Feature Request: `to_message_state()` and `as_message_state()` helpers

[davicaetano]

Checked other resources

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-cli
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-perplexity
• langchain-prompty
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

I would like to request two small additions to improve the ergonomics of LCEL pipelines when working with message-based agents:

1. A helper function

to_message_state(obj)
that converts strings, BaseMessage objects, lists of messages, or mixed inputs into the standard message-state format:

{"messages": [BaseMessage, ...]}

2. A Runnable utility

as_message_state()
which allows this conversion to be inserted cleanly in LCEL pipe chains (e.g. prompt | llm | as_message_state()).

This avoids the need for repeated:

RunnableLambda(lambda x: {"messages": [x]})

and standardizes a pattern used by many LangGraph-based workflows.

Use Case

Many LCEL pipelines ultimately need to output a message-state dictionary so they can be consumed by LangGraph or any message-history–based workflow.

Currently, users must manually wrap LLM outputs:

RunnableLambda(lambda x: {"messages": [x]})

This pattern:

• is verbose
• is error-prone
• appears repeatedly in real-world user code
• makes LCEL pipelines harder to read
• creates friction when integrating LangChain with LangGraph

Providing to_message_state() and as_message_state() would make this workflow clean, consistent, and beginner-friendly.

Proposed Solution

Helper function

to_message_state(obj)

Behavior:

• str → converted into HumanMessage
• BaseMessage → wrapped directly
• list → normalized into a list of messages
• {"messages": [...]} → returned unchanged
• None → returns {"messages": []}
• invalid types → TypeError

Always outputs:

{"messages": [BaseMessage, ...]}

Runnable utility

as_message_state()

A thin wrapper around the helper:

chain = prompt | llm | as_message_state()

Implementation notes

I am happy to submit a PR with:

• the helper
• the runnable
• complete test coverage
• necessary documentation updates
• updated examples where appropriate

Alternatives Considered

1. Keep using RunnableLambda manually

This works:

RunnableLambda(lambda x: {"messages": [x]})

But is verbose, error-prone, and inconsistent.

2. Implement this in LangGraph instead of LangChain

This is not ideal because:

• LCEL, Runnables, and message types live in LangChain
• LangGraph only consumes normalized state and does not transform LLM output

3. Modify LLM classes to output MessageState directly

Too intrusive; breaks separation of concerns.

Additional Context

• MessageState ({"messages": [...]}) is widely used in LangGraph and message-history–based workflows.
• There is currently no official helper or Runnable in LangChain to convert arbitrary LLM outputs into this state format.
• Several community examples show users doing this manually with RunnableLambda.
• This proposal aims to standardize and simplify that pattern.
• I can contribute a PR once maintainers approve the design.