langchain-ai/deepagentsjs

What it solves

It eliminates the need to manually wire together prompts, tools, and context management when building AI agents. Instead of starting from scratch, developers get a "batteries-included" agent harness that is ready to run out of the box with sensible defaults.

How it works

Deep Agents provides a pre-configured agent framework built on LangGraph. It integrates several core capabilities by default:

• Planning: Uses a write_todos tool for breaking down tasks and tracking progress.
• Filesystem Access: Includes tools like read_file, write_file, edit_file, ls, glob, and grep to serve as the agent's working memory.
• Sub-agents: Uses a task tool to delegate work to other agents with isolated context windows.
• Context Management: Employs file-based workflows to handle long-running tasks.

Because it returns a compiled LangGraph graph, it supports production features like streaming, checkpointers, and persistence.

Who it’s for

Developers who want to deploy functional AI agents quickly using TypeScript/JavaScript without spending time on the initial boilerplate of tool and prompt configuration.

Highlights

• Ready-to-run: Comes with built-in planning and filesystem tools immediately.
• Provider agnostic: Compatible with any tool-calling chat model.
• LangGraph native: Full access to LangGraph's streaming and state management features.
• Flexible: Allows for custom tools, models, and system prompts.
• Environment-aware: Provides specific entrypoints for both Node.js and browser environments.