Pydantic DeepAgents

v1.0.0Coding Agentsstable

Build Claude Code–style deep agents in Python: tool-calling, sandboxed execution, multi-agent teams, skills, checkpoints, and unlimited context — all on Pydantic AI.

agent-frameworkai-agentsanthropicclaude-codecli
Share:
811
Stars
0
Downloads
0
Weekly
0/5

What is Pydantic DeepAgents?

Pydantic DeepAgents is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to build claude code–style deep agents in python: tool-calling, sandboxed execution, multi-agent teams, skills, checkpoints, and unlimited context — all on pydantic ai.

Build Claude Code–style deep agents in Python: tool-calling, sandboxed execution, multi-agent teams, skills, checkpoints, and unlimited context — all on Pydantic AI.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Build Claude Code–style deep agents in Python: tool-calling,

Use Cases

Claude Code-style deep agents in Python
Multi-agent teams with checkpoints and skills
vstorm-co

Maintainer

LicenseMIT
Languagepython
Versionv1.0.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx pydantic-deepagents

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Pydantic DeepAgents

Pydantic DeepAgents is a Python-native agent harness that brings Claude Code-style deep agents to any codebase, featuring live run forking with copy-on-write filesystem isolation, sandboxed Docker execution, multi-agent swarms with task delegation, persistent memory, and built-in MCP server integration. Agents can call filesystem, shell, web search, and browser automation tools out of the box, and in-flight forking lets you split a task into N parallel branches that an AI judge then merges. It is the go-to framework for teams that want autonomous coding agents with reproducible, auditable execution traces.

Prerequisites

  • Python 3.10 or later installed
  • An ANTHROPIC_API_KEY or OPENAI_API_KEY for the LLM backend
  • Docker installed (optional, for sandboxed execution)
  • An MCP client such as Claude Desktop if using MCP integration
1

Install via the one-liner installer

The install script handles everything including CLI setup on macOS and Linux.

curl -fsSL https://raw.githubusercontent.com/vstorm-co/pydantic-deep/main/install.sh | bash
2

Or install via pip

Install the Python package directly, with optional CLI extras.

pip install pydantic-deep
pip install "pydantic-deep[cli]"  # with CLI extras
3

Set your API key

Export the API key for your chosen LLM provider. Anthropic and OpenAI are directly supported; OpenRouter, Gemini, and Ollama endpoints also work.

export ANTHROPIC_API_KEY=your_anthropic_key
# or
export OPENAI_API_KEY=your_openai_key
4

Configure the default model

Set the model you want agents to use by default.

pydantic-deep config set model anthropic:claude-sonnet-4-6
5

Run an agent task

Launch the interactive TUI or run a one-shot task in headless mode.

# Interactive terminal
pydantic-deep

# Headless one-shot
pydantic-deep run "Fix failing tests in auth module"
6

Add MCP servers

Connect additional MCP servers (such as DeepWiki or GitHub) using the /mcp command inside the TUI, or configure them programmatically.

# Inside pydantic-deep TUI:
/mcp

Pydantic DeepAgents Examples

Client configuration

Claude Desktop config to run pydantic-deepagents as an MCP server.

{
  "mcpServers": {
    "pydantic-deepagents": {
      "command": "npx",
      "args": ["pydantic-deepagents"],
      "env": {
        "ANTHROPIC_API_KEY": "your_anthropic_key"
      }
    }
  }
}

Prompts to try

Tasks well-suited to Pydantic DeepAgents once connected.

- "Refactor the authentication module to use JWT tokens and add unit tests"
- "Fork the current task into two branches: one using a decorator pattern and one using a context manager, then merge the better solution"
- "Search the web for the latest Pydantic v2 migration guide and apply the necessary changes to this codebase"
- "Run the test suite in a Docker sandbox and fix any failing tests"

Troubleshooting Pydantic DeepAgents

pydantic-deep command not found after install

Make sure the install script completed without errors and that ~/.local/bin (or the Python scripts directory) is on your PATH. Try running source ~/.bashrc or opening a new terminal.

Docker sandbox fails to start

Ensure Docker Desktop is running and your user has permission to run docker commands (add yourself to the docker group on Linux). Use pydantic-deep tui --sandbox none to run without Docker.

API authentication errors when running agents

Confirm ANTHROPIC_API_KEY or OPENAI_API_KEY is exported in the same shell session or is present in the MCP client env block. Run pydantic-deep config list to verify the current model setting.

Frequently Asked Questions about Pydantic DeepAgents

What is Pydantic DeepAgents?

Pydantic DeepAgents is a Model Context Protocol (MCP) server that build claude code–style deep agents in python: tool-calling, sandboxed execution, multi-agent teams, skills, checkpoints, and unlimited context — all on pydantic ai. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Pydantic DeepAgents?

Follow the installation instructions on the Pydantic DeepAgents GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Pydantic DeepAgents?

Pydantic DeepAgents works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Pydantic DeepAgents free to use?

Yes, Pydantic DeepAgents is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "pydantic-deepagents": { "command": "npx", "args": ["-y", "pydantic-deepagents"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Pydantic DeepAgents?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides