Extra Credit

LangChain

Official docs outline

17 sections
  1. LangChain overview

    LangChain framework purpose, standard model interface, agents, relationship to LangGraph and LangSmith.

  2. Models

    Chat models via init_chat_model, invoke/stream/batch, model parameters, provider integrations, tool calling and multimodal support.

  3. Messages

    Message types (system, human, AI, tool), content blocks, message history and metadata.

  4. Tools

    Defining tools with the @tool decorator, schemas, tool runtime context, ToolNode execution.

  5. Agents

    create_agent, the model-tool loop, system prompts, dynamic models, agent state and invocation.

  6. Structured output

    Returning typed responses with Pydantic, dataclass or TypedDict schemas, response_format strategies.

  7. Streaming

    Streaming tokens, updates and custom events from models and agents, stream modes.

  8. Short-term memory

    Conversation state with checkpointers and thread IDs, trimming, summarizing and deleting messages.

  9. Long-term memory

    Persistent cross-thread memory with stores, namespaces, semantic search over memories.

  10. Retrieval and RAG

    Retrieval-augmented generation: document loaders, text splitting, embeddings, vector stores, retrievers.

  11. Vector stores and embeddings

    Vector store integrations, embedding models, similarity search, indexing documents.

  12. Middleware

    Agent middleware hooks around model and tool calls, summarization, retries, custom middleware.

  13. Context engineering

    Controlling model context, tools and prompts per step through runtime context, state and store.

  14. Guardrails

    Input and output validation, PII handling, content filtering and safety checks in agents.

  15. Human-in-the-loop

    Interrupts for approving, editing or rejecting tool calls, resuming from checkpoints.

  16. Model Context Protocol in LangChain

    Loading MCP server tools via langchain-mcp-adapters, multi-server clients and transports.

  17. Observability with LangSmith

    Tracing runs, debugging agent steps, datasets and evaluation, monitoring in production.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.

All 39