agentsclimarketplace

Deepagents langchain

Skill mahmoud20138/Tradecraft/plugins/tradecraft/skills/deepagents-langchain

DeepAgents — production-ready LangGraph agent framework by LangChain. Batteries-included: planning (write_todos), filesystem ops, shell execution, sub-agents with isolated context, auto context summarization. pip install deepagents → create_deep_agenFrom its SKILL.md

Install
npx -y skills add mahmoud20138/Tradecraft --skill deepagents-langchain

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 18 stars18 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.2 KB, 931 tokens by cl100k_base, as published. Nobody here has run it

deepagents-langchain

USE FOR:

  • "production-ready agent with LangGraph"
  • "batteries-included coding/research agent"
  • "sub-agents with isolated context windows"
  • "LangChain agent framework"
  • "agent with planning + filesystem + shell"
  • "MCP tools in LangGraph agent" tags: [LangGraph, LangChain, agent, production, sub-agents, MCP, planning, filesystem, shell, open-source] kind: framework category: ai-agent-builder

What Is DeepAgents?

Production-ready, batteries-included LangGraph agent by LangChain. No manual setup — create_deep_agent() returns a fully functional agent.


Quick Start

pip install deepagents
from deepagents import create_deep_agent

agent = create_deep_agent()
result = agent.invoke({
    "messages": [{"role": "user", "content": "Research the latest AI agent frameworks and summarize"}]
})
print(result["messages"][-1].content)

Built-in Capabilities

CapabilityTools Included
Planningwrite_todos — task decomposition + progress tracking
Filesystemread, write, edit, search files
Shellexecute commands (with sandboxing)
Sub-agentsdelegate tasks with isolated context windows
Contextauto-summarization, large output → file handling

Architecture (LangGraph)

# Returns a compiled LangGraph graph
agent = create_deep_agent()

# Supports all LangGraph features:
# - Streaming
for chunk in agent.stream({"messages": [("user", "task")]}):
    print(chunk)

# - Persistence / checkpointing
from langgraph.checkpoint.memory import MemorySaver
agent = create_deep_agent(checkpointer=MemorySaver())

# - LangGraph Studio compatibility (visual debug)

Customization

from deepagents import create_deep_agent
from langchain_anthropic import ChatAnthropic

# Custom model
agent = create_deep_agent(
    model=ChatAnthropic(model="claude-opus-4-6")
)

# Add custom tools
from langchain_core.tools import tool

@tool
def my_tool(query: str) -> str:
    """Custom tool description"""
    return do_something(query)

agent = create_deep_agent(tools=[my_tool])

# Custom system prompt
agent = create_deep_agent(
    system_prompt="You are an expert financial analyst..."
)

MCP Integration

from langchain_mcp_adapters import MCPToolkit

# Connect any MCP server
toolkit = MCPToolkit(server_command=["npx", "gitnexus", "mcp"])
mcp_tools = toolkit.get_tools()

agent = create_deep_agent(tools=mcp_tools)

Sub-Agents Pattern

# Main agent delegates to sub-agents with isolated contexts
# Sub-agents don't share main agent's conversation history
# Useful for: parallel research, isolated code execution

agent = create_deep_agent(
    enable_subagents=True,
    subagent_model=ChatAnthropic(model="claude-haiku-4-5-20251001")  # cheaper for subtasks
)

Use Cases

  • Research pipeline: fetch + summarize + synthesize across sources
  • Code automation: read codebase → plan changes → edit files → run tests
  • Data processing: ingest files → transform → write outputs
  • Multi-step workflows: plan → delegate subtasks → aggregate results

vs. Other Agent Frameworks

FeatureDeepAgentsOpenAliceAutoHedge
FrameworkLangGraphCustom TSSwarms
Built-in tools✓ (full)✓ (trading)✓ (trading)
Sub-agents✓✗✗
MCP support✓✓ (planned)✗
DomainGeneralTradingTrading
Studio UI✓ LangGraph✓ Web✗

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 610,677. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.