anlyon-langchain

LangChain integration for Anlyon: AnlyonVectorStore and AnlyonRetriever over Anlyon Memory, AnlyonChatMessageHistory over Anlyon Sessions, and AnlyonTracer for nested run/span tracing.

Trace a LangChain invocation

from anlyon import Client
from anlyon_langchain import AnlyonTracer

client = Client(api_key="anlyon_live_...")
tracer = AnlyonTracer(client, application="support-agent")

# Pass the callback to any LangChain runnable/agent invocation.
result = chain.invoke({"question": "Where is my order?"}, config={"callbacks": [tracer]})

The handler creates one Anlyon run for the root callback and parent-linked workflow_step, model_call, and tool_call spans for nested callbacks. It is best-effort: Anlyon transport failures never change the application result.

Installation

pip install anlyon-langchain

Retrieval

from anlyon import Client
from anlyon_langchain import AnlyonVectorStore

client = Client()  # uses ANLYON_API_KEY
store = AnlyonVectorStore(client, collection="docs")
retriever = store.as_retriever()

Chat history

AnlyonChatMessageHistory puts a chain's conversation in an Anlyon Session durable across processes, and summarized automatically once the thread gets long.

from anlyon import Client
from anlyon_langchain import AnlyonChatMessageHistory
from langchain_core.runnables.history import RunnableWithMessageHistory

client = Client()

chain_with_history = RunnableWithMessageHistory(
    chain,
    lambda session_id: AnlyonChatMessageHistory(client, session_id),
    input_messages_key="input",
    history_messages_key="history",
)

# `session_id` is an Anlyon session id  create one with client.sessions.create()
chain_with_history.invoke(
    {"input": "Where is my order?"},
    config={"configurable": {"session_id": "ses_abc123"}},
)

.messages reads the thread back as LangChain messages, .add_messages() appends a turn in one round-trip, and .clear() truncates the thread while keeping the session id usable.

See the Anlyon docs for the full guide.

License

MIT