langchain-fundamentals

Build LangChain agents with create_agent(), middleware, and structured outputs.

Updated May 26, 2026

Search Tags:#middleware #langchain #agents #human-in-the-loop #structured output #state persistence #tool-calling

Ready to run

npx skills add https://github.com/anukkrit149/anukkrit-skills --skill langchain-fundamentals-anukkrit149

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: langchain-fundamentals
Source: https://github.com/anukkrit149/anukkrit-skills/tree/main/cloud/skills/langchain-fundamentals
Command: npx skills add https://github.com/anukkrit149/anukkrit-skills --skill langchain-fundamentals-anukkrit149

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you avoid fragile, outdated agent patterns by showing how to build production-ready LangChain agents that reliably orchestrate tools, state, and human approvals.

Core Features & Use Cases

Production agents with create_agent: Build agents using create_agent() so tool execution and state management work correctly and consistently.
Middleware-driven control flows: Add human-in-the-loop approvals, error handling, and logging using middleware patterns for safer automation.
Typed, structured results: Produce validated outputs via response_format or with_structured_output() so downstream code can trust the schema.
Real-world example: Use a human approval middleware when a tool is risky, pause for confirmation, then resume the same run to complete the task without losing context.

Quick Start

Create a LangChain agent using create_agent() with clear tools and middleware, and invoke it with your input messages.

Frequently Asked Questions about langchain-fundamentals

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I add human-in-the-loop approvals to a LangChain agent?▼

Human-in-the-loop approvals are added to LangChain agents by attaching a custom middleware to create_agent(). This middleware intercepts risky tool calls, pauses execution for confirmation, and resumes the run without losing context.

What is the best way to build production-ready LangChain agents?▼

Production-ready LangChain agents are built using create_agent() combined with middleware for control flows and checkpointers for state persistence. This pattern ensures reliable tool orchestration and consistent state management.

How do I get validated structured output from a LangChain agent?▼

Validated structured output from a LangChain agent is obtained by specifying response_format or using with_structured_output(). This enforces a schema so downstream code can safely trust the typed responses.

Can I use middleware to handle errors and logging in LangChain tool-calling pipelines?▼

Yes, middleware-driven control flows in LangChain handle errors, logging, and human approvals. Adding middleware to create_agent() provides safer automation for tool-calling pipelines by intercepting and managing execution steps.

How do I persist agent state across multiple LangChain runs?▼

Agent state is persisted across LangChain runs by configuring optional checkpointers with create_agent(). This maintains context and state safely throughout production workflows and human-in-the-loop pauses.