LangChain Integration

Kaplaix provides zero-code adapters for LangChain in both TypeScript and Python. The adapters use LangChain’s callback system to automatically capture tool calls as Kaplaix events.

TypeScript

Installation


npm install @kaplaix/adapter-langchain-ts @kaplaix/sdk-ts

Peer dependencies: @langchain/core >= 0.3.0, @langchain/langgraph >= 0.2.0

Usage


import { KaplaixClient } from "@kaplaix/sdk-ts";
import { KaplaixHandler } from "@kaplaix/adapter-langchain-ts";
 
const client = new KaplaixClient({ apiKey: "al_live_..." });
const session = client.session({ agentId: "my-langchain-agent" });
 
// Create the callback handler
const handler = new KaplaixHandler(session);
 
// Pass it to your LangChain agent
const result = await agent.invoke(
  { input: "What is the weather?" },
  { callbacks: [handler] },
);
 
await session.end();
await client.shutdown();

The handler automatically emits tool_api events for every tool invocation, capturing tool name, arguments hash, and response status.

Python

Installation


pip install kaplaix-langchain kaplaix

Dependencies: kaplaix >= 0.1.0, langchain-core >= 0.3

Usage


from kaplaix import KaplaixClient
from kaplaix_langchain import KaplaixHandler
 
client = KaplaixClient(api_key="al_live_...")
 
with client.session(agent_id="my-langchain-agent") as session:
    handler = KaplaixHandler(session)
 
    result = agent.invoke(
        {"input": "What is the weather?"},
        config={"callbacks": [handler]},
    )

What gets captured

The LangChain adapter captures:

LangChain event	Kaplaix category	Payload fields
Tool start/end	`tool_api`	`toolName`, `argumentsHash`, `responseStatus`

Other framework adapters

Kaplaix also provides adapters for:

LangGraph (@kaplaix/adapter-langgraph-ts) — automatic event capture for LangGraph.js agents
CrewAI (kaplaix-crewai) — Python adapter for CrewAI agents

These follow the same pattern: install the adapter, create a handler from a session, and pass it to your framework’s callback system.