What is the purpose of the SKILL.md file?

SKILL.md is a structured markdown file that uses progressive disclosure to load specialized instructions and tools only when they are relevant to the current task.

Does this skill support both Python and TypeScript?

Yes, Deep Agents Core provides implementation patterns and code examples for both Python and TypeScript environments.

What is the Deep Agents framework?

Deep Agents is an opinionated framework built on LangChain and LangGraph that provides a pre-configured harness for planning, memory, and delegation.

When should I use a Deep Agent instead of a standard agent?

Use Deep Agents for complex tasks that require multi-step planning, large context management, or persistent memory across multiple user sessions.

How do Deep Agents handle sensitive operations?

The framework includes Human-in-the-loop middleware that can be configured to interrupt the agent and request user approval for specific tool actions like writing files.

Deep Agents Core

Name: Deep Agents Core
Author: langchain-ai

bylangchain-ai

•

534

•

开发者工具

Streamlines the development of advanced AI agents with integrated task planning, persistent memory, and modular skill management.

Deep Agents Core provides the essential framework for building sophisticated, multi-step AI agents using LangChain and LangGraph. It offers an opinionated middleware harness that automatically handles complex behaviors like hierarchical task delegation, filesystem context management, and human-in-the-loop approvals. By standardizing how agents interact with tools and maintain state across threads, it enables developers to build production-ready autonomous systems that are both persistent and capable of high-level reasoning without manual architecture setup.

主要功能

01534 GitHub stars

02On-demand skill loading using the standardized SKILL.md progressive disclosure format

03SubAgent orchestration for delegating work to specialized assistant instances

04Integrated TodoListMiddleware for automated multi-step task planning

05Persistent long-term memory and session state via Store and Checkpointers

06Pluggable Filesystem and Store backends for flexible context management

使用场景

01Creating multi-agent research workflows where a lead agent manages specialized sub-workers

02Building autonomous coding assistants that require complex file manipulation and directory awareness

03Developing stateful customer support bots that maintain context across multiple user sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add langchain-ai/langchain-skills deep-agents-core

For use in Claude.ai and ChatGPT

主要功能

01534 GitHub stars

02On-demand skill loading using the standardized SKILL.md progressive disclosure format

03SubAgent orchestration for delegating work to specialized assistant instances

04Integrated TodoListMiddleware for automated multi-step task planning

05Persistent long-term memory and session state via Store and Checkpointers

06Pluggable Filesystem and Store backends for flexible context management

使用场景

01Creating multi-agent research workflows where a lead agent manages specialized sub-workers

02Building autonomous coding assistants that require complex file manipulation and directory awareness

03Developing stateful customer support bots that maintain context across multiple user sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add langchain-ai/langchain-skills deep-agents-core

For use in Claude.ai and ChatGPT