DOCSEARCHbyAlgolia
Submit Docs

LangChain Documentation

website·Indexed 31 days ago

https://langchain-doc.cn

Documentation for LangChain and LangGraph, focusing on building production-ready AI agents. Includes guides on the Graph API, state management, persistence, and streaming, as well as integration details for providers like OpenAI, Anthropic, and AWS. Covers migration to v1, structured output, and the functional API for creating complex workflows with LLMs.

Tokens
127.4K
Snippets
582
Records
746
Agent score
58%

What's inside LangChain

  1. Overview of LangChain Python Integration Providers

    v1
    LangChain provides a vast ecosystem of integrations for Python, allowing developers to connect LLMs and AI agents to various external tools, databases, and platforms. These integrations cover several categories including LLM providers, chat models, retrievers, vector stores, and document loaders.

  2. Overview of the deepagents library

    v1
    deepagents is a standalone library built on LangGraph designed for creating agents that handle complex, multi-step tasks. It is inspired by applications like Claude Code, Deep Research, and Manus, providing built-in capabilities for planning, context management via a virtual file system, and the ability to spawn specialized sub-agents.

  3. Overview of Agent Chat UI

    v1

    Agent Chat UI is an open-source Next.js application that provides a conversational interface for interacting with LangChain Agents. It is designed to work with Agents created using create_agent and supports both local and deployed environments (such as LangSmith). Key capabilities include:

    • Tool Visualization: Automatic intuitive rendering of tool calls and results.
    • Time-travel Debugging: Ability to navigate conversation history and fork new conversations from any point.
    • State Inspection: Viewing and modifying Agent state at any point during execution.
    • Human-in-the-loop: Built-in support for reviewing and responding to Agent requests.
    • Generative UI: Support for generative user interfaces via LangGraph.

  4. Overview of the LangChain Framework

    LangChain is an application development framework powered by Large Language Models (LLMs). It provides a complete toolchain for building, testing, and deploying reliable agents and complex AI applications, moving from prototype to production level. It is designed as a platform for Agent Engineering.

  5. Overview of Abso LLM Proxy

    v1
    Abso is an open-source LLM proxy designed to balance performance and cost by automatically switching between 'fast models' and 'high-precision models' based on the complexity of the prompt. It utilizes heuristic algorithms to ensure low latency and fast response speeds.

  6. Overview of Agent Chat UI features

    v1

    Agent Chat UI is an open-source Next.js application that provides a conversational interface for interacting with LangChain agents. Key capabilities include:

    • Tool Visualization: Automatically renders tool calls and results in an intuitive interface.
    • Time-Travel Debugging: Allows browsing session history and forking new sessions from any point in time.
    • State Inspection: Enables viewing and modifying the agent's state at any point during execution.
    • Human-in-the-Loop (HITL): Built-in support for reviewing and responding to agent requests.
    • Generative UI: Support for rendering dynamic UI components within the chat.

  7. Overview of LangGraph Studio

    v1
    LangGraph Studio is a free agent IDE integrated with LangSmith that allows developers to visualize, interact with, and debug agents locally. It provides capabilities for tracking decisions, evaluating performance, and prompt engineering, offering a clear view of how an agent 'thinks' and processes information.

  8. Overview of LangChain Open Source Frameworks

    LangChain provides two primary open-source frameworks for building AI agents:

    • LangChain: Designed for quickly getting started with any model provider to build agents.
    • LangGraph: Provides low-level orchestration, memory, and human-in-the-loop support to control every step of a custom agent. It supports durable execution for managing long-running tasks.

  9. Overview of the LangChain Framework

    LangChain is a framework designed for developing applications powered by large language models (LLMs). It provides standard interfaces for LLMs, embedding models, and vector stores, integrating with hundreds of providers. The framework supports the entire application lifecycle: development (using open-source components and integrations), productionization (via LangSmith for monitoring and evaluation), and deployment (via LangGraph Platform for production-ready APIs).

  10. Overview of Model Context Protocol (MCP) in LangChain

    v1
    The Model Context Protocol (MCP) is an open protocol that standardizes how applications provide tools and context to Large Language Models (LLMs). LangChain agents can utilize tools defined on MCP servers through the langchain-mcp-adapters library.

  11. Overview of LangGraph Subgraphs

    v1
    Subgraphs in LangGraph allow developers to decompose complex workflows into smaller, manageable, and reusable components. They are particularly useful for breaking down large workflows, reusing graph logic across multiple processes, building multi-agent systems where each agent is an independent graph, and implementing conditional workflows.

  12. Overview of Multi-Agent System Architectures

    v1

    Multi-agent systems decompose complex applications into specialized agents that collaborate to solve problems. This architecture is particularly useful when:

    1. A single agent has too many tools, making it difficult to select the correct one.
    2. Context or memory becomes too large for a single agent to track effectively.
    3. Tasks require specialized expertise (e.g., a planner, a researcher, and a mathematician).