runtime(ContextSchema).context.user_id return {"result": f"Hello {user_id}"} ``` <details> <summary>Before/After: Store and Stream Writer Access</summary> ### Before (Old pattern) ```py from langgraph.store.base import BaseStore from langchain_core.runnables import RunnableConfig def update_memory(state: MessagesState, config: RunnableConfig, *, store: BaseStore): user_id = config.get("configurable", {}).get("user_id") namespace = (user_id, "memories") memory_id = str(uuid.uuid4()) store.put(namespace, memory_id, {"memory": memory}) ``` ### After (new pattern) ```py from langgraph.runtime import Runtime def update_memory(state: MessagesState, runtime: Runtime[ContextSchema]): user_id = runtime.context.user_id namespace = (user_id, "memories") memory_id = str(uuid.uuid4()) runtime.store.put(namespace, memory_id, {"memory": memory}) ``` </details> ## Key Benefits - **Type Safety**: Type checked `context` input to `invoke` / `stream`, plus typed access to `Runtime` attributes - **Cleaner API**: Direct `context` parameter instead of nested `config['configurable']` - **Better DX**: IDE autocomplete for context fields and `Runtime` attributes - **Unified Runtime**: Single `Runtime` object provides access to context, store, and stream writer, with room for expanding to streamlined config/checkpoint information in the near future. ## Breaking Changes & Migration ### Deprecated APIs - `StateGraph(..., config_schema=X)` -> `StateGraph(..., context_schema=X)` - `Pregel.config_schema` → `Pregel.get_context_jsonschema()` - this is largely meant to be external and we don't anticipate this affecting many users ### Migration Details - Maintains backward compatibility with existing `config['configurable']` usage - Deprecation warnings guide users to the new API ## Future Work - [ ] Deprecate injection pattern for `store`, `stream_writer`, maybe `previous`, update docs to recommend popping from runtime. - [ ] Support `Runtime` injection for tools, right now only `get_runtime` is supported - [ ] LangGraph style guide with recommended best practices Eventually, I think we should move away from storing and popping things from `config["configurable"]`, which can be a fully internal refactor. Though I would love to do this pre v1, it should largely be internal, so can be done afterwards. Config changes should be in a different PR than this one to keep things reasonably scoped. This PR is already pretty big. Lots of plumbing. ## Related Issues Closes #5023
yindo added the pull-request label 2026-02-20 17:50:01 -05:00
yindo closed this issue 2026-02-20 17:50:01 -05:00
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 pull-request
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#4327
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?